أهمية وأداء في PLS-SEM وNCA: تقديم تحليل خريطة الأهمية والأداء المدمجة (cIPMA) Importance and performance in PLS-SEM and NCA: Introducing the combined importance-performance map analysis (cIPMA)

المجلة: Journal of Retailing and Consumer Services، المجلد: 78
DOI: https://doi.org/10.1016/j.jretconser.2024.103723
تاريخ النشر: 2024-01-23

جامعة جنوب الدنمارك SDÚ

الأهمية والأداء في PLS-SEM و NCA

تقديم تحليل خريطة الأهمية والأداء المدمجة (cIPMA)

هاوف، سفين؛ ريشتر، نيكول فرانزيسكا؛ سارستيدت، ماركو؛ رينجل، كريستيان م.

نُشر في:

مجلة البيع بالتجزئة وخدمات المستهلك

DOI:

10.1016/j.jretconser.2024.103723

تاريخ النشر:

2024

نسخة الوثيقة:

النسخة النهائية المنشورة

ترخيص الوثيقة:

CC BY-NC-ND

الاستشهاد بالنسخة المنشورة (APA):

هاوف، س.، ريشتر، ن. ف.، سارستيدت، م.، ورينجل، ك. م. (2024). الأهمية والأداء في PLS-SEM و NCA: تقديم تحليل خريطة الأهمية والأداء المدمجة (cIPMA). مجلة البيع بالتجزئة وخدمات المستهلك، 78 (مايو)، المقال 103723.https://doi.org/10.1016/j.jretconser.2024.103723
اذهب إلى إدخال النشر في بوابة أبحاث جامعة جنوب الدنمارك

شروط الاستخدام

تم تقديم هذا العمل من قبل جامعة جنوب الدنمارك.
ما لم يُذكر خلاف ذلك، فقد تم مشاركته وفقًا للشروط الخاصة بالأرشفة الذاتية.
إذا لم يتم ذكر ترخيص آخر، تنطبق هذه الشروط:
  • يمكنك تنزيل هذا العمل للاستخدام الشخصي فقط.
  • لا يمكنك توزيع المادة بشكل إضافي أو استخدامها لأي نشاط يهدف إلى الربح أو الكسب التجاري
  • يمكنك توزيع عنوان URL الذي يحدد هذه النسخة المفتوحة الوصول بحرية
إذا كنت تعتقد أن هذه الوثيقة تنتهك حقوق الطبع والنشر، يرجى الاتصال بنا مع تقديم التفاصيل وسنحقق في ادعائك. يرجى توجيه جميع الاستفسارات إلىpuresupport@bib.sdu.dk

أهمية وأداء في PLS-SEM وNCA: تقديم تحليل خريطة الأهمية والأداء المدمجة (cIPMA)

سفين هاوف ، نيكول فرانزيسكا ريشتر ، ماركو سارستيدت ، كريستيان م. رينجل جامعة هيلموت شميت، هولستنهوفويغ 85، 22043، هامبورغ، ألمانيا جامعة جنوب الدنمارك، كامبوسفاي 55، DK-5230، أودنسه م، الدنمارك جامعة لودفيغ ماكسيميليان في ميونيخ، مدرسة إدارة LMU ميونيخ، لودفيغ شتراسه 28 RG/III، 80539، ميونيخ، ألمانيا جامعة بابل بولياي، كلية الاقتصاد وإدارة الأعمال، رومانيا جامعة هامبورغ للتكنولوجيا، قسم علوم الإدارة والتكنولوجيا، أم شوارزنبرغ-كامبوس 4 (D)، 21073، هامبورغ، ألمانيا جامعة جيمس كوك، أستراليا

معلومات المقال

المحرر المعالج: البروفيسور ه. تيمرمانس

الكلمات المفتاحية:

cIPMA
تحليل الشرط الضروري
NCA
المربعات الصغرى الجزئية
PLS
PLS-SEM
نمذجة المعادلات الهيكلية
SEM
تحليل خريطة الأهمية والأداء
IPMA

الملخص

يقدم هذا البحث نهجًا جديدًا يوسع تطبيق تحليل خريطة الأهمية والأداء (IPMA) في نمذجة المعادلات الهيكلية باستخدام المربعات الصغرى الجزئية (PLS-SEM) من خلال دمج النتائج من تحليل الشرط الضروري (NCA). يتضمن IPMA تقييم أهمية وأداء المتغيرات الكامنة ومؤشراتها، بينما يقدم NCA بعدًا إضافيًا من خلال تحديد العوامل التي تعتبر حاسمة لتحقيق النتائج المرغوبة. يستخدم NCA منطق الضرورة لتحديد العوامل الأساسية المطلوبة لتحقيق نتيجة، بينما يتبع PLS-SEM منطق الكفاية الإضافية لتحديد العوامل التي يجب أن تكون موجودة والتي تساهم في مستويات الأداء العالية. إن دمج هذين المنطقين في بعد الأداء ذو قيمة خاصة لتحديد أولويات الإجراءات التي يمكن أن تحسن النتائج المستهدفة، مثل رضا العملاء والتزام الموظفين. على الرغم من أن الاستخدام المشترك لـ PLS-SEM و NCA هو اقتراح حديث، فإن هذه الدراسة هي الأولى التي تجمع بينهما مع IPMA (أي، في IPMA المدمجة؛ cIPMA). توضح دراسة حالة الاستخدام المشترك لـ PLS-SEM و NCA لإجراء cIPMA. يعزز هذا النهج المبتكر اتخاذ القرارات للباحثين والممارسين، مما يمكنهم من تحديد أولويات جهودهم بشكل فعال.

1. المقدمة

مقارنة أداء وأهمية السمات من أجل إنتاج نتيجة معينة له تقليد طويل في مجال الإدارة (مارتلا وجيمس، 1977). يقوم الباحثون عادةً بتصور هذا التفاعل في مخطط ثنائي الأبعاد، حيث يتم عادةً تجميع السمات في أربعة أرباع تجمع بين قيم الأهمية والأداء المنخفضة والعالية. يُشار إلى هذا المخطط عادةً باسم شبكة الأهمية والأداء (مارتلا وجيمس، 1977)، مصفوفة الأهمية والأداء (سلاك، 1994)، خريطة الجودة (كريستنسن وآخرون، 2000)، أو، بشكل أوسع، تحليل خريطة الأهمية والأداء (IPMA، رينجل وسارستيدت، 2016). أصبح IPMA أداة قياسية لفهم المكان الذي يجب أن تركز فيه جهود التحسين الإداري (على سبيل المثال، سيفر، 2015؛ سكووك وآخرون، 2001). ومع ذلك، واجه مؤلفون من مجالات وتوجهات منهجية مختلفة تحديات عند استخدام مجموعة الأدوات، والتي تتعلق – من بين أمور أخرى – بفهم الأهمية والأداء، فضلاً عن
تعريف مستويات العتبة (انظر، على سبيل المثال، أوه، 2001).
تم استخدام IPMA أيضًا في سياق نمذجة المعادلات الهيكلية باستخدام المربعات الصغرى الجزئية (PLS-SEM)، وهي طريقة متعددة المتغيرات لتقدير العلاقات المعقدة بين البنى ومؤشراتها (هاير وآخرون، 2022، الفصل 1؛ لوهمولر، 1989، الفصل 2؛ وولد، 1982). في العقود الماضية، تطور استخدام PLS-SEM في أبحاث التسويق (غوينتر وآخرون، 2023؛ هاير وآخرون، 2012؛ سارستيدت وآخرون، 2022)، بما في ذلك البيع بالتجزئة وخدمات المستهلك (على سبيل المثال، كاي وآخرون، 2023؛ هيو-تي-الكوسر وهيرنانديز-روخاس، 2022؛ رودريغيز وآخرون، 2020). يسهل PLS-SEM إجراء IPMA، بسبب الطريقة التي تقدر بها الطريقة معلمات النموذج. على وجه التحديد، يقوم PLS-SEM بحساب مركبات المتغيرات المؤشرية لتمثيل البنى في نموذج إحصائي. استخدم الباحثون هذه الدرجات المركبة كتمثيلات لأداء البنى وقارنوها مع التأثيرات الكلية التي تمارسها البنى على هدف معين (على سبيل المثال، كريستنسن وآخرون، 2000؛ رينجل وسارستيدت، 2016؛ ستروكينز وآخرون، 2017). يستند IPMA الأساسي إلى متوسط درجات الأداء والأهمية ومن ثم
يتصورها في خريطة الأهمية والأداء.
استخدمت العديد من الدراسات IPMA في سياق PLS-SEM. على سبيل المثال، في سياق البيع بالتجزئة وخدمات المستهلك، يستند زانغ وآخرون (2023) إلى IPMA للتحقيق في أهمية وأداء المحفزات البيئية لنوايا الشراء داخل المتجر لدى المستهلكين عبر اثنين من بنى الموقف الاستهلاكي. في نموذج قبول التكنولوجيا (TAM)، الذي يتميز أيضًا بشكل بارز في البيع بالتجزئة وخدمات المستهلك (على سبيل المثال، إنغ وآخرون، 2022؛ بيريز-أراندا وآخرون، 2023؛ شهيدي وآخرون، 2022)، يمكن استخدام IPMA لمقارنة التأثير الكلي لسهولة استخدام التكنولوجيا من حيث استخدامها المقصود (أهميتها للبنية المستهدفة) مع متوسط سهولة الاستخدام المدركة من قبل مستخدميها المحتملين (أدائها). وبالتالي، تسمح نتائج IPMA بتحديد العوامل السابقة ذات الأهمية العالية للبنية المستهدفة، والتي، مع ذلك، تظهر أداءً منخفضًا نسبيًا. هذه المعلومات حاسمة، حيث تشير إلى كيفية تحديد أولويات الإجراءات لتحسين هذه البنى السابقة، مما سيزيد بشكل كبير من البنية المستهدفة. على سبيل المثال، تشير الأهمية العالية لسهولة الاستخدام، جنبًا إلى جنب مع تقييم سهولة الاستخدام الضعيف للتكنولوجيا (أي، أداء ضعيف)، إلى حاجة محددة لتحسين سهولة استخدام هذه التكنولوجيا لزيادة استخدامها في النهاية.
بطبيعتها، يتبع تنفيذ IPMA منطق الكفاية الإضافية الذي يمكن أن تسهم فيه عدة عوامل سابقة في تحقيق نتيجة معينة. يتم إعطاء الأولوية للعوامل السابقة التي تحسن النتيجة، أي تلك التي تظهر أداءً منخفضًا ولكن لها تأثير إجمالي قوي وذو دلالة على النتيجة، بينما لا يتم إعطاء الأولوية لتلك التي تظهر تأثيرات ضعيفة. ومع ذلك، قد يكون من الممكن أن يتطلب عامل سابق معين ذو أهمية نسبية منخفضة (أي تأثير ضعيف فيما يتعلق بزيادة النتيجة المرغوبة) اهتمامًا خاصًا عندما تمنع غيابه تحقيق النتيجة. على سبيل المثال، قد يحدد IPMA في سياق TAM أن توافق التكنولوجيا له أهمية نسبية منخفضة لاستخدامها. ومع ذلك، إذا لم يحقق التوافق عتبة معينة، فلن يكون المستخدمون راغبين في استخدام التكنولوجيا. يتطلب النظر في مثل هذه العلاقات المتبادلة اعتماد منظور الضرورة، مما يعني أن النتيجة – أو مستوى معين من النتيجة – لا يمكن تحقيقها إلا إذا كانت السبب الضروري موجودًا أو عند مستوى معين.
لدمج مثل هذه المنطق الضروري، يمكن للباحثين الاستفادة من تحليل الشرط الضروري (NCA؛ دول، 2016؛ 2020)، الذي اكتسب مؤخرًا شهرة في العديد من مجالات البحث التجاري (للاطلاع على مراجعة حديثة للمواضيع التي تم تحليلها باستخدام NCA، انظر دول وآخرون، 2023)، سواء كطريقة مستقلة أو بالاشتراك مع PLS-SEM وطرق الانحدار الأخرى (ريختر وآخرون، 2022). يسعى NCA إلى تحديد الشروط الضرورية التي تمثل القيود أو الاختناقات أو العوامل الحرجة، والتي يجب حلها أو تلبيتها لتحقيق نتيجة معينة. إن تحديد مثل هذه الشروط الضرورية له أهمية كبيرة في ممارسة الإدارة، حيث لا يمكن تحقيق نتيجة معينة إلا إذا كانت الشرط الضروري موجودًا أو عند مستوى معين (دول وآخرون، 2021؛ هوف وآخرون، 2021؛ ريختر وهوف، 2022). من المهم أن تُظهر البنى السابقة التي تظهر تأثيرات ضعيفة وغير دالة في PLS-SEM أنها قد تكون شروطًا ضرورية؛ أي أنه بدون هذه البنى السابقة لا يمكن تحقيق نتيجة معينة. وبالتالي، فإن عدم النظر في هذه البنى السابقة قد يؤدي إلى توصيات غير مكتملة. على سبيل المثال، في سياق تجارة التجزئة وخدمات المستهلكين، يبرز باباس (2023) تأثير الركود ومخاطر الجودة كشرطين ضروريين دائمين لنوايا الشراء لدى المصطافين. يستخدم العيهي وآخرون (2023) NCA للكشف عن أن الالتزام الأخلاقي، والمساءلة الأخلاقية، والمخاطر المدركة، ومعرفة التكاليف هي شروط سابقة ضرورية لشراء المستهلكين للمنتجات المعاد تصنيعها.
تم الاعتراف بالاستخدام المشترك المقترح حديثًا لـ NCA و PLS-SEM (ريختر وآخرون، 2020) كـ “مساهمة فريدة من خلال مقارنة ودمج الأساليب، مما يوضح كيف يمكن استخدام NCA (الذي يركز على الشروط الضرورية) بالاشتراك مع PLS-SEM (الذي يركز على الكفاية) لإنشاء طريقة غير معترف بها سابقًا لتقييم السببية” (بيرغ وآخرون، 2022، ص. 1842). تماشيًا مع الأبحاث الحديثة التي اعتمدت هذا النهج متعدد الأساليب (مثل…
نحن نؤكد أنه يمكن أن يثري تحديد الشروط الضرورية أيضًا تحليل الأهمية النسبية (IPMA) من خلال توفير معلومات حول ضرورة البنى السابقة. توفر هذه المعلومات للباحثين والممارسين صورة أكثر اكتمالاً عن البنى السابقة المهمة والضرورية، بينما تساعد أيضًا في تكملة نتائج تحليل الأهمية النسبية. من خلال اتباع نتائج نموذج المسار لفائدة التكنولوجيا المدركة، قد يجد الباحثون الذين يقيمون قبول المستهلك للتقنيات الجديدة، على سبيل المثال، أن هذه الفائدة ليست ذات أهمية كبيرة، على الرغم من أنها شرط ضروري لاستخدام التكنولوجيا. بعبارة أخرى، لن يتم استخدام التكنولوجيا ما لم يتم إدراكها على أنها مفيدة. بينما قد يتلقى بناء سابق ذو أهمية قليلة وفقًا لنتائج تحليل الأهمية النسبية (PLS-SEM) اهتمامًا ضئيلًا من تحليل الأهمية النسبية الكلاسيكي، إلا أنه قد يمثل شرطًا ضروريًا، مما يجعله مرة أخرى محور اتخاذ القرار عندما يتم دمج تحليل الأهمية النسبية مع تحليل الشروط الضرورية (NCA).
في هذا السياق، نساهم أولاً في هذا المجال من خلال مناقشة كيفية فهم الأهمية والأداء في سياق PLS-SEM وNCA، بالإضافة إلى كيفية تكامل النهجين في تحليلات الأهمية والأداء. بناءً على الأبحاث السابقة في كل مجال (مثل، دول، 2016، 2020؛ رينجل وسارستيدت، 2016)، نناقش مفهومي الأهمية والأداء في سياق PLS-SEM – كما تم تنفيذه في IPMA – وسياق NCA. بالإضافة إلى ذلك، نقدم إرشادات لـ IPMA المدمجة التي تعتمد على نتائج PLS-SEM وNCA. من خلال القيام بذلك، نوسع الإرشادات المقترحة في ريشتر وآخرون (2020)، الذين لم يتناولوا IPMA (انظر أيضًا ريشتر وآخرون، 2023ب). أخيرًا، نوضح IPMA المدمجة باستخدام TAM موسع ونقدم للباحثين الذين ينفذون هذا النهج مجموعة أدوات، قبل أن نختم بحثنا ونحدد أبحاث المستقبل.

2. الأهمية والأداء في PLS-SEM

PLS (Lohmöller، 1989، الفصل 2؛ Wold، 1982) هو نهج قائم على المركبات في SEM حيث يمثل البنى كمجموعات موزونة من المؤشرات (Hair et al.، 2017؛ Sarstedt et al.، 2016). استنادًا إلى هذه الخاصية، استخدم الباحثون الدرجات المركبة التي تنتجها PLS-SEM لتطوير مؤشرات مثل مقياس رضا العملاء السويدي (Fornell، 1992) ومؤشر رضا العملاء الأمريكي (Fornell et al.، 1996). في حساب هذه المؤشرات، تُعتبر الدرجات المركبة كقيم أداء، تعكس رضا المستجيبين عن ميزات معينة أو عن البنية العامة. على سبيل المثال، إذا أظهر جميع المستجيبين مستوى أقصى من الرضا العام، فإن ذلك يترجم إلى أداء بنسبة 100 في المئة لهذه البنية. بدءًا من Anderson وMittal (2000) وKristensen et al. (2000)، قارن البحث اللاحق هذه الدرجات بأوزان النموذج القياسي أو الهيكلي التي تمثل أهمية المؤشرات الفردية أو البنى لتحسين هدف معين، مثل ربح الشركة. معًا، تحدد هذان البعدان محاور خريطة الأهمية والأداء، التي تمثل متوسط درجات الأهمية والأداء فيما يتعلق ببنية هدف معينة.
مع تزايد تعقيد النماذج، التي غالبًا ما تمتد عبر عدة طبقات من البنى، يعتمد الباحثون عادةً على التأثيرات الكلية للنموذج الهيكلي – أي، مجموع التأثيرات المباشرة وغير المباشرة لبنية سابقة على نتيجة معينة – لتمثيل بُعد الأهمية في تحليل الأهمية والأداء (IPMA) (Hair et al.، 2024، الفصل 4؛ Ringle و Sarstedt، 2016). أي، بدلاً من تقييد التحليل بمقدمات الهدف المباشرة، يأخذ تحليل الأهمية والأداء في الاعتبار أيضًا البنى ذات الصلة بشكل غير مباشر (أي، تلك التي تؤثر على النتيجة من خلال واحدة أو أكثر من البنى الوسيطة).
في خريطة الأهمية والأداء، يتم رسم قيم أهمية البنى السابقة على -محور وقيم أدائها على -المحور. باستخدام هذه التوضيحات، يمكن للباحثين، على سبيل المثال، تحديد البنى السابقة التي تعتبر مهمة نسبيًا فيما يتعلق بشرح البنى المستهدفة الرئيسية ذات الاهتمام (أي، تلك التي لها تأثير إجمالي قوي) مع أداء نسبي منخفض (أي، درجات متغيرة كامنة منخفضة). ستكون هذه البنى، بشكل خاص، ذات أولوية عالية من أجل تحقيق التحسين، لأنها خاصة جدًا
مهم للبناء المستهدف، على الرغم من أنهم يؤدون بشكل ضعيف نسبيًا في الوقت نفسه.
لتوضيح مفهوم IPMA بشكل أفضل، اعتبر نموذج المسار الافتراضي في الشكل 1 (اللوحة A) مع الأربعة مفاهيم إلى في هذا نموذج المسار، هو البناء المستهدف. استنادًا إلى معاملات المسار، فإن البنى السابقة ، و تؤثر بشكل مباشر على . بالإضافة إلى ذلك، و تؤثر بشكل غير مباشر على عبر ، والتي تعود إلى ناتج التأثيرات المباشرة المقابلة (للمزيد من التفاصيل، انظر Hair وآخرون، 2024، الفصل 4). في هذا المثال، التأثيرات غير المباشرة لـ على يتم حسابها على النحو التالي:



لذا، فإن التأثير غير المباشر الإجمالي لـ على هو . إضافة التأثير المباشر لـ على (أي، 0.50)، بالإضافة إلى التأثير غير المباشر الإجمالي للبنى السابقة (أي، 0.34)، نحصل ليس فقط على التأثير الإجمالي لـ ، ولكن أيضًا على أهمية للبنية المستهدفة الرئيسية .
على النقيض من ذلك، تمثل القيم المتوسطة لدرجات البنية أدائها. هنا، يمكن للباحثين استخدام درجات البنية الموحدة أو درجات البنية غير الموحدة. يتم تقدير درجات البنية الموحدة على أساس – بيانات المؤشر الموحد المحولة. درجات البنية الموحدة الناتجة دائمًا ما يكون لها قيمة متوسطة تساوي صفر وانحراف معياري يساوي 1. نظرًا لأن الإبلاغ عن القيم المتوسطة التي تساوي صفر لجميع البنى أقل فائدة لـ IPMA، فإن الإجراء يشير إلى درجات البنية غير الموحدة.
تعتمد درجات البنية غير الموحدة على مقاييس المؤشرات
الشكل 1. مثال على تحليل خريطة الأهمية والأداء؛ معتمد من Hair et al. (2024، الفصل 4).
(على سبيل المثال، من 1 إلى 5 أو من 1 إلى 7). إذا تم قياس جميع المؤشرات على نفس المقياس (على سبيل المثال، جميعها على مقياس من 1 إلى 5)، فإن تفسيرها على مستوى البنية يكون مباشرًا. ومع ذلك، يصبح تفسيرها غامضًا إذا كانت المقاييس مختلفة (على سبيل المثال، Ringle و Sarstedt، 2016؛ Hair et al.، 2024، الفصل 4). من أجل مقارنة أفضل لمستويات الأداء، يمكن إعادة قياس بيانات المؤشر غير الموحدة بحيث تتراوح جميعها بين 0 و 100، حيث يمثل 0 أدنى أداء و 100 أعلى أداء. تتم إعادة قياس ملاحظة بالنسبة للمؤشر عبر

حيث هو -th المؤشر في نموذج مسار PLS، E[.] تمثل الدرجة الفعلية للمؤشر i للمستجيب ، و min[.] و max[.] تمثل القيم الدنيا والقصوى للمؤشر (Ringle و Sarstedt، 2016؛ Hair et al.، 2024، الفصل 4). يمكن أن تشير القيم الدنيا والقصوى للمؤشر إلى القيم الدنيا والقصوى النظرية (على سبيل المثال، 1 و 7 عند استخدام مقياس من سبع نقاط) أو القيم الدنيا والقصوى التجريبية الناتجة عن بيانات العينة (على سبيل المثال، 2 و 7، لأن أي مستجيب لم يقيم المؤشرات بأقل من 2). من السهل تحديد القيم الدنيا والقصوى النظرية لمقياس الفاصل (على سبيل المثال، 1 و 7 عند استخدام مقياس من سبع نقاط). قد لا تحتوي استخدامات مقاييس أخرى على حد أدنى أو أقصى نظري، وقد يُنصح الباحث أو حتى يُجبر على استخدام الحد الأدنى أو الأقصى التجريبي للبيانات المتاحة.
تكون درجات المتغيرات الكامنة المعاد قياسها عبارة عن تركيبة خطية من بيانات المؤشر المعاد قياسها والأوزان الخارجية المعاد قياسها. للحصول على الأوزان المعاد قياسها، يجب علينا أولاً حساب الأوزان غير الموحدة عن طريق قسمة الأوزان الموحدة على الانحراف المعياري لمؤشرها المعني. يتم تطبيع هذه الأوزان لكل نموذج قياس بحيث يكون مجموع هذه الأوزان الخارجية النهائية المعاد قياسها يساوي واحدًا. تتيح لنا هذه الأوزان الخارجية المعاد قياسها تحديد درجات البنية غير الموحدة المعاد قياسها التي تتراوح بين 0 و 100 (أي، من خلال التركيبة الخطية لبيانات المؤشر على مقياس من 0 إلى 100 والأوزان الخارجية المعاد قياسها لكل بنية). تمثل متوسط هذه القيم درجة أداء البنية المشار إليها في IPMA في PLSSEM (انظر القيم بين الأقواس في البنى في الشكل 1، اللوحة A).
يوضح الشكل 1 (اللوحة B) كيف تجمع IPMA بين هذين الجانبين رسوميًا للبنية المستهدفة . قسم خطان رماديان متقطعان خريطة الأهمية والأداء إلى أربعة أرباع. يمثل الخط العمودي الأهمية المتوسطة والخط الأفقي يتوافق مع الأداء المتوسط للبنى، المحسوب كما يلي (الشكل 1، اللوحة A):
الأهمية المتوسطة ، و
الأداء المتوسط .
يمكن للباحثين الذين يستخدمون IPMA استخدام هذه الدرجات المتوسطة ورسمها على خريطة الأهمية والأداء إلى أربعة أرباع منفصلة. ومع ذلك، قد يعتمد الباحثون على منطق مختلف عند تقسيم الأرباع الأربعة، مثل المعرفة السابقة أو تقييم الخبراء. يؤدي الجمع بين الأهمية العالية/المنخفضة والأداء العالي/المنخفض إلى توصيات محددة، تستهدفها البنى السابقة من خلال الأنشطة الإدارية (Ringle و Sarstedt، 2016).

3. الأهمية والأداء في NCA

على عكس IPMA الكلاسيكية، يبني NCA على منطق الضرورة لتقييم أهمية وأداء البنى السابقة لنتيجة. يعني منطق الضرورة أن النتيجة لا يمكن تحقيقها إلا إذا كانت حالة معينة موجودة (Dul، 2016، 2020، الفصل 2). وبالتالي، تمثل الحالة الضرورية قيدًا أو عنق زجاجة يجب التغلب عليه للسماح بوجود النتيجة. من أجل اكتشاف الشروط الضرورية في مجموعات البيانات، يستخدم NCA تقنيات خط السقف لفصل
المناطق في الرسوم البيانية المبعثرة بدون ملاحظات عن تلك التي تحتوي على ملاحظات. من خلال تقسيم المساحة ‘الفارغة’ (المعروفة أيضًا باسم منطقة السقف) على المساحة الكاملة التي تشمل الملاحظات (المعروفة أيضًا باسم النطاق)، يحسب NCA حجم تأثير الضرورة ، والذي يمكن أن يكون 0 إذا لم تكن هناك مساحة فارغة، و 1 إذا كانت المساحة القصوى الممكنة فارغة.
خذ، على سبيل المثال، حالة ثلاثية حيث تحتوي الحالة والنتيجة كل منهما على ثلاثة مستويات، وهي منخفض ومتوسط وعالي (الشكل 2). يتم حساب حجم التأثير على أنه ، مع كون هو منطقة السقف و هو النطاق. يتم حساب حجم عن طريق عد عدد ‘الخلايا الفارغة’. يمكن حساب كما يلي: ؛ حيث هو عدد المستويات، و هو عدد المستويات. في مثالنا، . في السيناريو A في الشكل 2، يكون حجم التأثير هو ؛ في السيناريو B في الشكل 2، يكون حجم التأثير هو 0.75 (للحصول على تفاصيل حول حساب أحجام التأثير، انظر Dul، 2020، الفصل 4). وبالتالي، يعتمد حجم التأثير على عدد الخلايا الفارغة من الإجمالي. إذا كانت خلية واحدة فقط فارغة، فإن حجم التأثير هو 0.25 (الشكل 2، السيناريو A). إذا كانت ثلاث خلايا فارغة، فإن حجم التأثير هو 0.75 (الشكل 2، السيناريو B).
لذا، يحدد حجم التأثير مدى تقييد بواسطة . كمعيار تعسفي، يمكن اعتبار ‘تأثيرًا صغيرًا’، و ‘تأثيرًا متوسطًا’، و ‘تأثيرًا كبيرًا’، و d ‘تأثيرًا كبيرًا جدًا’ (Dul، 2016). ومع ذلك، من أجل الاستنتاج بأن فرضية الشرط الضروري يمكن قبولها، يجب أن يكون لـ على الأقل حجم تأثير متوسط (على سبيل المثال، ) وأهمية إحصائية (على سبيل المثال، مع قيمة أقل من 0.05). علاوة على ذلك، يجب أن يكون الشرط الضروري مدعومًا دائمًا نظريًا (Dul، 2021a، الفصل 2؛ Dul et al.، 2023).
من أجل اتخاذ قرار بشأن ما إذا كان حجم التأثير المحدد مرضيًا، يجب على الباحثين دائمًا تضمين المعرفة السياقية (Dul، 2020، الفصل 4). في الواقع، قد يُنظر إلى تقييم الأهمية لـ NCA على أنه أقل وضوحًا من تطبيق PLS-SEM. على سبيل المثال، قد يكون الاستنتاج بأن حجم تأثير 0.75 أكثر أهمية من 0.25 مضللًا إذا كان مستوى عالٍ من مطلوبًا دائمًا أو مطلوبًا من منظور عملي. في مثالنا الثلاثي، يمكن اعتبار كلا حجم التأثيرين متساويين في الأهمية، حيث يشير كلاهما إلى أن مستوى معين من مطلوب لتحقيق مستوى عالٍ من (أي، مستوى متوسط من في السيناريو A ومستوى عالٍ من في السيناريو B). على العكس، قد لا يكون مستوى على الأقل متوسط وحجم تأثير ذو دلالة إحصائية مهمًا على الإطلاق إذا لم يقيد المستوى المطلوب للنتيجة. على سبيل المثال، في السيناريو A، لا يتم تقييد نتيجة بواسطة إذا كان المرء يسعى فقط لتحقيق مستوى متوسط من . في هذه الحالة، فإن حجم التأثير 0.25 ليس له أهمية عملية. على النقيض من ذلك، في السيناريو B، يمثل شرطًا ضروريًا لتحقيق مستوى نتيجة متوسط من .
في سياق NCA، يجب فهم الأداء من حيث مقدار المستوى المطلوب لحالة معينة الذي تم تحقيقه بالفعل. يمكن تحليل ذلك من منظور حالة أو عينة. من منظور مستوى الحالة، يمكننا تحديد ما إذا كانت حالة واحدة تؤدي بشكل أفضل من حالة أخرى من حيث تحقيق المستوى الضروري. على سبيل المثال، في الشكل 2، السيناريو B، تؤدي حالة بمستوى منخفض بشكل أسوأ من حالة بمستوى متوسط (على الرغم من أن هذا لا يزال غير كافٍ لتحقيق مستوى عالٍ من المستوى). قد تكون هذه النظرة من منظور الحالة، على سبيل المثال، مثيرة للاهتمام بشكل خاص إذا كانت منظمة ما ترغب في معرفة كيف تقارن مع منظمات أخرى. من منظور العينة، يمكننا تحليل عدد الحالات في عينة ما التي لا تزال بحاجة إلى تحقيق المستوى المطلوب من الشرط. يمكن تطبيق هذا المنظور العيني، على سبيل المثال، إذا كانت منظمة ما ترغب في الحصول على رؤى حول عملائها أو موظفيها. تركز تحليلات PLS-SEM وتفسير نتائج تحليل خريطة الأداء المشترك (cIPMA) على العينات (أو العينات الفرعية، مثل مجموعات محددة من العملاء في)
مجموعة البيانات)، مما يجعل المنظور الأخير ذا صلة خاصة عند دمج نتائج IPMA المستندة إلى PLS-SEM مع رؤى NCA.
في حالة وجود شروط ضرورية متعددة، يجب أن تساعد مقارنة مستويات الأداء المختلفة في تحديد الشروط التي تتطلب اهتمامًا خاصًا. على سبيل المثال، إذا كانت معظم الحالات قد حققت المستوى المطلوب بالفعل، فقد لا تكون هذه الحالة ملحة كما هو الحال في حالة لم تحقق فيها العديد من الحالات ذلك (دول، 2021أ، الفصل 4). وبالتالي، يمكن أن تساعد مقارنة أداء الشروط الضرورية المختلفة في تحديد أولويات الإجراءات.
باختصار، من أجل تقييم أهمية الشروط الضرورية المختلفة في سياق cIPMA، نقترح على الباحثين أولاً تقييم ما إذا كان الشرط الضروري مدعومًا نظريًا وما إذا كان له على الأقل حجم تأثير متوسط ويكون ذا دلالة إحصائية. ومع ذلك، لا نوصي بالتركيز فقط على حجم تأثير الحجم لفهم أهميته العملية في سياق مشروع بحث معين. بدلاً من ذلك، نقترح تحديد مستوى النتيجة المرغوبة (أو البناء المستهدف) وتقييم ما إذا كان يمكن تحقيق هذا المستوى فقط بمستوى معين من الشرط. يمكن القيام بذلك باستخدام المعلومات الكامنة في خط السقف، والذي يمكن أيضًا توضيحه في شكل جدولي في جدول عنق الزجاجة. العمود الأول من هذا الجدول يظهر مستويات النتيجة المختلفة، بينما يعرض العمود التالي المستويات الحرجة المقابلة للشرط؛ أي المستويات التي يجب تلبيتها لتحقيق النتيجة (كما سنوضح في الأقسام التالية). وبالتالي، يعتمد التقييم النهائي لأهمية cIPMA على المستوى المرغوب للبناء المستهدف أو النتيجة (أي، المستوى)، الذي يحدده الباحث. من أجل تحليل الأداء، نقترح أن تشير CIPMA إلى عدد الحالات التي تقل عن المستويات اللازمة، مما سيعطي الأولوية لتحديد الشروط الضرورية.

4. إرشادات لتحليل خريطة الأهمية والأداء المدمجة

قدم ريشتر وآخرون (2020) إرشادات لدمج PLS-SEM مع NCA، والتي سنقوم بتوسيعها لتناسب cIPMA (انظر أيضًا ريشتر وآخرون، 2023ب). يوضح الشكل 3 إرشاداتنا الموسعة، حيث سنناقش بالتفصيل الخطوات الإضافية (المطبوعة بالخط العريض). تتعلق هذه الخطوات الإضافية بالتحقق من المتطلبات الإضافية في الخطوة 4، وتشغيل cIPMA ونقل درجات المتغيرات الكامنة في الخطوة 5، والإعدادات المحددة لتشغيل NCA في الخطوة 6، وتفسير موسع للنتائج في الخطوة 8. يرجى ملاحظة أننا نقوم أيضًا بتحديث بعض الخطوات الموضحة في ريشتر وآخرون (2020) (المعلمة بـ * في الشكل 3).

4.1. مقاييس قياس المؤشرات (الخطوة 2)

عادةً ما تعتمد IPMA على درجات المتغيرات الكامنة غير المعيارية، والتي يتم إعادة قياسها على مقياس من 0 إلى 100 لتسهيل تفسير أداء البنى. استخدام درجات المتغيرات الكامنة المعاد قياسها في NCA سيوفر نفس أحجام تأثير الضرورة كما عند تطبيق الطريقة على درجات المتغيرات الكامنة (المعيارية أو غير المعيارية) من تحليل PLS-SEM، طالما أن المؤشرات المستخدمة لقياس بنية ما مقاسة على نفس المقياس. لهذا السبب، نوصي بتطبيق cIPMA على النماذج التي يعتمد فيها كل بناء على مؤشرات تقاس على نفس (المقياس النظري). أي أنه بينما قد تستند مؤشرات بناء واحد فقط إلى مقياس ليكرت من 7 نقاط، قد تعتمد مؤشرات بناء آخر فقط على مقياس تقييم من 5 نقاط. يجب على الباحثين عدم استخدام مؤشرات بمقاييس مختلفة في نموذج القياس لبناء واحد عند تشغيل CIPMA.
الشكل 2. الشروط الضرورية الثلاثية.

4.2. التحقق من المتطلبات الإضافية عند تقييم نماذج القياس (الخطوة 4)

في سياق PLS-SEM، تتطلب IPMA أن تكون تقديرات الأوزان الخارجية إيجابية، بغض النظر عما إذا كان نموذج القياس محددًا بشكل تكويني أو انعكاسي. إذا كانت الأوزان الخارجية سلبية، فلن تقع درجات المتغيرات الكامنة (المعاد قياسها) ضمن النطاق (لكن سيكون، على سبيل المثال، بين -5 و 95) (رينجل وسارستيدت، 2016). إذا كان الوزن الخارجي لمؤشر ما سلبياً وغير متوقع وذو دلالة إحصائية، يُنصح الباحثون بفحص المؤشر ومقياسه. يجب ألا يكون للمقياس أو السؤال اتجاه مختلف (مثل العناصر ذات المقياس العكسي) مقارنةً بمؤشرات النموذج القياسي الأخرى. إذا كانت الأوزان الخارجية السلبية غير دالة، يجب على الباحثين النظر في إزالتها. أخيراً، قد تكون الأوزان الخارجية السلبية نتيجة لارتفاع التداخل بين المؤشرات، والذي تشير إليه قيم عامل التضخم في التباين التي تبلغ 5 أو أكثر (هير وآخرون، 2019). في هذه الحالة، يجب على الباحثين مرة أخرى النظر في إزالة المؤشرات. ومع ذلك، يجب عليهم ملاحظة أن إزالة المؤشرات من النماذج القياسية تعني أخذ المزيد من الاعتبارات في الحسبان، وهو ما يشرحه هير وآخرون (2022، الفصل 5) بمزيد من التفصيل.

4.3. قم بتشغيل IPMA ونقل درجات المتغيرات الكامنة (الخطوة 5)

يجب على الباحثين الاعتماد على الحد الأدنى والحد الأقصى النظري لدرجات مقاييس المؤشرات عند إجراء تحليل الأهمية النسبية للمتغيرات (IPMA)؛ على سبيل المثال، 1 و5 على مقياس يتراوح من 1 إلى 5. بهذه الطريقة، يمكن تفسير درجات المتغيرات الكامنة في شكل قيم نسبية من 0 إلى 100. إذا أجاب أحد المستجيبين، على سبيل المثال، بـ 3 على مقياس يتراوح من 1 إلى 5، فإن إعادة القياس تشير إلى أداء متوسط للصفة التي عبرت عنها المؤشر. بعد إجراء IPMA، يجب تصدير درجات المتغيرات الكامنة المعاد قياسها إلى ملف بيانات جديد يعمل كمدخل لتحليل الشبكة (NCA). إذا كان الهدف هو توسيع IPMA لاحقًا إلى مستوى المؤشر، نوصي بأن يقوم الباحثون أيضًا بتصدير بيانات المؤشرات المعاد قياسها.

4.4. قم بتشغيل NCA (الخطوة 6)

في الخطوة 6، يجب على الباحثين إجراء تحليل الشروط الضرورية (NCA) لتحديد الشروط الضرورية المحتملة. يمكن إجراء NCA باستخدام حزمة R مجانية (Dul، 2022). بدلاً من ذلك، يدعم برنامج SmartPLS 4 (Ringle et al.، 2022) أيضًا NCA (انظر أيضًا Cheah et al.، 2023؛ Richter et al.، 2023b؛ Hair et al.، 2024، الفصل 4؛ للحصول على إرشادات أساسية حول كيفية إجراء وتقرير نتائج NCA، انظر أيضًا Dul et al.، 2023). يعتمد NCA على درجات المتغيرات الكامنة للبناء الداخلي (البنايات) وبناياتها السابقة المباشرة، بالإضافة إلى الإمكانية.
درجات المؤشر إذا كانت هناك نماذج قياس تكوينية (كما أوضح ريشتر وآخرون، 2020؛ ريشتر وآخرون، 2023ب؛ هير وآخرون، 2024، الفصل 4). يمكن أن تشير NCA إلى النطاق التجريبي أو النظري لبيانات المدخلات. في سياق NCA، التوصية العامة هي استخدام النطاق التجريبي، حيث إنه ينتج تقديرًا أكثر تحفظًا لحجم التأثير (دول، 2020، الفصل 4) – نحن نتفق مع هذه التوصية. أي أننا نستخدم درجات المتغير الكامن المعاد قياسها من IPMA في PLS-SEM كبيانات مدخلات ونشير إلى نطاقها التجريبي (الذي قد لا يكون بالضرورة من 0 إلى 100).
إنشاء جداول الاختناق يساعد الباحثين في تقييم أهمية وأداء الشروط الضرورية. جدول الاختناق هو تمثيل جدولي لخط السقف. العمود الأول من هذا الجدول يعرض مستويات مختلفة من النتيجة، بينما يعرض العمود أو الأعمدة التالية المستويات المقابلة للشرط أو الشروط. يمكن عرض مستويات النتيجة ومستويات الشرط (أو الشروط) بطرق مختلفة (Dul، 2021a، الفصل 4.3). حزمة NCA في R تنتج (كإعداد افتراضي) جداول اختناق حيث، على سبيل المثال، يتم عرض كل من النتيجة والشرط (أو الشروط) كنطاقات نسبية، مع مستوى 0% يتوافق مع أدنى القيم الملاحظة ومستوى إلى أعلى القيم الملاحظة (أي أنها تشير إلى نطاق القيم التجريبي). بديل آخر هو عرض النتيجة والشرط (أو الشروط) من حيث قيمها الفعلية. القيم الفعلية هي تلك المستخدمة كمدخلات للتحليل؛ أي، إذا استخدم الباحثون درجات المتغيرات الكامنة المعاد قياسها، فإن هذه هي القيم الفعلية. لاحظ أن استخدام نطاقات النسب المئوية والقيم الفعلية سيقدم نفس النتائج عندما يستخدم الباحثون درجات المتغيرات الكامنة المعاد قياسها مع حد أدنى فعلي قدره 0 وحد أقصى قدره 100.
من خلال الإشارة إلى هذه المستويات الضيقة، يمكن للباحثين اختيار مستوى نتيجة محدد وتحديد أي مستوى من الحالة ضروري لتحقيقه. يمكن أيضًا استخدام جدول الاختناق لتحديد أداء البنى السابقة من منظور منطق الضرورة. لذلك، يجب عرض جدول الاختناق كنسب مئوية للبنى السابقة، والتي توفر النسبة وعدد الحالات التي لا تصل إلى المستوى المطلوب لمستوى النتيجة المقابل.
عادةً ما تنتج NCA جداول عنق الزجاجة تحتوي على عشرة خطوات متساوية المسافة عندما، على سبيل المثال، تستخدم نطاقات النسب المئوية من 0 إلى ، 20 وما إلى ذلك). من أجل تقديم تفسير مشترك للنتائج، نوصي بتحديد مستوى أداء معين يهم المتغيرات التابعة أو المستهدفة التي سيتم تقييم مستويات المتغيرات السابقة لها بمزيد من التفصيل. يمكن للباحثين تحديد مستوى الأداء المرغوب للمتغير المستهدف أو المتغيرات المستهدفة من خلال الاستناد إلى المعرفة الخبيرة، أو المعايير النموذجية في سياق البحث، أو الأهداف الطموحة التي تشكل خلفية المشروع. على سبيل المثال، في العمل
الخطوة 1: تحديد هدف البحث والخلفية النظرية
حدد الفرضيات وفقًا لمنطق الكفاية والضرورة (للحالة الأخيرة، انظر بوكرا نتز ودول، 2023؛ ريشتر وهاوف، 2022).*
الخطوة 2: إعداد البيانات والتحقق منها
حجم العينة: اتبع الإرشادات المتعلقة بحجم العينة كما هو موضح في سياق PLS-SEM، على سبيل المثال، من خلال الرجوع إلى جداول القوة المنشورة (Hair et al.، 2022، الفصل 1).
توزيع البيانات: تقرير المعلومات حول توزيع البيانات (Hair وآخرون، 2012).
القيم الشاذة: قم بإجراء تحليل للقيم الشاذة من خلال اتباع الإرشادات الشائعة، مثل فحص الملاحظات التي تظهر قيمة zscore > 3 (سارستيد وموي، 2019، الفصل 5).*
مستوى القياس/ترميز المقاييس: استخدم بيانات مترية أو شبه مترية (أي، مقاييس فترية، مثل مقاييس ليكرت)؛ تأكد من أن اتجاه المقياس/الترميز يتوافق مع العلاقات المفترضة نظريًا، وإلا عد إلى المقياس أو قم بعكسه. تأكد من أن المؤشرات المستخدمة لقياس بناء ما تعتمد على نفس المقياس (على سبيل المثال، جميع المؤشرات على مقياس من 1 إلى 5).
الخطوة 3: قم بتشغيل تحليل PLS-SEM
استخدم PLS-SEM لتقدير درجات المتغيرات الكامنة، وعلاقات النموذج الهيكلي، وأهميتها (هار وآخرون، 2022، الفصل 6).
الخطوة 4: تقييم موثوقية وصلاحية نماذج القياس
استخدم إرشادات التقييم في سياق PLS-SEM لتقييم جودة نماذج القياس (مثل، غونتر وآخرون، 2023؛ هير وآخرون، 2022، الفصول 4 و5؛ سارستيد وآخرون 2021). فيما يتعلق بالبناءات الانعكاسية، قم بتقييم تحميلاتها، وكرونباخ. / موثوقية مركبة / متوسط التباين المستخرج، ونسبة التباين المتباين. فيما يتعلق بالبناءات التكوينية، قم بإجراء تحليل التكرار، وتقييم عوامل تضخم التباين، وأهمية وملاءمة أوزان المؤشرات. إذا لزم الأمر، قم بإجراء تحسينات.
تحقق من المتطلبات الإضافية: يجب أن تكون تقديرات الأوزان الخارجية إيجابية في جميع نماذج القياس (أي، النماذج التكوينية والانعكاسية). في حالة الأوزان الخارجية السلبية غير المهمة، و/أو وجود VIF بمقدار 5 أو أعلى في نماذج القياس، يمكن للباحثين، بالنظر إلى الاعتبارات في Hair et al. (2022، الفصل 5)، أن يفكروا في إزالة المؤشر.
الخطوة 5: قم بتشغيل IPMA ونقل درجات المتغيرات الكامنة
قم بتشغيل IPMA في PLS-SEM باستخدام القيم النظرية الدنيا والقصوى لمقاييس المؤشرات كمدخلات (رينجل وسارستيد، 2016).
قم بتصدير درجات المتغيرات الكامنة المعاد قياسها (وإذا رغبت، بيانات المؤشرات المعاد قياسها) إلى ملف جديد واستورد ملف البيانات الجديد إلى R أو SmartPLS.
الخطوة 6: تشغيل NCA
قم بتشغيل تحليل NCA في SmartPLS (انظر أيضًا Richter et al.، 2023b) أو R (الإعدادات الافتراضية؛ أي، استخدم النطاق التجريبي لبيانات الإدخال؛ 10,000 تكرار) (Dul، 2020، الفصل 4). قم بتشغيل التحليل للعلاقات المفترضة أو جميع العلاقات: المعتمد = درجات المتغيرات الكامنة المستجيبة، المستقل = درجات المتغيرات الكامنة السابقة. اختر خط السقف بناءً على النظرية أو نمط البيانات (Dul، 2020، الفصل 4).* راجع النطاق التجريبي لبيانات الإدخال.
إنشاء جدول عنق الزجاجة: اختر المستوى المرغوب فيه للبناء المستهدف (استنادًا إلى معرفتك في هذا المجال) وتأكد من أن جدول عنق الزجاجة يحدد هذا المستوى.
الخطوة 7: تقييم النموذج الهيكلي
بعد تقييم قيم VIF للنموذج الداخلي، قم بتقييم نموذج PLS-SEM وفقًا لمعايير التقييم القياسية، وخاصة معامل التحديد ( )، دلالة معامل المسار الإحصائية وأهميتها، والقوة التنبؤية (Guenther et al.، 2023؛ Hair et al.، 2022، الفصل 6؛ Sarstedt et al. 2021؛ Sharma et al.، 2023؛ Shmueli et al.، 2019).*
قم بفحص بصري لمخطط التشتت لنتائج NCA؛ تحقق من القيم الشاذة كاختبار موثوقية محدد لـ NCA (Dul، 2021a، الفصل 3.8 والفصل 4.6). في حالة استبعاد القيم الشاذة، عد إلى الخطوة 3؛ وإلا، تابع.
قيم حجم تأثير الضرورة وأهميتها الإحصائية (دول، 2016؛ 2020، الفصل 4).
الخطوة 8: تفسير النتائج
قم بإنتاج مخطط IPMA موسع: ضع البنى بناءً على التأثيرات الإجمالية المعيارية على -محور (الأهمية) وأداء متوسط درجات المتغيرات الكامنة على -محور (الأداء)؛ استخدم المعلومات من NCA للإشارة إلى ما إذا كان البناء ضروريًا أم لا (على سبيل المثال، باستخدام رمز لون)، وللإشارة إلى كيفية أدائها من حيث الضرورة (على سبيل المثال، عن طريق تعديل حجم البنى السابقة الموضوعة). يمكن للباحثين استخدام قالب Microsoft Excel الخاص بنا لهذا الغرض (https://www.pls-sem.net/downloads/).
ملاحظة: * تشير إلى أن الخطوات تم تحديثها مقارنةً بـ ريشتر وآخرون (2020)
الشكل 3. دليل موسع خطوة بخطوة في سياق IPMA المدمجة (cIPMA).
ملاحظة: *تشير إلى أن الخطوات تم تحديثها مقارنةً بـ ريشتر وآخرون (2020).
دراسات الرضا، قد يكون هذا هو درجة رضا الموظفين عن العمل بمعدل 70 (على مقياس من 0 إلى 100) (كما في هاف وآخرون، 2015)، بينما في مشروع تسويقي حول رضا العملاء قد يكون هذا مستوى مرجعي يبلغ 75 أو أكثر، اعتمادًا على الصناعة (على مقياس من 0 إلى 100) (مؤشر رضا العملاء الأمريكي، 2022؛ ريجدون وآخرون، 2010). اعتمادًا على هذه المستويات، قد يحتاج الباحثون إلى تغيير عرض الخطوات في جداول الاختناق الخاصة بهم لإظهار مستوى الأداء المحدد للبناء المستهدف (على سبيل المثال، زيادة عدد الخطوات من 10 إلى 20 ستوفر معلومات حول إلخ. المستويات).

4.5. تفسير النتائج (الخطوة 8)

تستفيد تفسير النتائج من كل من نتائج PLS-SEM المتعلقة بأهمية القيم والأداء للمتغيرات، ونتائج NCA المتعلقة بالشروط الضرورية المقابلة. لتسهيل التفسير، ينبغي على الباحثين إنتاج شكل يوضح أهمية المتغيرات السابقة ومتوسط أدائها – في تحليل الأثر القائم على PLS-SEM – بالإضافة إلى تقديم معلومات حول ما إذا كانت هذه المتغيرات السابقة شروطًا ضرورية لمستوى النتيجة المختارة، وإذا كان الأمر كذلك، كيف تؤدي. الشكل 4 يظهر مثالاً على مثل هذا التوضيح الرسومي. يوضح الرسم البياني نتائج تحليل المسار الهيكلي القائم على PLS-SEM؛ أي التأثيرات الكلية للبناءات السابقة على البناء المستهدف (الأهمية) على -محور ومتوسط درجاتهم المعاد قياسها (الأداء) على -المحور. بالإضافة إلى ذلك، يميز الرسم البياني بين البنى الضرورية وغير الضرورية؛ أي، البنى التي تظهر على الأقل حجم تأثير متوسط ولها دلالة إحصائية وتلك التي لا تظهر ذلك. تُعرض البنى غير الضرورية لتحقيق المستوى المطلوب للبنية المستهدفة كدوائر سوداء، بينما تُعرض البنى الضرورية كدوائر بيضاء. بالنسبة للمثال الموضح في الشكل 4، نفترض مستوى مطلوب للبنية المستهدفة من . بعد ذلك، بناءً على المستوى المطلوب للبناء المستهدف، يستخدم الرسم البياني معلومات جدول الاختناق لتحديد النسبة المئوية للحالات المتبقية التي تقل عن المستويات المطلوبة للبناءات السابقة (المقدمة في عرض النسبة المئوية). تعكس حجم الفقاعات لكل من البناءات السابقة التي تعتبر شروطًا ضرورية هذه المعلومات. كلما كانت الفقاعات أكبر، كانت النسبة المئوية للحالات التي لم تحقق المستوى المطلوب للشروط الضرورية أكبر. وبالتالي، تشير الفقاعات الكبيرة إلى أنه من منظور الضرورة، يجب على الباحثين توجيه انتباههم إلى هذا الجانب. عند فحص النتائج في الشكل 4، نجد أن ليس شرطًا ضروريًا لتحسين البناء المستهدف، بينما و هي. كما أن
الشكل 4. IPMA المدمجة (cIPMA).
ملاحظة: توسيع المثال المعطى في الشكل 1، اللوحة ب. بالنسبة للمستوى المطلوب من النتيجة: – البناء غير ضروري؛ البناء ضروري. تمثل أحجام الفقاعات النسبة المئوية للحالات التي لم تحقق المستوى المطلوب من النتائج المرغوبة.
يصبح واضحًا أن العديد من الحالات لم تحقق المطلوب المستوى. هذا يجعل ذو صلة عالية، حيث إن الاستثمارات في هياكل أخرى لن تزيد من النتيجة في هذه الحالات ما لم يكن تم حل عنق الزجاجة. تحسين لذا فإن أداء ‘ مهم جدًا في هذه الحالة لتمكين البناء المستهدف الأداء المطلوب.
من خلال دمج المعلومات من NCA، وبشكل خاص من جداول الاختناق، يمكّن cIPMA من تفسير غني للنتائج، والتي نميزها في الشكل 5. على سبيل المثال، فإن البنى المرسومة في الزاوية العليا اليمنى – على الرغم من أهميتها العالية – غالبًا ما تكون خارج نطاق تركيز الباحث، حيث إنها قد حققت بالفعل مستوى أداء عالي. ومع ذلك، إذا كانت نتائج NCA تشير إلى أن هذه البنى لم تحقق مستوى ضرورة كافٍ، يجب أن تظل تحسيناتها أولوية. وبالمثل، فإن البنى الموجودة في الربع السفلي الأيسر (أي تلك التي تتمتع بأهمية وأداء منخفضين) لا ينبغي اعتبارها غير ذات صلة بحد ذاتها. إذا لم تحقق هذه البنى مستوى ضرورة مرغوب، يجب اتخاذ إجراءات لتحسينها. تفسير الحالات الأخرى في هذه الأرباع (مثل البنى التي تحقق مستوى ضرورة مرضٍ) وفي الأرباع الأخرى هو مشابه.

5. مثال توضيحي لتحليل خريطة الأهمية والأداء المدمجة

بينما تعتبر نماذج مختلفة مناسبة لتكون مثالاً (مثل، لي وآخرون، 2024؛ سيال وآخرون، 2024)، نوضح تطبيق cIPMA من خلال الاستناد إلى نسخة موسعة من نموذج قبول التكنولوجيا (Davis، 1989)، الذي كان بمثابة مخطط للبحث في سلوك المستهلك في سياقات التسويق وسلوك المستهلك المختلفة. قدم ريشتر وآخرون (2020) النموذج المفاهيمي والحجج والنظريات ذات الصلة في عرضهم للاستخدام المشترك لـ PLS-SEM وNCA. كما أن البيانات وتحليل PLS-SEM موثقة جيدًا. وبالتالي، لا نقدم جميع الخطوات بالتفصيل، ولكن نركز على الجوانب ذات الصلة بـ cIPMA.

5.1. النموذج المفاهيمي، البيانات، وتحليلات PLS-SEM (الخطوات 1 إلى 3)

النموذج المعني يتضمن بنائين داخليين: النية السلوكية لتبني تقنية، والتي تؤدي إلى الاستخدام الفعلي للتقنية (على سبيل المثال، أجن، 1991؛ ديفيس وآخرون، 1989؛ شابارد وآخرون، 1988؛ تيرنر وآخرون، 2010). يحتوي على أربعة بنى خارجية رئيسية تسبق النية السلوكية واستخدام التقنية: التوافق، الذي يوضح مدى ملاءمة الابتكار مع نمط حياة العميل وقيمه، والفائدة المدركة وسهولة الاستخدام المدركة، والقيمة العاطفية، التي تقيس ما إذا كان العملاء يستمتعون أو يشعرون بمشاعر إيجابية عند استخدام منتج ما (شيث وآخرون، 1991). الشكل 6 يشير إلى المفهوم
الشكل 5. التوصيات الإدارية استنادًا إلى تحليل الأهمية والأداء المشترك (cIPMA).
الشكل 6. النموذج المفاهيمي.

نموذج.

استخدمنا عينة من متبني أجهزة قراءة الكتب الإلكترونية في فرنسا ) وجمعت الردود من خلال استبيان عبر الإنترنت. قمنا بتنفيذ عنصر واحد ومقياس ليكرت من 7 نقاط لقياس البناء المستهدف استخدام التكنولوجيا؛ بينما استخدمت البنى المتبقية نماذج قياس عاكسة مع عناصر مقاسة على مقاييس ليكرت من 5 نقاط (لمزيد من المعلومات حول العينة والإحصائيات الوصفية، انظر الجدول A1 في الملحق وRichter et al.، 2020). أي أن المؤشرات المستخدمة لقياس بناء معين مقاسة جميعها على نفس المقياس. مجموعة البيانات متاحة مجانًا عبر Richter وHauff وKolev وSchubring (2023a). استخدمنا البيانات لتقدير TAM الموسع بواسطة برنامج SmartPLS 4 (Ringle et al.، 2022) – لمزيد من التفاصيل حول تقدير النموذج، انظر Richter et al. (2023b).

5.2. تقييم موثوقية وصلاحية نماذج القياس (الخطوة 4)

أظهرت نماذج القياس والبنية جودة إحصائية مناسبة عند تقييمها وفقًا لمعايير التقييم القياسية (انظر الجدول A2 في الملحق وRichter et al.، 2020). لإنهاء الخطوة 4، نحتاج إلى إجراء فحص إضافي لمتطلبات IPMA. نقوم بذلك من خلال تقييم ما إذا كانت الأوزان الخارجية في نموذجنا إيجابية، وهو ما هو عليه (الجدول A2 في الملحق). لذلك نستمر في التحليلات دون إجراء تغييرات إضافية.

5.3. قم بتشغيل IPMA ونقل درجات المتغيرات الكامنة (الخطوة 5)

نقوم بتشغيل تحليل الأهمية النسبية للأداء (IPMA) في نمذجة المعادلات الهيكلية باستخدام برنامج SmartPLS 4 (رينجل وآخرون، 2022). نظرًا لأن جميع المستجيبين استخدموا نطاق المقياس بالكامل على مستوى المؤشر، فإن المقاييس التجريبية والنظرية على مستوى المؤشر متطابقة. لذلك، لا نحتاج إلى تعديل القيم الدنيا والعليا النظرية للمقياس في IPMA. وفقًا لإرشاداتنا، قمنا بتحديد التحليل بحيث ينتج درجات متغيرة كامنة معادلة على مقياس من 0 إلى 100. بعد تشغيل IPMA، نقوم بإنشاء ملف بيانات جديد مع الدرجات المعادلة لتحليلنا.

5.4. قم بتشغيل NCA (الخطوة 6)

نستخدم درجات المتغيرات الكامنة المعاد قياسها لتشغيل تحليل الحاجة الشرطية (NCA) على درجات البناء التي تم إنشاؤها في تحليل المعادلات الهيكلية باستخدام أسلوب المربعات الصغرى (PLS-SEM). في تحليل الحاجة الشرطية، نستخدم النطاق التجريبي (لتشغيل تحليل الحاجة الشرطية في R، انظر Dul (2021b)؛ يرجى ملاحظة أن تحليل الحاجة الشرطية نفسه هو تحليل ثنائي المتغيرات غير متأثر بالعلاقات الأخرى التي تم تحليلها). نقوم بإنشاء جداول عنق الزجاجة لتحديد المستويات الضرورية للبناءات السابقة التي يجب تلبيتها. نظرًا لأننا نريد تقييم نتائجنا بناءً على مستوى استخدام التكنولوجيا البالغ 85 (انظر الخطوات 7 و8)، فإننا نزيد عدد الخطوات إلى 20 من أجل إنتاج جداول عنق الزجاجة التي تعرض هذا المستوى من النتائج. نقوم بإنشاء جدولين عنق زجاجة (استنادًا إلى تقنية خط السقف CE-FDH)، أحدهما مع
القيم الفعلية، التي توفر المستويات اللازمة للبنى السابقة التي تحتاج مستويات استخدام التكنولوجيا المختلفة لتحقيقها، وواحدة مع النسب المئوية للظروف، التي تخبرنا عن عدد ونسبة الحالات التي لم تحقق المستويات المطلوبة للبنى السابقة للمستويات المقابلة لاستخدام التكنولوجيا (انظر الخطوة التالية).

5.5. تقييم النموذج الهيكلي (الخطوة 7)

في هذه الخطوة، نقوم أولاً بتقييم النتائج من تحليل PLS-SEM، بما في ذلك IPMA. بعد ذلك، نقوم بتقييم نتائج NCA. نحن نوضح تقييم البنية المستهدفة لاستخدام التكنولوجيا.
تحليل التأثيرات الكلية المعيارية من تحليل PLS-SEM (الجدول 1) يظهر أن نية التبني لها أقوى تأثير ملحوظ على استخدام التكنولوجيا (0.437)، تليها القيمة العاطفية (0.362). التأثيرات الكلية لبقية المتغيرات ليست ذات دلالة. الأهمية المتوسطة للمتغيرات هي 0.225. التباين المفسر لاستخدام التكنولوجيا هو .
تظهر نتائج IPMA متوسط درجات الأداء للمتغيرات الكامنة المعاد قياسها (الجدول 2). عند التركيز على البنى السابقة المباشرة لاستخدام التكنولوجيا، نجد أن التوافق لديه أدنى متوسط أداء (61.557)، بينما سهولة الاستخدام لديها أعلى متوسط (75.640). الحد الأدنى لدرجة الأداء لهذا البناء (المحدد بالحالة) هو 16.871، مما يشير إلى أن أي مستجيب لم يقيم جميع مؤشرات سهولة الاستخدام عند أدنى مستوى. عند النظر إلى البنى السابقة مجتمعة، يكون متوسط الأداء 68.731. عند فحص نتائج IPMA بشكل كامل، نتعلم أن جميع البنى تؤدي بمستويات متشابهة إلى حد ما، حيث أن نية التبني والقيمة العاطفية لهما أكبر أهمية لاستخدام التكنولوجيا.
بعد ذلك، ننتقل إلى نتائج NCA. أولاً، نقوم بفحص الرسوم البيانية المتناثرة بصريًا. نجد أن جميع الرسوم البيانية المتناثرة تظهر مساحة فارغة في الزاوية العليا اليسرى، مما يشير إلى شروط ضرورية محتملة (انظر الشكل A1 في الملحق). لا تشير أي من الرسوم البيانية المتناثرة إلى حالات فردية لها تأثير خاص على خط السقف (الاستثناءات العليا) أو النطاق (الاستثناءات النطاقية).
بعد ذلك، نعتبر أحجام تأثير الضرورة أهمية وحجم بالنسبة لاستخدام تكنولوجيا البناء المستهدف. اتبعنا ريشتر وآخرون (2020) ونشير إلى خط السقف CE-FDH (الجدول 3). نلاحظ أن اختيار خط السقف (CE-FDH مقابل CR-FDH) يجب مناقشته بناءً على (أ) مقياس البيانات (بيانات متقطعة مقابل بيانات مستمرة)، (ب) نمط الملاحظات بالقرب من خط السقف (غير منتظم مقابل خطي)، و (ج) الأفكار النظرية المحتملة (تشير إلى خط سقف مستقيم، انظر دول، 2020، الفصل 4). نجد أن جميع أحجام التأثير ذات دلالة إحصائية عند وعلى الأقل بحجم متوسط.
الجدول 1
إجمالي التأثيرات من تحليل PLS-SEM.
بناء استخدام التكنولوجيا ذو دلالة إحصائية
التأثيرات الكلية فترات الثقة بوساطة 95%
القيمة العاطفية 0.362 [0.200; 0.521] نعم
سهولة الاستخدام 0.049 [-0.110; 0.215] لا
الفائدة المدركة 0.149 [-0.075; 0.357] لا
التوافق 0.127 [-0.108; 0.365] لا
نية التبني 0.437 [0.268; 0.609] نعم
الجدول 2
قيم الأداء للمتغيرات الكامنة (معاد ضبطها).
بناء قيمة الأداء، متوسط قيمة الأداء، الحد الأدنى قيمة الأداء، القصوى
استخدام التكنولوجيا ٤٩.٧١٣ 0.000 ١٠٠.٠٠٠
نية التبني 72.041 0.000 ١٠٠.٠٠٠
القيمة العاطفية 70.171 0.000 ١٠٠.٠٠٠
سهولة الاستخدام 75.640 16.871 ١٠٠.٠٠٠
الفائدة المدركة 64.248 0.000 ١٠٠.٠٠٠
التوافق ٦١.٥٥٧ 0.000 ١٠٠٫٠٠٠
الجدول 3
أحجام تأثير الضرورة (خط سقف CE-FDH).
بناء استخدام التكنولوجيا
حجم التأثير d -قيمة
القيمة العاطفية 0.33 0.000
سهولة الاستخدام 0.24 0.016
الفائدة المدركة 0.24 0.001
التوافق 0.21 0.000
نية التبني 0.29 0.000
الجدول 4
استخدام تقنية جدول الاختناق، القيم الفعلية (استنادًا إلى درجات المتغيرات الكامنة في PLS-SEM المعاد قياسها من 0 إلى 100).
استخدام التكنولوجيا نية التبني القيمة العاطفية سهولة الاستخدام الفائدة المدركة التوافق
القيم الفعلية (معادلة من 0 إلى 100)
0 NN NN NN NN NN
٥ NN NN 25.373 NN NN
10 NN NN 25.373 NN NN
15 NN NN 25.373 NN NN
20 NN NN 25.373 NN NN
٢٥ NN NN 25.373 NN NN
30 NN NN 25.373 NN NN
٣٥ ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 25.518
40 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 25.518
٤٥ ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 25.518
50 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 25.518
٥٥ ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 ٣٣.٦٩٦
60 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٤٦٩ 15.702 ٣٣.٦٩٦
65 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣٫٤٦٩ 15.702 ٣٣.٦٩٦
70 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٨٧٥ ٤٨.١٢٧ ٣٣.٦٩٦
75 ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٨٧٥ ٤٨.١٢٧ ٣٣.٦٩٦
٨٠ ٣٣.٨٣٣ ٤٩.٦٥٠ ٣٣.٨٧٥ ٤٨.١٢٧ ٣٣.٦٩٦
85 75.000 ٤٩.٦٥٠ 66.904 66.212 ٣٣.٦٩٦
90 75.000 ٤٩.٦٥٠ 66.904 66.212 ٣٣.٦٩٦
95 75.000 ٤٩.٦٥٠ 66.904 66.212 ٣٣.٦٩٦
100 75.000 ٤٩.٦٥٠ 66.904 66.212 ٣٣.٦٩٦
استنادًا إلى الدرجات المقدمة في الجدول 4، نقوم بعد ذلك بتحديد مستويات القيمة لمكوناتنا السابقة التي يجب تلبيتها لتحقيق مستوى مرغوب من استخدام التكنولوجيا. فيما يتعلق بتحليلنا، نفترض مستوى نتيجة مرغوب من استخدام التكنولوجيا يبلغ 85، وهو معيار متحفظ إلى حد ما (أي مرتفع). المستويات المقابلة هي 75 لنية التبني، 50 للقيمة العاطفية، 67 لسهولة الاستخدام، 66 للفائدة المدركة، و34 للتوافق. أي أنه لتحقيق درجة استخدام تكنولوجيا تبلغ 85 (على مقياس من 0 إلى 100)، نحتاج إلى تحقيق درجة 75 لنية التبني (على مقياس من 0 إلى 100)، و50 للقيمة العاطفية (على مقياس من 0 إلى 100)، إلخ.
بالإضافة إلى ذلك، في الجدول 5 نحدد النسبة المئوية للحالات التي لا تحقق المستويات المطلوبة للبنى السابقة. على سبيل المثال، لم تحقق جميع الحالات المستوى الضروري من نية التبني لتمكين درجة استخدام التكنولوجيا من 85 (على مقياس من 0 إلى 100). النسب المئوية المقابلة للعوامل الأخرى هي لـ
القيمة العاطفية لسهولة الاستخدام، للمنفعة المدركة، و للتوافق. في ما يلي، سيتم استخدام هذه المعلومات لتوسيع نتائج IPMA الأولية.

5.6. تفسير النتائج (الخطوة 8)

فيما يلي، نجمع النتائج من الجداول 1 و 2 و 5 لتصميم خريطة أهمية-أداء مجمعة تُظهر أهميتها في شكل التأثيرات الكلية لـ PLS-SEM، وكذلك أدائها المتوسط في شكل متوسط درجات المتغيرات الكامنة المعاد قياسها لنية التبني، والقيمة العاطفية، وسهولة الاستخدام، والفائدة المدركة، والتوافق كما تم الحصول عليها من تحليلات PLS-SEM. بالإضافة إلى ذلك، تُظهر خريطة الأهمية-الأداء المجمعة ما إذا كانت البنى السابقة شروطًا ضرورية لاستخدام التكنولوجيا أم لا، وإذا كانت ضرورية، كم عدد الحالات التي لا تحقق المستويات المطلوبة. الشكل 7 يُظهر خريطة الأهمية-الأداء المجمعة.
تماشياً مع توضيحاتنا السابقة، نجد أن نية التبني، التي تقع في الزاوية العليا اليمنى من الخريطة، مهمة للغاية وتظهر بالفعل أداءً عالياً. ومع ذلك، لا تزال نية التبني بحاجة إلى أولوية، حيث إنها الشرط الضروري للبناء المستهدف، وهو استخدام التكنولوجيا. بشكل أكثر تحديداً، لم تحقق العديد من الحالات المستوى المطلوب من نية التبني. علاوة على ذلك، فإن الفائدة المدركة وسهولة الاستخدام، التي تقع في الربع الأيسر، تتطلب أولوية على الرغم من أنها ذات أهمية قليلة وتؤدي بالفعل بشكل جيد نسبيًا. كلاهما شرط ضروري، لكن العديد من الحالات لا تحقق المستويات المطلوبة. للمنفعة المدركة، لتسهيل الاستخدام). إن الفشل في إعطاء الأولوية لهذه العناصر، التي قد تبدو غير مهمة للوهلة الأولى بالنظر إلى نتائج PLS-SEM، سيحد من نجاح الأنشطة الأخرى المتعلقة بتحسين استخدام التكنولوجيا. يجب معالجة هذه المجالات، لأنها شروط ضرورية. من ناحية أخرى، فإن القيمة العاطفية والتوافق هما أيضًا شروط ضرورية، ولكن القليل من هذه الحالات لا تحقق المستويات المطلوبة (أي ، و وبالتالي، فإن هذه البنى تعتبر أقل أهمية نسبيًا في بيئة عملية.

6. الخاتمة

في هذه الورقة، قدمنا فهمًا معززًا للأهمية والأداء يجمع بين الرؤى من PLS-SEM وNCA. تقدم ورقتنا عدة مساهمات مفيدة للباحثين والممارسين الذين يهدفون إلى الاستفادة من PLS-SEM وNCA وتحليلات الأهمية والأداء بشكل عام.
هدفنا الشامل هو خلق الوعي بمنطق مختلف للباحثين والممارسين الذين يستخدمون PLS-SEM و IPMA؛ وهو منطق الضرورة، الذي يقدم عدة طرق لتكملة النتائج المستخلصة من تفكير الكفاية في PLS-SEM. أولاً، يكمل النتائج من خلال تقديم منظور جديد حول الأهمية. قد تكون بعض البنى لها تصنيف أهمية منخفض بناءً على IPMA في PLS-SEM (أي عندما تظهر ارتباطًا نسبيًا منخفضًا مع البنية المستهدفة وبالتالي، لديها قدرة أقل على تحفيز زيادة في النتيجة المستهدفة). ومع ذلك، قد تكون هذه البنى ذات أهمية عالية من منظور الضرورة. إذا كانت هذه البنى تمثل عنق الزجاجة، فلن يتم تحقيق المستوى المطلوب منها. ثانياً، يوفر cIPMA للباحثين فهماً أعمق لمستويات الأداء ذات الصلة التي يجب تحقيقها، حيث تقدم معلومات عن عنق الزجاجة من NCA منظورًا دقيقًا حول مقدار المعلومات المطلوبة فعليًا لدفع النتيجة المستهدفة إلى المستوى التالي. هذه الرؤى تفيد الأبحاث التي تحقق في سلف النجاح في سياقات مختلفة (مثل، رضا الوظيفة، أداء الفريق، ورضا العملاء على سبيل المثال لا الحصر). بالإضافة إلى ذلك، فإن هذه الرؤى ذات قيمة عملية عالية، حيث تسمح بتحديد الإجراءات بوضوح وترتيب أولوياتها، وعلاوة على ذلك، تقدم معلومات حول مقدار الاستثمار المطلوب في تحسينات معينة. ثالثاً، لتقييم الأداء المحقق في عينة (وهو النهج العادي في سياق PLS-SEM)، نقترح تحليل عدد الحالات التي لا تزال بحاجة لتحقيق المستوى المطلوب للشرط. في حالة وجود ضرورات متعددة.
الجدول 5
استخدام تقنية جدول الاختناق: القيم الفعلية لاستخدام التقنية (استنادًا إلى درجات المتغيرات الكامنة في PLS-SEM المعاد قياسها من 0 إلى 100) والنسب المئوية للبنى السابقة.
استخدام التكنولوجيا نية التبني القيمة العاطفية سهولة الاستخدام الفائدة المدركة التوافق
القيم الفعلية (معادلة من 0 إلى 100) نسبة (ورقم) الحالات التي لا تحقق المستويات اللازمة
0 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
٥ 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
10 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
15 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
20 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
٢٥ 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
30 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
٣٥ ٤.٦ (8) ٥.٧ (10) 1.1 (2) 1.7 (3) ٥.٧ (10)
40 ٤.٦ (8) ٥.٧ (10) 1.1 (2) 1.7 (3) 5.7 (10)
٤٥ ٤.٦ (8) 5.7 (10) 1.1 (2) 1.7 (3) 5.7 (10)
50 ٤.٦ (8) 5.7 (10) 1.1 (2) 1.7 (3) ٥.٧ (10)
٥٥ ٤.٦ (8) 5.7 (10) 1.1 (2) 1.7 (3) ٨.٦ (15)
60 ٤.٦ (8) 5.7 (10) 1.1 (2) 1.7 (3) ٨.٦ (15)
65 ٤.٦ (8) 5.7 (10) 1.1 (2) 1.7 (3) ٨.٦ (15)
70 ٤.٦ (8) ٥.٧ (10) 2.9 (5) 17.2 (30) ٨.٦ (15)
75 ٤.٦ (8) 5.7 (10) 2.9 (5) 17.2 (30) ٨.٦ (15)
٨٠ ٤.٦ (8) 5.7 (10) 2.9 (5) 17.2 (30) ٨.٦ (15)
85 ٣٩.١ (68) 5.7 (10) ٢٨.٧ (50) ٤٧.١ (82) ٨.٦ (15)
90 ٣٩.١ (68) 5.7 (10) ٢٨.٧ (50) ٤٧.١ (82) ٨.٦ (15)
95 ٣٩.١ (68) 5.7 (10) ٢٨.٧ (50) ٤٧.١ (82) ٨.٦ (15)
100 ٣٩.١ (68) 5.7 (10) ٢٨.٧ (50) ٤٧.١ (82) 8.6 (15)
الشكل 7. خريطة الأهمية والأداء المدمجة لاستخدام التكنولوجيا.
في ظل هذه الظروف، يمكن أن تساعد هذه المعلومات في تحديد أولويات الإجراءات. علاوة على ذلك، تساهم دراستنا أيضًا في البحث الأوسع حول تحليلات الأهمية والأداء، التي تواجه تحدي تعريف الأهمية والأداء، بالإضافة إلى مستويات الأداء الحرجة. نعتقد أن cIPMA هو خطوة إلى الأمام في هذا الاتجاه.
بالإضافة إلى هذه المساهمات، هناك أيضًا مجالات متنوعة يمكن أن تركز عليها الأبحاث المستقبلية. أولاً، تستخدم طرق PLS-SEM وغيرها من الطرق التي تولد معلومات حول العوامل التي يجب أن تكون موجودة، قوة العلاقة بين البنى لتحديد التأثير الذي يحدثه زيادة في البنية السابقة على البنية المستهدفة. يتم توليد هذه النتائج في الغالب بالنسبة للعينة الكاملة؛ أي لجميع تركيبات البيانات التي تتضمن قيم بنى منخفضة ومتوسطة وعالية. ومع ذلك، عند دمج PLS-SEM مع NCA، قد يكون من المثير للاهتمام فهم العلاقة بين البنى المعنية عندما يتم تجاوز عنق الزجاجة المحدد. على سبيل المثال، إذا أشارت NCA إلى أن البنية السابقة ذات المستوى المتوسط حاسمة لتحقيق مستوى النتيجة العالية المرغوب، فقد يكون من المثير للاهتمام فهم العلاقة بين البنية السابقة والبنية المستهدفة في الحالات التي حققت بالفعل المستوى المتوسط المطلوب. من منطق الكفاية.
من منظور، يمكن أن يمثل هذا في النهاية زيادة في الإمكانيات المستقبلية للبناء السابق في IPMA. ثانياً، استناداً إلى نتائج PLS-SEM، تشير أهمية IPMA إلى التأثيرات الكلية لجميع البنى المعنية. بعبارة أخرى، تشمل هذه التأثيرات كل من الروابط المباشرة وغير المباشرة. هذه التأثيرات غير المباشرة أو التأثيرات الوسيطة لا تؤخذ حالياً بعين الاعتبار بشكل صريح في CIPMA المقترح. لذلك، نشجع الأبحاث المستقبلية على تطوير أفكار لتحليل علاقات الوساطة في سياق NCA، والتي يمكن استخدامها في النهاية في cIPMA.

بيان مساهمة مؤلفي CRediT

سفين هاوف: الكتابة – مراجعة وتحرير، الكتابة – المسودة الأصلية، التصور، التحقق، الإشراف، المنهجية، التحقيق، التحليل الرسمي، إدارة البيانات، التصور. نيكول فرانزيسكا ريشتر: الكتابة – مراجعة وتحرير، الكتابة – المسودة الأصلية، التصور، التحقق، الإشراف، إدارة المشروع، المنهجية، التحقيق، التحليل الرسمي، إدارة البيانات، التصور. ماركو سارستيدت: الكتابة – مراجعة وتحرير، الكتابة – المسودة الأصلية، التصور، التحقق، المنهجية، التحقيق، التصور. كريستيان م. رينغلي: الكتابة – مراجعة وتحرير، الكتابة – المسودة الأصلية، التحقق، البرمجيات، إدارة المشروع، المنهجية، التحقيق، التحليل الرسمي، إدارة البيانات، التصور.

إعلان عن تضارب المصالح

تستخدم هذه الدراسة برنامج التحليل الإحصائي SmartPLShttps://www. سمارت بلس). يعترف كريستيان م. رينجل بمصلحة مالية في SmartPLS.

توفر البيانات

البيانات المتاحة كجزء من مقال بيانات مختصرة (https://doi. org/10.1016/j.dib.2023.109190)

الملحق

الجدول A1
وصف البيانات
متغير كامن (القياس مأخوذ من) مؤشر معنى النطاق [الحد الأدنى; الحد الأقصى] س.د. الكورتوز الزائد الانحراف
القيمة العاطفية، التأملية (سوينى وسوتار، 2001) EMV_01 استمتاع ٣.٩٠٢ [1; 5] 0.842 1.942 -1.036
EMV_02 متعة ٣.٧٢٤ [1; 5] 0.887 0.940 -0.675
EMV_03 استرخاء ٣.٧٩٩ [1; 5] 0.877 1.465 -0.675
سهولة الاستخدام، عاكسة (مور وبن باسات، 1991) EOU_01 مدة التعلم ٤.٠١١ [1; 5] 0.988 0.800 -0.996
EOU_02 عملية ٤.٠٩٢ [1; 5] 0.811 0.798 -0.822
EOU_03 تنقل القائمة 3.971 [1; 5] 0.867 1.201 -0.904
الفائدة المدركة، التأملية (أنتون وآخرون، 2013؛ مور و PU_01 الميزة العامة ٣.٣٩٧ [1; 5] 0.970 -0.176 -0.296
بن باسات، 1991 PU_02 تطبيق عملي ٣.٥٩٨ [1; 5] 1.055 -0.106 -0.585
PU_03 تحسين القراءة 3.293 [1; 5] 1.109 -0.534 -0.474
التوافق، الانعكاسي (هوانغ وهسيه، 2012؛ مور وبن باسات، 1991) CO_01 سلوك القراءة ٣.٢٩٩ [1; 5] 0.996 -0.238 -0.419
CO_02 نمط الاستهلاك ٣.٤٢٧ [1; 5] 0.991 0.259 -0.646
CO_03 احتياجات القراءة ٣.٦٥٥ [1; 5] 0.992 0.430 -0.829
نية التبني، التأملية (فينكاتيش وآخرون، 2012) إعلان_01 الاستخدام المستقبلي ٤.٠٢٣ [1; 5] 0.928 1.210 -1.046
AD_02 الاستخدام اليومي ٣.٧٧٦ [1; 5] 0.972 0.360 -0.712
AD_03 الاستخدام المتكرر ٣.٨٤٥ [1; 5] 0.925 0.869 -0.785
استخدام التكنولوجيا، عنصر واحد (فينكاتيش وآخرون، 2012) استخدم_01 الكتب الإلكترونية ٣.٩٨٣ [1; 7] 1.610 -0.894 -0.063
الجدول A2
ملخص نتائج نماذج القياس (الانعكاسية)
المتغيرات الكامنة مؤشرات أوزان الأحمال تشارك المؤشر (الأحمال المربعة) AVE موثوقية مركبة كرونباخ فترة الثقة بتقنية Bootstrap بنسبة 95% لـ HTMT لا تشمل 1
القيمة العاطفية EMV_01 0.338 0.891 0.794 0.853 0.946 0.914 0.917 نعم
EMV_02 0.375 0.950 0.903
EMV_03 0.368 0.929 0.863
سهولة الاستخدام EOU_01 0.453 0.784 0.615 0.697 0.873 0.783 0.873 نعم
EOU_02 0.371 0.878 0.771
EOU_03 0.380 0.840 0.706
الفائدة المدركة PU_01 0.319 0.722 0.521 0.642 0.842 0.723 0.753 نعم
PU_02 0.426 0.819 0.671
PU_03 0.491 0.856 0.737
التوافق CO_01 0.396 0.901 0.812 0.779 0.914 0.858 0.859 نعم
CO_02 0.357 0.906 0.821
CO_03 0.381 0.840 0.706
نية التبني إعلان_01 0.345 0.933 0.870 0.889 0.960 0.938 0.939
AD_02 0.347 0.935 0.874
AD_03 0.368 0.960 0.922
الجدول A3
جدول عنق الزجاجة بصيغ مختلفة
استخدام تكنولوجيا عنق الزجاجة، نطاقات النسب المئوية لدرجات المتغيرات الكامنة الموحدة (كما في ريشتر وآخرون، 2020)
نية التبني القيمة العاطفية سهولة الاستخدام* الفائدة المدركة التوافق
0 NN NN NN NN NN
10 NN NN 10.2 NN NN
20 NN NN 10.2 NN NN
30 NN NN 10.2 NN NN
40 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 ٢٥.٥
50 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 ٢٥.٥
60 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 ٣٣.٧
70 ٣٣.٨ ٤٩.٦ ٢٠.٥ ٤٨.١ ٣٣.٧
٨٠ ٣٣.٨ ٤٩.٦ ٢٠.٥ ٤٨.١ ٣٣.٧
90 75.0 ٤٩.٦ 60.2 66.2 ٣٣.٧
100 75.0 ٤٩.٦ 60.2 66.2 ٣٣.٧
استخدام تكنولوجيا عنق الزجاجة، درجات المتغيرات الكامنة المعاد قياسها، نطاقات النسب المئوية
0 NN NN NN NN NN
10 NN NN 10.2 NN NN
20 NN NN 10.2 NN NN
30 NN NN 10.2 NN NN
40 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 ٢٥.٥
50 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 25.5
60 ٣٣.٨ ٤٩.٦ ٢٠.٠ 15.7 ٣٣.٧
70 ٣٣.٨ ٤٩.٦ ٢٠.٥ ٤٨.١ ٣٣.٧
٨٠ ٣٣.٨ ٤٩.٦ ٢٠.٥ ٤٨.١ ٣٣.٧
90 75.0 ٤٩.٦ 60.2 66.2 ٣٣.٧
100 75.0 ٤٩.٦ 60.2 66.2 ٣٣.٧
استخدام تكنولوجيا عنق الزجاجة، درجات المتغيرات الكامنة المعاد قياسها، الدرجات الفعلية
0 NN NN NN NN NN
10 NN NN ٢٥.٤ NN NN
20 NN NN ٢٥.٤ NN NN
30 NN NN 25.4 NN NN
40 ٣٣.٨ ٤٩.٦ ٣٣.٥ 15.7 ٢٥.٥
50 ٣٣.٨ ٤٩.٦ ٣٣.٥ 15.7 ٢٥.٥
60 ٣٣.٨ ٤٩.٦ ٣٣.٥ 15.7 ٣٣.٧
70 ٣٣.٨ ٤٩.٦ ٣٣.٩ ٤٨.١ ٣٣.٧
٨٠ ٣٣.٨ ٤٩.٦ ٣٣.٩ ٤٨.١ ٣٣.٧
90 75.0 ٤٩.٦ 66.9 66.2 ٣٣.٧
100 75.0 ٤٩.٦ 66.9 66.2 ٣٣.٧
*تنتج درجات المتغيرات الكامنة الموحدة والمعاد قياسها نفس نتائج الاختناق إذا تم تقديمها كنسب مئوية، لكنها تختلف فيما يتعلق بالقيم الفعلية لبناء سهولة الاستخدام. هذه الاختلافات تعود إلى أن الحد الأدنى لقيمة سهولة الاستخدام هو 16.9. وبالتالي، تتراوح القيم المحتملة لسهولة الاستخدام من 16.9 إلى 100؛ أي 83.1. لذا، فإن تحويل النسب المئوية، على سبيل المثال 10.2 أو 20.0، إلى درجات فعلية معاد قياسها ينتج عنه أو ٣٣.٥.
-المحور: نية التبني، -محور: استخدام التكنولوجيا
الشكل A1. مخططات تشتت NCA مع خطوط السقف CE-FDH.

References

Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50, 179-211.
Alyahya, M., Agag, G., Aliedan, M., Abdelmoety, Z.H., Daher, M.M., 2023. A sustainable step forward: understanding factors affecting customers’ behaviour to purchase remanufactured products. J. Retailing Consum. Serv. 70, 103172.
American Customer Satisfaction Index, 2022. Unparalleled Customer Intelligence. Retrieved from. https://www.theacsi.org/our-industries/.
Anderson, E.W., Mittal, V., 2000. Strengthening the satisfaction-profit chain. J. Serv. Res. 3 (2), 107-186.
Antón, C., Camarero, C., Rodríguez, J., 2013. Usefulness, enjoyment, and self-image congruence: the adoption of e-book readers. Psychol. Market. 30 (4), 372-384.
Bergh, D.D., Boyd, B.K., Byron, K., Gove, S., Ketchen, D.J., 2022. What constitutes a methodological contribution? J. Manag. 48 (7), 1835-1848.
Bolívar, L.M., Roldán, J.L., Castro-Abancéns, I., Casanueva, C., 2022. Speed of international expansion: the mediating role of network resources mobilisation. Manag. Int. Rev. 62, 541-568.
Cai, X., Cebollada, J., Cortiñas, M., 2023. Impact of seller- and buyer-created content on product sales in the electronic commerce platform: the role of informativeness, readability, multimedia richness, and extreme valence. J. Retailing Consum. Serv. 70, 103141.
Cheah, J.-H., Magno, F., Cassia, F., 2023. Reviewing the smartPLS 4 software: the latest features and enhancements. Journal of Marketing Analytics. Advance online publication.
Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13 (3), 319-340.
Davis, F.D., Bagozzi, R.P., Warshaw, P.R., 1989. User acceptance of computer technology: a comparison of two theoretical models. Manag. Sci. 35 (8), 982-1003.
Dul, J., 2016. Necessary condition analysis (NCA): logic and methodology of “necessary but not sufficient” causality. Organ. Res. Methods 19 (1), 10-52.
Dul, J., 2020. Conducting Necessary Condition Analysis. Sage, London.
Dul, J., 2021a. Advances in Necessary Condition Analysis. Online book retrieved from:, Version 0.1. https://bookdown.org/ncabook/advanced_nca2/.
Dul, J., 2021b. Necessary Condition Analysis (NCA) with R. A Quick Start Guide: 2 March 2021. Online document retreived from:, Version 3.1.0. https://www.erim.eur.nl
/fileadmin/user_upload/_generated/download/Quick_Start_Guide_NCA_3.1.0_Ma rch_2_2021.pdf.
Dul, J., 2022. R Package NCA: Necessary Condition Analysis. Version:, 3.2.0. https://cr an.r-project.org/web/packages/NCA/index.html.
Dul, J., Hauff, S., Bouncken, R.B., 2023. Necessary condition analysis (NCA): review of research topics and guidelines for good practice. Rev. Manag. Sci. 17 (2), 683-714.
Dul, J., Hauff, S., Tóth, Z., 2021. Necessary condition analysis in marketing research. In: Nunkoo, R., Teeroovengadum, V., Ringle, C.M. (Eds.), Handbook of Research Methods for Marketing Management. Edward Elgar, pp. 51-72.
Fornell, C., 1992. A national customer satisfaction barometer: the Swedish experience. J. Market. 56 (1), 6-21.
Fornell, C., Johnson, M.D., Anderson, E.W., Cha, J., Bryant, B.E., 1996. The American satisfaction index: nature, purpose, and findings. J. Market. 60 (4), 7-18.
Guenther, P., Guenther, M., Ringle, C.M., Zaefarian, G., Cartwright, S., 2023. Improving PLS-SEM use for business marketing research. Ind. Market. Manag. 111, 127-142.
Hair, J.F., Hult, G.T.M., Ringle, C.M., Sarstedt, M., 2022. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), 3 ed. Sage, Thousand Oaks, CA.
Hair, J.F., Hult, G.T.M., Ringle, C.M., Sarstedt, M., Thiele, K.O., 2017. Mirror, mirror on the wall: a comparative evaluation of composite-based structural equation modeling methods. J. Acad. Market. Sci. 45 (5), 616-632.
Hair, J.F., Risher, J.J., Sarstedt, M., Ringle, C.M., 2019. When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 31 (1), 2-24.
Hair, J.F., Sarstedt, M., Ringle, C.M., Gudergan, S.P., 2024. Advanced Issues in Partial Least Squares Structural Equation Modeling (PLS-SEM), 2 ed. Sage, Thousand Oaks, CA.
Hair, J.F., Sarstedt, M., Ringle, C.M., Mena, J.A., 2012. An assessment of the use of partial least squares structural equation modeling in marketing research. J. Acad. Market. Sci. 40 (3), 414-433.
Hauff, S., Guerci, M., Dul, J., van Rhee, H., 2021. Exploring necessary conditions in HRM research: fundamental issues and methodological implications. Hum. Resour. Manag. J. 31 (1), 18-36.
Hauff, S., Richter, N.F., Tressin, T., 2015. Situational job characteristics and job satisfaction: the moderating role of culture. Int. Bus. Rev. 24 (4), 710-723.
Huang, L.-Y., Hsieh, Y.-J., 2012. Consumer electronics acceptance based on innovation attributes and switching costs: the case of e-book readers. Electron. Commer. Res. Appl. 11 (3), 218-228.
Huete-Alcocer, N., Hernandez-Rojas, R.D., 2022. Do SARS-CoV-2 safety measures affect visitors experience of traditional gastronomy, destination image and loyalty to a world heritage city? J. Retailing Consum. Serv. 69, 103095.
Kristensen, K., Martensen, A., Grønholdt, L., 2000. Customer satisfaction measurement at post Denmark: results of application of the European customer satisfaction index methodology. Total Qual. Manag. 11 (7), S1007-S1015.
Le, H.T.P.M., Kim, D., Park, J., 2024. The way to generate customer citizenship behavior with customer experience. J. Retailing Consum. Serv. 76, 103608.
Lohmöller, J.-B., 1989. Latent Variable Path Modeling with Partial Least Squares. Physica, Heidelberg.
Martilla, J.A., James, J.C., 1977. Importance-performance analysis. J. Market. 41 (1), 77-79.
Moore, G.C., Benbasat, I., 1991. Development of an instrument to measure the perceptions of adopting an information technology innovation. Inf. Syst. Res. 2 (3), 192-222.
Ng, F.Z.-X., Yap, H.-Y., Tan, G.W.-H., Lo, P.-S., Ooi, K.-B., 2022. Fashion shopping on the go: a dual-stage predictive-analytics SEM-ANN analysis on usage behaviour, experience response and cross-category usage. J. Retailing Consum. Serv. 65, 102851.
Oh, H., 2001. Revisiting importance-performance analysis. Tourism Manag. 22 (6), 617-627.
Pappas, N., 2023. Came and gone? A longitudinal study of the effects of COVID-19 on tourism purchasing intentions. J. Retailing Consum. Serv. 72, 103269.
Perez-Aranda, J., Gonzáles Robles, E.M., Alarcón Urbistondo, P., 2023. Understanding antecedents of continuance and revisit intentions: the case of sport apps. J. Retailing Consum. Serv. 72, 103288.
Richter, N.F., Hauff, S., 2022. Necessary conditions in international business research: advancing the field with a new perspective on causality and data analysis. J. World Bus. 57 (5), 101310.
Richter, N.F., Hauff, S., Gudergan, S.P., Ringle, C.M., 2022. The use of partial least squares structural equation modeling and complementary methods in international management research. Manag. Int. Rev. 62 (4), 449-470.
Richter, N.F., Hauff, S., Kolev, A.E., Schubring, S., 2023a. Dataset on an extended technology acceptance model: a combined application of PLS-SEM and NCA. Data Brief 48, 109190.
Richter, N.F., Hauff, S., Ringle, C.M., Sarstedt, M., Kolev, A.E., Schubring, S., 2023b. How to apply necessary condition analysis in PLS-SEM. In: Latan, H., Hair, J.F., Noonan, R. (Eds.), Partial Least Squares Path Modeling: Basic Concepts, Methodological Issues and Applications. Springer, pp. 267-297.
Richter, N.F., Martin, J., Hansen, S.V., Taras, V., Alon, I., 2021. Motivational configurations of cultural intelligence, social integration, and performance in global virtual teams. J. Bus. Res. 129, 351-367.
Richter, N.F., Schubring, S., Hauff, S., Ringle, C.M., Sarstedt, M., 2020. When predictors of outcomes are necessary: guidelines for the combined use of PLS-SEM and NCA. Ind. Manag. Data Syst. 120 (12), 2243-2267.
Rigdon, E.E., Ringle, C.M., Sarstedt, M., 2010. Structural modeling of heterogeneous data with partial least squares. In: Malhotra, N.K. (Ed.), Review of Marketing Research, vol. 7. Sharpe, Armonk, NY, pp. 255-296.
Ringle, C.M., Sarstedt, M., 2016. Gain more insight from your PLS-SEM results: the importance-performance map analysis. Ind. Manag. Data Syst. 116 (9), 1865-1886.
Ringle, C.M., Wende, S., Becker, J.-M., 2022. SmartPLS 4. Oststeinbek: SmartPLS. Retrieved from. https://www.smartpls.com/.
Rodríguez, P.G., Villarreal, R., Valiño, P.C., Blozis, S., 2020. A PLS-SEM approach to understanding e-SQ, e-satisfaction and e-loyalty for fashion e-retailers in Spain. J. Retailing Consum. Serv. 57, 102201.
Sarstedt, M., Hair, J.F., Pick, M., Liengaard, B.D., Radomir, L., Ringle, C.M., 2022. Progress in partial least squares structural equation modeling use in marketing research in the last decade. Psychol. Market. 39 (5), 1035-1064.
Sarstedt, M., Hair, J.F., Ringle, C.M., Thiele, K.O., Gudergan, S.P., 2016. Estimation issues with PLS and CBSEM: where the bias lies. J. Bus. Res. 69 (10), 3998-4010.
Sever, I., 2015. Importance-performance analysis: a valid management tool? Tourism Manag. 48, 43-53.
Shahidi, N., Tossan, V., Bourliataux-Lajoinie, S., Cacho-Elizondo, S., 2022. Behavioral intention to use a contact tracing application: the case of StopCovid in France. J. Retailing Consum. Serv. 68, 102998.
Sheppard, B.H., Hartwick, J., Warshaw, P.R., 1988. The theory of reasoned action: a meta-analysis of past research with recommendations for modifications and future research. J. Consum. Res. 15 (3), 325-343.
Sheth, J.N., Newman, B.I., Gross, B.L., 1991. Why we buy what we buy: theory of consumption values. J. Bus. Res. 22 (2), 159-170.
Siyal, A.W., Chen, H., Jamal Shah, S., Shahzad, F., Bano, S., 2024. Customization at a glance: investigating consumer experiences in mobile commerce applications. J. Retailing Consum. Serv. 76, 103602.
Skok, W., Kophamel, A., Richardson, I., 2001. Diagnosing information systems success: importance-performance maps in the health club industry. Inf. Manag. 38 (7), 409-419.
Slack, N., 1994. The importance-performance matrix as a determinant of improvement Priority. Int. J. Oper. Prod. Manag. 44 (5), 59-75.
Streukens, S., Leroi-Erelds, S., Willems, K., 2017. Dealing with nonlinearity in importance-performance map analysis (IPMA): an integrative framework in a PLSSEM context. In: Latan, H., Noonan, R. (Eds.), Partial Least Squares Path Modeling. Springer, Cham, pp. 367-403.
Sukhov, A., Olsson, L.E., Friman, M., 2022. Necessary and sufficient conditions for attractive public transport: combined use of PLS-SEM and NCA. Transport. Res. Pol. Pract. 158, 239-250.
Sweeney, J.C., Soutar, G.N., 2001. Consumer perceived value: the development of a multiple item scale. J. Retailing 77 (2), 203-220.
Tiwari, P., Kaurav, R.P.S., Koay, K.Y., 2023. Understanding travel apps usage intention: findings from PLS and NCA. Journal of Marketing Analytics. advance online publication.
Turner, M., Kitchenham, B., Brereton, P., Charters, S., Budgen, D., 2010. Does the technology acceptance model predict actual use? A systematic literature review. Inf. Software Technol. 52 (5), 463-479.
Venkatesh, V., Thong, J.Y.L., Xu, X., 2012. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Quarterly 36 (1), 157-178.
Wold, H., 1982. Soft modeling: the basic design and some extensions. In: Jöreskog, K.G., Wold, H. (Eds.), Systems under Indirect Observations: Part II. North-Holland, Amsterdam, pp. 1-54.
Zhang, P., Chao, C.-W., Chiong, R., Hasan, N., Aljaroodi, H.M., Tian, F., 2023. Effects of in-store live stream on consumers’ offline purchase intention. J. Retailing Consum. Serv. 72, 103262.
Bokrantz, J., Dul, J., 2023. Building and testing necessity theories in supply chain management. J. Supply Chain Manag. 59 (1), 48-65.
Sarstedt, M., Mooi, E.A., 2019. A Concise Guide to Market Research: The Process, Data, and Methods Using IBM SPSS Statistics. 3, ed. Springer, Berlin, Heidelberg.
Sharma, P.N., Liengaard, B.D., Hair, J.F., Sarstedt, M., Ringle, C.M., 2023. Predictive model assessment and selection in composite-based modeling using PLS-SEM: extensions and guidelines for using CVPAT. Eur. J. Market. 57 (6), 1662-1677.
Shmueli, G., Sarstedt, M., Hair, J.F., Cheah, J., Ting, H., Vaithilingam, S., Ringle, C.M., 2019. Predictive model assessment in PLS-SEM: guidelines for using PLSpredict. Eur. J. Market. 53 (11), 2322-2347.

    • Corresponding author. Ludwig-Maximilians-University Munich, LMU Munich School of Management, Ludwigstraße 28 RG/III, 80539, Munich, Germany.
    E-mail addresses: hauff@hsu-hh.de (S. Hauff), nicole@sam.sdu.dk (N.F. Richter), sarstedt@lmu.de (M. Sarstedt), c.ringle@tuhh.de (C.M. Ringle).
  1. We refer to a trichotomous illustration here to facilitate interpretation; however, the NCA is not limited to specific measurement levels. For more information, see Dul (2020, Chapter 3). Also note that the symbols used in the NCA context have a different meaning than in the PLS-SEM context. While the latter usually uses to denote latent constructs and to denote indicators, the former uses Y for outcomes and X for conditions.
  2. Note that the bottleneck levels differ due to the different scales involved. Also note that this statement refers to the standardized, unstandardized or rescaled scores calculated in the IPMA context on the basis of a linear transformation. Other forms of transformation, particularly non-linear transformations, should be carefully considered, since they could affect the NCA’s results.
  3. Under https://www.pls-sem.net/downloads, we provide a Microsoft Excel template for downloading in order to produce this kind of indication.
  4. In the Appendix, we offer a comparison of bottleneck tables for the standardized latent variable scores based on percentage ranges, the rescaled latent variable scores based on percentage ranges, and the bottleneck table for rescaled latent variable scores based on the actual values. We also comment on the differences (see Table A3 in the Appendix).

Journal: Journal of Retailing and Consumer Services, Volume: 78
DOI: https://doi.org/10.1016/j.jretconser.2024.103723
Publication Date: 2024-01-23

SDÚ University of Southern Denmark

Importance and performance in PLS-SEM and NCA

introducing the Combined Importance-Performance Map Analysis (cIPMA)

Hauff, Sven; Richter, Nicole Franziska; Sarstedt, Marko; Ringle, Christian M.

Published in:

Journal of Retailing and Consumer Services

DOI:

10.1016/j.jretconser.2024.103723

Publication date:

2024

Document version:

Final published version

Document license:

CC BY-NC-ND

Citation for pulished version (APA):

Hauff, S., Richter, N. F., Sarstedt, M., & Ringle, C. M. (2024). Importance and performance in PLS-SEM and NCA: introducing the Combined Importance-Performance Map Analysis (cIPMA). Journal of Retailing and Consumer Services, 78(May), Article 103723. https://doi.org/10.1016/j.jretconser.2024.103723
Go to publication entry in University of Southern Denmark’s Research Portal

Terms of use

This work is brought to you by the University of Southern Denmark.
Unless otherwise specified it has been shared according to the terms for self-archiving.
If no other license is stated, these terms apply:
  • You may download this work for personal use only.
  • You may not further distribute the material or use it for any profit-making activity or commercial gain
  • You may freely distribute the URL identifying this open access version
If you believe that this document breaches copyright please contact us providing details and we will investigate your claim. Please direct all enquiries to puresupport@bib.sdu.dk

Importance and performance in PLS-SEM and NCA: Introducing the combined importance-performance map analysis (cIPMA)

Sven Hauff , Nicole Franziska Richter , Marko Sarstedt , Christian M. Ringle Helmut Schmidt University, Holstenhofweg 85, 22043, Hamburg, Germany University of Southern Denmark, Campusvej 55, DK-5230, Odense M, Denmark Ludwig-Maximilians-University Munich, LMU Munich School of Management, Ludwigstraße 28 RG/III, 80539, Munich, Germany Babeş-Bolyai University, Faculty of Economics and Business Administration, Romania Hamburg University of Technology, Department of Management Sciences and Technology, Am Schwarzenberg-Campus 4 (D), 21073, Hamburg, Germany James Cook University, Australia

A R T I C L E I N F O

Handling Editor: Prof. H. Timmermans

Keywords:

cIPMA
Necessary condition analysis
NCA
Partial least squares
PLS
PLS-SEM
Structural equation modeling
SEM
Importance-performance map analysis
IPMA

Abstract

This research offers a novel approach that extends the application of importance-performance map analysis (IPMA) in partial least squares structural equation modeling (PLS-SEM) by incorporating findings from a necessary condition analysis (NCA). The IPMA comprises assessing latent variables and their indicators’ importance and performance, while an NCA introduces an additional dimension by identifying factors that are crucial for achieving the desired outcomes. An NCA employs necessity logic to identify the must-have factors required for an outcome, while PLS-SEM follows an additive sufficiency logic to identify the should-have factors that contribute to high performance levels. Integrating these two logics into the performance dimension is particularly valuable for prioritizing actions that could improve the target outcomes, such as customer satisfaction and employee commitment. Although the combined use of PLS-SEM and NCA is a recent suggestion, this study is the first to combine them with an IPMA (i.e., in a combined IPMA; cIPMA). A case study illustrates the combined use of PLS-SEM and an NCA to undertake a cIPMA. This innovative approach enhances researchers’ and practitioners’ decision making, enabling them to prioritize their efforts effectively.

1. Introduction

Comparing attributes’ performance and importance in order to produce a given outcome has a long tradition in the management discipline (Martilla and James, 1977). Researchers routinely visualize this interplay in a two-dimensional plot, in which the attributes are normally grouped into four quadrants that combine low and high importance and performance values. This plot is usually referred to as an importance-performance grid (Martilla and James, 1977), importance-performance matrix (Slack, 1994), a quality map (Kristensen et al., 2000), or, more broadly, an importance-performance map analysis (IPMA, Ringle and Sarstedt, 2016). The IPMA has become a standard tool for understanding where managerial improvement efforts should be focused (e.g., Sever, 2015; Skok et al., 2001). Nevertheless, authors from different disciplines and methodological backgrounds have encountered challenges when using the toolset, which – among others relate to understanding importance and performance, as well as the
threshold levels’ definition (see, for instance, Oh, 2001).
The IPMA has also been used in the context of partial least squares structural equation modeling (PLS-SEM), a multivariate method for estimating complex interrelationships between constructs and their indicators (Hair et al., 2022, Chapter 1; Lohmöller, 1989, Chapter 2; Wold, 1982). In the past decades, the use of PLS-SEM evolved in marketing research (Guenther et al., 2023; Hair et al., 2012; Sarstedt et al., 2022), including retailing and consumer services (e.g., Cai et al., 2023; Hue-te-Alcocer and Hernandez-Rojas, 2022; Rodríguez et al., 2020). PLS-SEM facilitates an IPMA, due to the way the method estimates model parameters. Specifically, PLS-SEM calculates indicator variables’ composites to represent the constructs in a statistical model. Researchers have used these composite scores as representations of the constructs’ performance and compared them with the total effects that the constructs exert on a specific target (e.g., Kristensen et al., 2000; Ringle and Sarstedt, 2016; Streukens et al., 2017). The underlying IPMA draws on the average performance and importance scores and subsequently
visualizes them in an importance-performance map.
Numerous studies have used the IPMA in a PLS-SEM context. For example, in the context of retailing and consumer services, Zhang et al. (2023) draw on the IPMA to investigate environmental stimuli’s importance and performance for consumers’ in-store purchase intentions via two consumer attitude constructs. In a technology acceptance model (TAM), whose use also features prominently in retailing and consumer services (e.g., Ng et al., 2022; Perez-Aranda et al., 2023; Shahidi et al., 2022), the IPMA can be used to compare the total effect of a technology’s ease of use in terms of its intended usage (its importance for the target construct) with its potential users’ average perceived ease of use (its performance). An IPMA’s results therefore allow the identification of a target construct’s highly important antecedents, that, however, exhibit relatively low performance. This information is crucial, since it indicates how action should be prioritized to improve these antecedent constructs, which will increase the target construct substantially. For instance, ease of use’s high importance, combined with a technology’s poor ease of use rating (i.e., a poor performance), indicates a specific need to improve this technology’s ease of use to ultimately increase its usage.
Inherently, the IPMA’s implementation follows an additive sufficiency logic according to which multiple antecedents could contribute to an outcome. Antecedent constructs that improve an outcome, namely those that show a low performance, but have a strong and significant total effect on an outcome, are prioritized, while those showing weak effects are not prioritized. However, it may well be that a certain antecedent construct with a relatively low importance (i.e., a weak effect regarding increasing the desired outcome) might still require particular attention when its absence prevents the outcome. For instance, an IPMA in the context of the TAM might ascertain that a technology’s compatibility has relatively low importance for its use. Yet, if the compatibility doesn’t meet a specific threshold, users would be unwilling to use the technology. Considering such interrelationships requires adopting a necessity perspective, which implies that an outcome – or a certain level of an outcome – could only be achieved if the necessary cause is in place or at a certain level.
To incorporate such a necessity logic, researchers can draw on the necessary condition analysis (NCA; Dul, 2016; 2020), which has recently gained prominence in many business research fields (for a recent review of the topics analyzed with an NCA, see Dul et al., 2023), either as a stand-alone method or in combination with PLS-SEM and other regression-based methods (Richter et al., 2022). The NCA seeks to identify necessary conditions representing constraints, bottlenecks or critical factors, which need to be solved or satisfied to achieve a certain outcome. Identifying such necessary conditions is highly relevant for management practice, since an outcome can only be achieved if the necessary condition is in place or is at a certain level (Dul et al., 2021; Hauff et al., 2021; Richter and Hauff, 2022). Importantly, antecedent constructs showing weak and nonsignificant PLS-SEM effects might be necessary conditions; that is, without these antecedents a certain outcome cannot be achieved. Failure to consider these antecedent constructs could therefore lead to incomplete recommendations. For example, in the context of retailing and consumer services, Pappas (2023) highlights the impact of recession and quality risks as constant necessary conditions for holidaymakers’ purchasing intentions. Alyahya et al. (2023) use the NCA to reveal that moral obligation, moral accountability, perceived risk, perceived risk, and cost knowledge are necessary antecedents for consumers’ purchase of remanufactured products.
The recently proposed combined use of NCA and PLS-SEM (Richter et al., 2020) has been acknowledged as “a unique contribution by comparing and combining approaches, demonstrating how NCA (which focuses on necessary conditions) can [sic] be used in combination with PLS-SEM (which focuses on sufficiency) to create a previously unrecognized way of assessing causality” (Bergh et al., 2022, p. 1842). In line with recent research work that adopted this multimethod approach (e.
g., Bolívar et al., 2022; Richter et al., 2021; Sukhov et al., 2022; Tiwari et al., 2023), we argue that identifying necessary conditions could also enrich an IPMA by providing information about the antecedent constructs’ necessity. This information provides researchers and practitioners with a more complete picture of the important and necessary antecedents, while also helping to complement an IPMA’s results. By following the path model results of a technology’s perceived usefulness, researchers assessing consumer acceptance of novel technologies might, for instance, find that these are of little importance, although they are a necessary condition for the technology’s usage. In other words, the technology will not be used unless it is perceived as useful. While an antecedent construct with little importance according to its PLS-SEM results might receive little attention from a classic IPMA, it might represent a necessary condition, which would once again make it the focus of decision making when PLS-SEM is combined with an NCA.
Against this background, we first contribute to the field by discussing how importance and performance are understood in the context of PLSSEM and an NCA, as well as how the two approaches complement one another in importance-performance analyses. Building on prior research in each field (e.g., Dul, 2016, 2020; Ringle and Sarstedt, 2016), we discuss the concepts of importance and performance in a PLS-SEM context – as implemented in an IPMA – and an NCA context. In addition, we present guidelines for a combined IPMA that builds on PLS-SEM and NCA results. By doing so, we extend the guidelines proposed in Richter et al. (2020), who did not address the IPMA (see also Richter et al., 2023b). Finally, we illustrate a combined IPMA by using an extended TAM and offering researchers implementing the approach a toolset, before concluding our research and outlining future research.

2. Importance and performance in PLS-SEM

PLS (Lohmöller, 1989, Chapter 2; Wold, 1982) is a composite-based approach to SEM in that it represents constructs as weighted sums of indicators (Hair et al., 2017; Sarstedt et al., 2016). Building on this characteristic, researchers have used composite scores produced by PLS-SEM to develop indices such as the Swedish Customer Satisfaction Barometer (Fornell, 1992) and the American Customer Satisfaction Index (Fornell et al., 1996). In the computation of these indices, the composite scores are conceived as performance values, reflecting the respondents’ satisfaction with certain features or with the overall construct. For example, if all respondents show a maximum level of overall satisfaction, this translates into a 100 percent performance of this construct. Starting with Anderson and Mittal (2000) and Kristensen et al. (2000), follow-up research has contrasted these scores with measurement or structural model weights representing the importance of individual indicators or constructs for improving a certain target, such as firm profit. Jointly, these two dimensions define an importance-performance map’s axes, which represent the average importance and performance scores with regard to a certain target construct.
With models’ increasing complexity, which often span multiple layers of constructs, researchers usually draw on structural model total effects – that is, the sum of an antecedent construct’s direct and all of its indirect effects on an outcome – to represent an IPMA’s importance dimension (Hair et al., 2024, Chapter 4; Ringle and Sarstedt, 2016). That is, rather than restricting the analysis to a target’s direct antecedents, the IPMA also considers indirectly related constructs (i.e., those that impact the outcome via one or more mediating constructs).
In the importance-performance map, the antecedent constructs’ importance values are plotted on the -axis and their performance values on the -axis. Using this illustration, researchers could, for example, identify antecedent constructs that are relatively important regarding explaining the key target constructs of interest (i.e., those with a strong total effect) with a relatively low performance (i.e., low average latent variable scores). Such constructs would, specifically, be highly prioritized in order to achieve improvement, because they are especially
important for the target construct, although they simultaneously perform relatively poorly.
To better illustrate the IPMA concept, consider the hypothetical path model in Fig. 1 (Panel A) with the four constructs to . In this path model, is the target construct. Based on the path coefficients, the antecedent constructs , and have direct effects on . In addition, and have indirect effects on via , which are due to the corresponding direct effects’ product (for further details, see Hair et al., 2024, Chapter 4). In this example, the indirect effects of on are computed as follows:



Hence, the total indirect effect of on is . Adding the direct effect of on (i.e., 0.50), as well as the antecedent constructs’ total indirect effect (i.e., 0.34), we not only obtain the total effect of , but also the importance of for the key target construct .
In contrast, the construct scores’ average values represent their performance. Here, researchers could use the standardized construct scores or the unstandardized construct scores. The standardized construct scores are estimated on the basis of -transformed standardized indicator data. The resulting standardized construct scores always have a mean value of zero and a standard deviation of 1 . Since reporting mean values of zero for all the constructs is less useful for the IPMA, the procedure refers to the unstandardized construct scores.
The unstandardized construct scores depend on the indicators’ scales
Fig. 1. Importance-performance map analysis example; adopted from Hair et al. (2024, Chapter 4).
(e.g., 1 to 5 or 1 to 7 ). If all the indicators are measured on the same scale (e.g., all on a scale from 1 to 5), their interpretation on the construct level is straightforward. However, their interpretation becomes ambiguous if different scales are involved (e.g., Ringle and Sarstedt, 2016; Hair et al., 2024, Chapter 4). For a better comparison of the performance levels, the unstandardized indicator data can be rescaled so that they all range between 0 and 100, with 0 representing the lowest and 100 the highest performance. The rescaling of an observation with respect to indicator proceeds via

where is the -th indicator in the PLS path model, E[.] represents indicator i’s actual score of respondent , and the min[.] and the max[.] represent the indicator’s minimum and maximum values (Ringle and Sarstedt, 2016; Hair et al., 2024, Chapter 4). The indicator’s minimum and maximum can refer to the theoretical minimum and maximum values (e.g., 1 and 7 when using a seven-point scale) or the empirical minimum and maximum values resulting from the sample data (e.g., 2 and 7, because no respondent evaluated the indicators below 2). It is easy to determine the theoretical minimum and maximum of an interval scale (e.g., 1 and 7 when using a seven-point scale). The use of other scales might not have a theoretical minimum or maximum, and the researcher could be advised or even forced to use the data at hand’s empirical minimum or maximum.
The rescaled latent variable scores are a linear combination of the rescaled indicator data and the rescaled outer weights. To obtain the rescaled weights, we must first compute the unstandardized weights by dividing the standardized weights by the standard deviation of its respective indicator. These weights are normalized per measurement model so that the sum of these final rescaled outer weights equals one. These rescaled outer weights allow us to determine the unstandardized rescaled construct scores that range between 0 and 100 (i.e., by the linear combination of the indicator data on a scale from 0 to 100 and the rescaled outer weights per construct). The average of these values represents the construct’s performance score referred to in the IPMA in PLSSEM (see values in brackets in the constructs in Fig. 1, Panel A).
Fig. 1 (Panel B) shows how the IPMA combines these two aspects graphically for the target construct . Two grey dashed lines divide the importance-performance map into four quadrants. The vertical line represents the mean importance and the horizontal line corresponds to the constructs’ mean performance, computed as follows (Fig. 1, Panel A):
mean importance , and
mean performance .
Researchers using the IPMA could use these average scores and plot them on an importance-performance map into four separate quadrants. However, researchers might rely on a different logic when dividing the four quadrants, such as previous knowledge or expert assessment. The combination of high/low importance and high/low performance induces specific recommendations, which antecedent constructs target through managerial activities (Ringle and Sarstedt, 2016).

3. Importance and performance in NCA

Unlike the classic IPMA, the NCA builds on a necessity logic to assess the antecedent constructs’ importance and performance for an outcome. A necessity logic implies that an outcome can only be achieved if a specific condition is present (Dul, 2016, 2020, Chapter 2). A necessary condition therefore represents a constraint or a bottleneck that must be overcome to allow the outcome to exist. In order to detect necessary conditions in data sets, NCA uses ceiling line techniques to separate
areas in scatter plots without observations from those with observations. By dividing the ’empty’ space (also called the ceiling zone) by the entire area that includes observations (also called the scope), NCA calculates the necessity effect size , which could be 0 if there is no empty space, and 1 if the maximum possible space is empty.
Consider, for example, a trichotomous case where the condition and the outcome each have three levels, namely low, medium, and high (Fig. 2). The effect size is calculated as , with being the ceiling zone and the scope. The size of is calculated by counting the number of ’empty’ cells. could be calculated as follows: ; where is the number of levels, and is the number of levels. In our example, . In Scenario A in Fig. 2, the effect size is ; in Scenario B in Fig. 2, the effect size is 0.75 (for details on calculating the effect sizes see Dul, 2020, Chapter 4). The effect size therefore depends on how many of the total cells are empty. If only one cell is empty, the effect size is 0.25 (Fig. 2, Scenario A). If three cells are empty, the effect size is 0.75 (Fig. 2, Scenario B).
The effect size therefore specifies the extent to which is constrained by . As an arbitrary benchmark, could be considered a ‘small effect,’ a ‘medium effect,’ a ‘large effect,’ and d a ‘very large effect’ (Dul, 2016). However, in order to conclude that a necessary condition hypothesis could be accepted, should at least have a medium effect size (e.g., ) and a statistical significance (e.g., with a value lower than 0.05 ). Moreover, a necessary condition should always be theoretically supported (Dul, 2021a, Chapter 2; Dul et al., 2023).
In order to decide whether an identified effect size is satisfactory, researchers should always include contextual knowledge (Dul, 2020, Chapter 4). In fact, evaluating importance for an NCA could be perceived as less straightforward than for a PLS-SEM application. For example, to conclude that an effect size of 0.75 is more important than one of 0.25 might be misleading if a high level is always desired or needed from a practical perspective. In our trichotomous example, both effect sizes could be considered equally important, since both indicate that a specific level is needed to achieve a high (i.e., a medium level of in Scenario A and a high level of in Scenario B). Conversely, an at least medium level and statistically significant effect size might not be important at all if it does not constrain the outcome’s desired level. For example, in Scenario A, the outcome of is not constrained by if one only endeavors to achieve a medium level. In this case, the effect size of 0.25 is therefore of no practical importance. In contrast, in Scenario B, represents a necessary condition to achieve a medium outcome level.
In the NCA context, performance should be understood in terms of how much of a specific condition’s required level has already been achieved. This can be analyzed from a case or sample perspective. From a case-level perspective, we could determine whether one case performs better than another case in terms of achieving the necessary level. For example, in Fig. 2, Scenario B, a case with a level of low performs worse than a case with a level of medium (although this is still not enough to achieve a high level). This case perspective could, for example, be particularly interesting if an organization wants to know how it compares to other organizations. From a sample perspective, we could analyze how many cases in a sample still need to achieve the condition’s required level. This sample perspective could, for example, be applied if an organization wants to obtain insights into its customers or employees. PLS-SEM analyses and the interpretation of a combined importance performance map analysis’s (cIPMA’s) findings focus on samples (or subsamples, such as specific groups of customers in a
dataset), which makes the latter perspective particularly relevant when combining PLS-SEM-based IPMA results with NCA insights.
In the case of multiple necessary conditions, a comparison of the different performance levels should help identify the conditions that require particular attention. For example, if almost all the cases have already achieved the required level, this condition might not be as imperative as one where many cases have not done so (Dul, 2021a, Chapter 4). Consequently, comparing different necessary conditions’ performance could help prioritize actions.
In sum, in order to assess the importance of different necessary conditions in the context of a cIPMA, we suggest that researchers should first evaluate whether a necessary condition is theoretically supported and whether it at least has a medium effect size and is statistically significant. Nevertheless, we do not recommend merely focusing on the effect size’s magnitude in order to understand its practical importance in a specific research project’s context. Instead, we propose defining a desired outcome (or target construct) level and assessing whether this level could only be achieved with a specific level of the condition. This could be done by using the information inherent in the ceiling line, which could also be illustrated in tabular form in a bottleneck table. The first column of such a table shows the outcome’s different levels, while the next one displays the condition’s corresponding critical levels; that is, the levels that need to be satisfied to achieve the outcome (as we will illustrate in the following sections). Consequently, the ultimate assessment of a cIPMA’s importance depends on the outcome or target construct’s desired level (i.e., the level), which the researcher specifies. In order to analyze performance, we suggest that CIPMA refers to the number of cases below the necessary levels, which will prioritize the necessary conditions’ identification.

4. Guidelines for a combined importance-performance map analysis

Richter et al. (2020) provided guidelines to combine PLS-SEM with NCA, which we will extend to accommodate the cIPMA (see also Richter et al., 2023b). Fig. 3 shows our extended guidelines, whose additional steps (printed in bold) we will discuss in detail. These additional steps relate to the further requirements checks in Step 4, to running the cIPMA and transferring the latent variable scores in Step 5, to the specific settings to run the NCA in Step 6, and to an enriched interpretation of the findings in Step 8. Please note that we also update some of the steps outlined in Richter et al. (2020) (marked with * in Fig. 3).

4.1. Measurement scales of indicators (step 2)

The IPMA usually relies on the unstandardized latent variable scores, which are being rescaled on a scale from 0 to 100 to aid the interpretation of the constructs’ performances. Using the rescaled latent variable scores in the NCA will provide the same necessity effect sizes as when applying the method to the (standardized or unstandardized) latent variable scores from a PLS-SEM analysis, as long as the indicators used to measure a construct are measured on the same scale. For this reason, we recommend applying the cIPMA to models where each construct relies on indicators that are measured on the same (theoretical) scale. That is, while one constructs’ indicators may draw on a 7-point Likert scale only, another construct’s indicators may rely on a 5-point rating scale only. Researchers must not use differently scaled indicators in the measurement model of a single construct when running the CIPMA.
Fig. 2. Trichotomous necessary conditions.

4.2. Additional requirements check when evaluating measurement models (step 4)

In the context of PLS-SEM, IPMA requires the outer weights estimates to be positive, regardless of whether the measurement model is formatively or reflectively specified. If the outer weights are negative, the (rescaled) latent variable scores will not fall within the range (but would, e.g., be between -5 and 95) (Ringle and Sarstedt, 2016). If an indicator’s outer weight is unexpectedly negative and significant, researchers are advised to examine the indicator and its scale. The scale or question should not have a different direction (e.g., reverse-scale items) compared to the measurement model’s other indicators. If the negative outer weights are nonsignificant, researchers should consider removing them. Finally, negative outer weights could be a result of high indicator collinearity, which the variance inflation factor values of 5 and higher indicate (Hair et al., 2019). In this case, researchers should again consider removing the indicators. However, they should note that removing indicators from measurement models means taking more considerations into account, which Hair et al. (2022, Chapter 5) explain in more detail.

4.3. Run the IPMA and transfer the latent variable scores (step 5)

Researchers should rely on the theoretical minimum and maximum scores of the indicator scales when running the IPMA; for instance, 1 and 5 on a scale that ranges from 1 to 5 . This way, the latent variable scores can be interpreted in the form of percentage values from 0 to 100 . If a respondent, for instance, answered a 3 on a scale that ranges from 1 to 5, the rescaling indicates a mediocre performance of the trait that the indicator expressed. After running the IPMA, the rescaled latent variable scores need to be exported into a new data file that serves as input for the NCA. If the aim is to later extend the IPMA to the indicator level, we recommend that researchers also export the rescaled indicator data.

4.4. Run the NCA (step 6)

In step 6, researchers should run the NCA in order to identify potential necessary conditions. The NCA can be conducted with a free R package (Dul, 2022). Alternatively, the SmartPLS 4 software (Ringle et al., 2022) also supports the NCA (see also Cheah et al., 2023; Richter et al., 2023b; Hair et al., 2024, Chapter 4; for basic guidelines on how to conduct and report NCA results, see also Dul et al., 2023). The NCA draws on the endogenous construct’s (constructs’) and their direct antecedent constructs’ latent variable scores, as well as on the potential
indicator scores if there are formative measurement models (as explained by Richter et al., 2020; Richter et al., 2023b; Hair et al., 2024, Chapter 4). The NCA could refer to the input data’s empirical or theoretical scope. In the NCA context, the general recommendation is to use the empirical scope, as it produces a more conservative estimation of the effect size (Dul, 2020, Chapter 4) – we concur with this recommendation. That is, we use the rescaled latent variable scores from the IPMA in PLS-SEM as input data and refer to their empirical scope (which may not necessarily be from 0 to 100).
Creating bottleneck tables helps researchers assess their necessary conditions’ importance and performance. A bottleneck table is a tabular representation of the ceiling line. The first column of this table displays different levels of the outcome, while the next column or columns display the condition or conditions’ corresponding levels. The levels of the outcome and the levels of the condition(s) can be displayed in different ways (Dul, 2021a, Chapter 4.3). The NCA package in R produces (as a default) bottleneck tables in which, for example, both the outcome and the condition(s) are displayed as percentage ranges, with a level of 0% corresponding to the lowest observed values and a level of to the highest observed values (that is, it refers to the values’ empirical range). An alternative is to display the outcome and the condition(s) in terms of their actual values. Actual values are those used as input for the analysis; that is, if researchers use rescaled latent variable scores, these are the actual values. Note that the use of percentage ranges and actual values will provide the same results when researchers use rescaled latent variable scores with an actual minimum of 0 and a maximum of 100 .
By referring to these bottleneck levels, researchers can select a specific outcome level and determine which level of a condition is necessary for its achievement. The bottleneck table can also be used to determine the antecedent constructs’ performance from a necessity logic perspective. Therefore, the bottleneck table needs to be displayed as percentiles of the antecedent constructs, which provide the percentage and number of cases that do not reach the required level for a corresponding outcome level.
The NCA usually produces bottleneck tables with ten equidistant steps when, for instance, using percentage ranges from 0 to , 20 etc.). In order to offer a combined interpretation of the findings, we recommend specifying a performance level of interest for the dependent or target constructs in respect of which the antecedent constructs’ levels will be evaluated in more detail. Researchers could determine the target construct’s or constructs’ desired performance level by building on expert knowledge, typical standards in the research context, or on ambitious aims forming the project’s background. For instance, in job
Step 1: Specify the research objective and theoretical background
Outline hypotheses along the sufficiency and the necessity logic (for the latter, see Bokrantz & Dul, 2023; Richter & Hauff, 2022).*
Step 2: Prepare and check the data
Sample size: Follow the guidelines on sample size as outlined in a PLS-SEM context by, for instance, referring to published power tables (Hair et al., 2022, Chapter 1).
Data distribution: Report information on the data distribution (Hair et al., 2012).
Outliers: Perform an outlier analysis by following common guidelines, such as examining observations that show a zscore > 3 (Sarstedt & Mooi, 2019, Chapter 5).*
Measurement level/coding of scales: Use metric or quasi-metric data (i.e., interval-scaled, such as Likert scales); ensure that the direction of the scale/coding corresponds to the theoretically assumed relationships, otherwise revert to or flip the scale. Ensure that the indicators used to measure a construct draw on the same scale (e.g., all indicators on a scale from 1 to 5).
Step 3: Run the PLS-SEM analysis
Use PLS-SEM to estimate the latent variable scores, the structural model relationships, and their significance (Hair et al., 2022, Chapter 6).
Step 4: Evaluate the reliability and validity of the measurement models
Make use of the assessment guidelines in the PLS-SEM context to evaluate the measurement models’ quality (e.g., Guenther et al., 2023; Hair et al., 2022, Chapters 4 and 5; Sarstedt et al. 2021). In respect of reflective constructs, evaluate their loadings, Cronbach’s / composite reliability / , the average variance extracted, and the heterotraitmonotrait ratio. In respect of formative constructs, perform a redundancy analysis, evaluate the variance inflation factors, and the indicator weights’ significance and relevance. If required, make improvements.
Additional requirements check: The outer weights’ estimates need to be positive in all the measurements models (i.e., the formative and the reflective). In the case of nonsignificant negative outer weights, and / or a VIF of 5 and higher in the measurement models, researchers could, given the considerations in Hair et al. (2022, Chapter 5 ), consider removing the indicator.
Step 5: Run the IPMA and transfer the latent variable scores
Run the IPMA in PLS-SEM using the theoretical minimum and maximum values of the indicator scales as input (Ringle & Sarstedt, 2016).
Export the rescaled latent variable scores (and, if desired, the rescaled indicator data) to a new file and import the new data file into R or SmartPLS.
Step 6: Run the NCA
Run the NCA in SmartPLS (see also Richter et al., 2023b) or R (default settings; i.e., use the empirical scope of the input data; 10,000 permutations) (Dul, 2020, Chapter 4). Run the analysis for the hypothesized relations or all the relations: Dependent = endogenous latent variable score(s), independent = preceding latent variable scores. Select the ceiling line based on theory or data pattern (Dul, 2020, Chapter 4).* Refer to the input data’s empirical scope.
Create a bottleneck table: Choose the desired level for the target construct (building on your knowledge of the field) and ensure that the bottleneck table identifies this level.
Step 7: Evaluate the structural model
After evaluating the VIFs of the inner model, evaluate the PLS-SEM model along the standard assessment criteria, especially the coefficient of determination ( ), the path coefficients’ statistical significance and relevance, and the predictive power (Guenther et al., 2023; Hair et al., 2022, Chapter 6; Sarstedt et al. 2021; Sharma et al., 2023; Shmueli et al., 2019).*
Visually inspect the scatter plot of the NCA results; inspect outliers as an NCA-specific robustness check (Dul, 2021a, Chapter 3.8 and Chapter 4.6). In case of outlier elimination, return to Step 3; otherwise continue.*
Evaluate the necessity effect size and its statistical significance (Dul, 2016; 2020, Chapter 4).
Step 8: Interpret the findings
Produce an extended IPMA chart: Position constructs on the basis of the standardized total effects on the -axis (importance) and the average latent variable scores’ performance on the -axis (performance); use the information from the NCA to indicate whether a construct is necessary or not (e.g., using a color code), and to indicate how they perform in terms of necessity (e.g., by adjusting the positioned antecedent constructs’ size). Researchers may use our Microsoft Excel template for this purpose (https://www.pls-sem.net/downloads/).
Note: *indicates that steps are updated compared to Richter et al. (2020)
Fig. 3. An extended step-by-step guide in the context of a combined IPMA (cIPMA).
Note: *indicates that steps are updated compared to Richter et al. (2020).
satisfaction studies, this could be employees’ job satisfaction score of 70 (on a scale from 0 to 100) (as in Hauff et al., 2015), while in a marketing project on customer satisfaction this could be a benchmark level of 75 or more, depending on the industry (on a scale from 0 to 100) (American Customer Satisfaction Index, 2022; Rigdon et al., 2010). Depending on these levels, researchers might need to change the presentation of the steps in their bottleneck tables to show the target constructs’ selected performance level (e.g., increasing the number of steps from 10 to 20 will provide information on the etc. levels).

4.5. Interpret the findings (step 8)

The findings’ interpretation leverages both the PLS-SEM results regarding the constructs’ importance and performance values, and the NCA results regarding the corresponding necessity conditions. In order to facilitate the interpretation, researchers should produce a figure that clarifies the antecedent constructs’ importance and average performance – in a standard PLS-SEM-based IPMA – as well as providing information on whether these antecedent constructs are necessary conditions for the selected outcome level and, if so, how they perform. Fig. 4 shows an example of such a graphical illustration. The chart presents the PLS-SEM-based IPMA results; that is, the antecedent constructs’ total effects on the target construct (importance) on the -axis and their average rescaled scores (performance) on the -axis. In addition, the chart distinguishes between necessary and not necessary constructs; that is, constructs that show at least a medium effect size and have statistical significance and those that do not. Constructs that are not necessary for achieving the target construct’s desired level are displayed as black circles, while necessary constructs are displayed as white circles. In respect of the example shown in Fig. 4, we assume a desired level of the target construct of . Next, building on the target construct’s desired level, the chart uses the bottleneck table information to identify the percentage of cases remaining below the antecedent constructs’ required levels (provided in the percentile display). The size of the bubbles of each of the antecedent constructs that are necessary conditions reflects this information. The larger the bubbles, the larger the percentage of cases that have not achieved the necessary condition’s required level. Large bubbles therefore indicate that, from a necessity perspective, researchers should focus their attention on this aspect. Examining the results in Fig. 4, we find that is not a necessary condition for improving the target construct, while and are. It also
Fig. 4. Combined IPMA (cIPMA).
Note: Extension of the example given in Fig. 1, Panel B. For the desired level of the outcome: – construct is not necessary; construct is necessary. The bubble sizes represent the percentage of cases that have not achieved the required desired level outcome.
becomes obvious that many cases have not achieved the required level. This makes highly relevant, since investments in other constructs will not increase the outcome in these cases unless the bottleneck is solved. Improving ‘s performance is therefore very important in this situation in order to enable the target construct ‘s desired performance.
By integrating information from the NCA, specifically from the bottleneck tables, the cIPMA enables an enriched interpretation of the findings, which we characterize in Fig. 5. For example, constructs mapped in the upper right corner – despite their high importance – are often outside the researcher’s focus, as they have already achieved a high performance level. However, if the NCA results suggest that these constructs have not achieved a sufficient necessity level, their improvement should still be prioritized. Similarly, constructs positioned in the lower left quadrant (i.e., those that have a low importance and performance) should not be written off as irrelevant per se. If these constructs do not achieve a desired necessity level, actions for improvement should be taken. The interpretation of the other cases in these quadrants (e.g., constructs achieving a satisfactory necessity level) and in the other two quadrants is analogous.

5. Illustrative example of a combined importance-performance map analysis

While various models are suitable to serve as an example (e.g., Le et al., 2024; Siyal et al., 2024), we illustrate the application of a cIPMA by drawing on an extended TAM version (Davis, 1989), which has served as a blueprint for researching consumer behavior in various marketing and consumer behavior contexts. Richter et al. (2020) introduced the conceptual model and the relevant theoretical arguments and hypotheses in their demonstration of PLS-SEM’s and NCA’s combined use. The data and PLS-SEM analysis are also well documented. Consequently, we do not present all the steps in detail, but concentrate on aspects that are relevant for the cIPMA.

5.1. The conceptual model, data, and PLS-SEM analyses (steps 1 to 3)

The TAM under consideration comprises two endogenous constructs: the behavioral intention to adopt a technology, which leads to the actual technology use (e.g., Ajzen, 1991; Davis et al., 1989; Sheppard et al., 1988; Turner et al., 2010). It has four key exogenous constructs that precede behavioral intention and technology use: compatibility, which reports the innovation’s fit with the customer’s lifestyle and values, the perceived usefulness and perceived ease of use, and the emotional value, which measures whether customers enjoy or have positive feelings when using a product (Sheth et al., 1991). Fig. 6 indicates the conceptual
Fig. 5. Managerial recommendations based on the combined IPMA (cIPMA).
Fig. 6. Conceptual model.

model.

We used a sample of e-book reader adopters in France ( ) and collected responses via an online survey. We implemented a single item and a 7-point Likert scale to measure the target construct technology use; the remaining constructs used reflective measurement models with items measured on 5-point Likert scales (for more information on the sample and descriptive statistics, see Table A1 in the Appendix and Richter et al., 2020). That is, the indicators used to measure a specific construct are all measured on the same scale. The dataset is freely accessible via Richter, Hauff, Kolev, and Schubring (2023a). We used the data to estimate the extended TAM by means of the SmartPLS 4 software (Ringle et al., 2022) – for details on the model estimation, see Richter et al. (2023b).

5.2. Evaluate the reliability and validity of the measurement models (step 4)

The measurement and structural models demonstrated appropriate statistical quality when evaluated by means of the standard assessment criteria (see Table A2 in the Appendix and Richter et al., 2020). To finalize Step 4, we need to conduct an additional IPMA requirements check. We do so by evaluating whether the outer weights in our model are positive, which they are (Table A2 in the Appendix). We therefore continue with the analyses without making further changes.

5.3. Run the IPMA and transfer the latent variable scores (step 5)

We run the IPMA in PLS-SEM, using the SmartPLS 4 software (Ringle et al., 2022). Since all the respondents made use of the entire scale range at the indicator level, the empirical and theoretical scales at the indicator level are identical. Hence, we do not need to adjust the scale’s theoretical minimum and maximum values in the IPMA. Following our guideline, we specified the analysis such that it produces rescaled latent variable scores on a scale from 0 to 100 . After running the IPMA, we create a new data file with the rescaled scores for our NCA.

5.4. Run the NCA (step 6)

We use the rescaled latent variable scores to run the NCA on the construct scores generated in PLS-SEM. In the NCA, we use the empirical scope (to run the NCA in R, see Dul (2021b); please note that the NCA itself is a bivariate analysis not influenced by the other analyzed relationships). We create bottleneck tables to identify the antecedent constructs’ necessary levels that need to be satisfied. Since we want to evaluate our findings based on an outcome level of technology use of 85 (see Steps 7 and 8), we increase the number of steps to 20 in order to produce bottleneck tables that display this outcome level. We create two bottleneck tables (based on the CE-FDH ceiling line technique), one with
actual values, which provides the antecedent constructs’ necessary levels that the different levels of technology use need to achieve, and one with percentiles for the conditions, which informs us about the number and percentage of cases that have not achieved the required antecedent construct levels for the corresponding levels of technology use (see next step).

5.5. Evaluate the structural model (step 7)

In this step, we first evaluate the findings from the PLS-SEM analysis, including the IPMA. Thereafter we evaluate the NCA’s findings. We illustrate the evaluation of the target construct technology use.
Analyzing the standardized total effects from the PLS-SEM analysis (Table 1), we find that adoption intention has the strongest significant impact on technology use (0.437), followed by emotional value (0.362). The other constructs’ total effects are not significant. The constructs’ average importance is 0.225 . The explained variance of technology use is .
The IPMA results show the rescaled latent variables’ average performance scores (Table 2). Focusing on the direct antecedent constructs of technology use, we find that compatibility has the lowest average performance (61.557), while ease of use has the highest (75.640). This construct’s minimum (case-specific) performance score is 16.871, which suggests that no respondent evaluated all of the ease-of-use indicators at the lowest level. Taken jointly, the antecedent constructs have an average performance of 68.731. On examining the IPMA results in full, we learn that all constructs perform at fairly similar levels, with adoption intention and emotional value having the largest importance for technology use.
Next, we turn to the NCA results. First, we inspect the scatter plots visually. We find that all the scatter plots show an empty space in the upper left corner, indicating potential necessary conditions (see Figure A1 in the Appendix). None of the scatter plots indicates single cases with a particular influence on the ceiling line (ceiling outliers) or the scope (scope outliers).
Next, we consider the necessity effect sizes ‘s significance and size with regard to the target construct technology use. We followed Richter et al. (2020) and refer to the CE-FDH ceiling line (Table 3). We note that the selection of the ceiling (CE-FDH versus CR-FDH) line should be discussed based on (a) the data scaling (discrete versus continuous data), (b) the pattern of observations near the ceiling line (irregular versus linear), and (c) the potential theoretical ideas (indicating a straight ceiling line, see Dul, 2020, Chapter 4). We find that all the effect sizes are significant at and at least medium in size.
Table 1
Total effects from the PLS-SEM analysis.
Construct Technology use Statistically significant ( )?
Total effects 95% bootstrap confidence intervals
Emotional value 0.362 [0.200; 0.521] Yes
Ease of use 0.049 [-0.110; 0.215] No
Perceived usefulness 0.149 [-0.075; 0.357] No
Compatibility 0.127 [-0.108; 0.365] No
Adoption intention 0.437 [0.268; 0.609] Yes
Table 2
Performance values of latent variables (rescaled).
Construct Performance value, average Performance value, minimum Performance value, maximum
Technology use 49.713 0.000 100.000
Adoption intention 72.041 0.000 100.000
Emotional value 70.171 0.000 100.000
Ease of use 75.640 16.871 100.000
Perceived usefulness 64.248 0.000 100.000
Compatibility 61.557 0.000 100.000
Table 3
Necessity effect sizes (CE-FDH ceiling line).
Construct Technology use
Effect size d -value
Emotional value 0.33 0.000
Ease of use 0.24 0.016
Perceived usefulness 0.24 0.001
Compatibility 0.21 0.000
Adoption intention 0.29 0.000
Table 4
Bottleneck table technology use, actual values (based on the rescaled PLS-SEM latent variable scores from 0 to 100).
Technology use Adoption intention Emotional value Ease of use Perceived usefulness Compatibility
Actual values (rescaled 0-100)
0 NN NN NN NN NN
5 NN NN 25.373 NN NN
10 NN NN 25.373 NN NN
15 NN NN 25.373 NN NN
20 NN NN 25.373 NN NN
25 NN NN 25.373 NN NN
30 NN NN 25.373 NN NN
35 33.833 49.650 33.469 15.702 25.518
40 33.833 49.650 33.469 15.702 25.518
45 33.833 49.650 33.469 15.702 25.518
50 33.833 49.650 33.469 15.702 25.518
55 33.833 49.650 33.469 15.702 33.696
60 33.833 49.650 33.469 15.702 33.696
65 33.833 49.650 33.469 15.702 33.696
70 33.833 49.650 33.875 48.127 33.696
75 33.833 49.650 33.875 48.127 33.696
80 33.833 49.650 33.875 48.127 33.696
85 75.000 49.650 66.904 66.212 33.696
90 75.000 49.650 66.904 66.212 33.696
95 75.000 49.650 66.904 66.212 33.696
100 75.000 49.650 66.904 66.212 33.696
Building on the scores presented in Table 4, we subsequently identify the value levels of our antecedent constructs that need to be satisfied for a desired level of technology use. In respect of our analysis, we assume a desired outcome level of technology use of 85, which is a rather conservative (i.e., high) benchmark. The corresponding levels are 75 for adoption intention, 50 for emotional value, 67 for ease of use, 66 for perceived usefulness, and 34 for compatibility. That is, to achieve a technology use score of 85 (on a scale from 0 to 100), we need to achieve a score of 75 for adoption intention (on a scale from 0 to 100), 50 for emotional value (on a scale from 0 to 100), etc.
In addition, in Table 5 we identify the percentage of cases that do not achieve the antecedent constructs’ required levels. For example, of all cases did not achieve the necessary level of adoption intention to enable a score of technology use of 85 (on a scale from 0 to 100). The corresponding percentages for the other antecedents are for
emotional value, for ease of use, for perceived usefulness, and for compatibility. In the following, this information will be used to extend the initial IPMA results.

5.6. Interpret the findings (step 8)

In the following, we combine the results from Tables 1, 2 and 5 to design a combined importance-performance map that depicts its importance in the form of PLS-SEM’s total effects, and as well as its average performance in the form of the average rescaled latent variable scores of adoption intention, emotional value, ease of use, perceived usefulness, and compatibility as obtained from the PLS-SEM analyses. In addition, the combined importance-performance map shows whether the antecedent constructs are necessary conditions for technology use or not, and, if they are necessary, how many of the cases do not achieve the required levels. Fig. 7 shows the combined importance-performance map.
In line with our previous elaborations, we find that adoption intention, which is in the map’s upper right corner, is highly important and already shows high performance. However, adoption intention still needs prioritization, as it is the necessary condition for the target construct, technology use. More specifically, of the cases have not achieved adoption intention’s required level. Moreover, perceived usefulness and ease of use, which are in the left quadrants, require prioritization even though they are of little importance and already perform relatively well. Both are necessary conditions, but many cases do not achieve the required levels ( for perceived usefulness, for ease of use). Failure to prioritize these constructs, which may seem negligible at first glance given the PLS-SEM results, would limit other activities’ success regarding improving the technology use. These areas need to be addressed, because they are necessary conditions. Conversely, emotional value and compatibility are also necessary conditions, but only a few of these cases do not achieve the required levels (namely , and ); consequently, these constructs are relatively less critical in a practical setting.

6. Conclusion

In this paper, we introduced an enriched understanding of importance and performance that combines insights from PLS-SEM and NCA. Our paper makes several beneficial contributions for researchers and practitioners aiming to profit from PLS-SEM, NCA, and importanceperformance analyses in general.
Our overarching goal is to create awareness of a different logic for researchers and practitioners using PLS-SEM and IPMA; that is, the necessity logic, which offers several routes to complement findings from PLS-SEM’s sufficiency thinking. First, it complements findings with a new perspective on importance. Constructs might have a low IPMAbased importance classification in PLS-SEM (i.e., when they show a relatively low association with a target construct and, therewith, less potential to induce an increase in the targeted outcome). However, such constructs might be highly important from a necessity perspective. If these constructs are bottlenecks, their desired level will not be achieved. Second, the cIPMA provides researchers with a deeper understanding of the relevant performance levels to be achieved, since the NCA’s bottleneck information offers a fine-grained perspective on how much such information is actually needed to advance a target outcome to the next level. These insights benefit research investigating success’s antecedents in various contexts (e.g., job satisfaction, team performance, and customer satisfaction to name but a few). In addition, such insights are of high practical value, as they allow actions to be clearly identified and prioritized and, moreover, provide information on how much to invest in certain improvements. Third, to further assess the performance achieved in a sample (which is the normal approach in the PLS-SEM context), we propose analyzing how many cases still need to achieve the condition’s required level. In the case of multiple necessary
Table 5
Bottleneck table technology use: actual values for technology use (based on the rescaled PLS-SEM latent variable scores from 0 to 100) and the percentiles of antecedent constructs.
Technology use Adoption intention Emotional value Ease of use Perceived usefulness Compatibility
Actual values (rescaled 0-100) Percentage (and number) of cases that do not achieve the necessary levels
0 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
5 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
10 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
15 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
20 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
25 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
30 0 (0) 0 (0) 0.6 (1) 0 (0) 0 (0)
35 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 5.7 (10)
40 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 5.7 (10)
45 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 5.7 (10)
50 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 5.7 (10)
55 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 8.6 (15)
60 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 8.6 (15)
65 4.6 (8) 5.7 (10) 1.1 (2) 1.7 (3) 8.6 (15)
70 4.6 (8) 5.7 (10) 2.9 (5) 17.2 (30) 8.6 (15)
75 4.6 (8) 5.7 (10) 2.9 (5) 17.2 (30) 8.6 (15)
80 4.6 (8) 5.7 (10) 2.9 (5) 17.2 (30) 8.6 (15)
85 39.1 (68) 5.7 (10) 28.7 (50) 47.1 (82) 8.6 (15)
90 39.1 (68) 5.7 (10) 28.7 (50) 47.1 (82) 8.6 (15)
95 39.1 (68) 5.7 (10) 28.7 (50) 47.1 (82) 8.6 (15)
100 39.1 (68) 5.7 (10) 28.7 (50) 47.1 (82) 8.6 (15)
Fig. 7. Combined importance-performance map of technology use.
conditions, this information could help prioritize actions. Furthermore, our study also contributes to the broader research on importance and performance analyses, which face the challenge of having to define importance and performance, as well as the critical performance levels. We believe that the cIPMA is a step forward in this direction.
In addition to these contributions, there are also various areas on which future research could focus on. First, PLS-SEM and other methods that generate information on should-have factors, use the strength of the association between the constructs to determine the effect that an increase in the antecedent construct has on the target construct. These findings are mostly generated in respect of the full sample; that is, for all combinations of data involving low, medium, and high construct values. However, when combining PLS-SEM with NCA, it could be interesting to understand the association between the constructs involved when a specific bottleneck is bypassed. For instance, if the NCA points to a medium level antecedent construct being critical to achieve a desired high outcome level, it could be interesting to understand the association between the antecedent and the target construct of cases that have already achieved the required medium level. From a sufficiency logic
perspective, this could ultimately represent an increase in the antecedent construct’s further potential in the IPMA. Second, based on the PLS-SEM findings, the IPMA’s importance refers to all of the involved constructs’ total effects. In other words, they involve both the direct and indirect associations. These indirect or mediation effects are currently not explicitly considered in the proposed CIPMA. We therefore encourage future research to develop ideas for mediation relationships’ analysis in the NCA context, which could ultimately be used in a cIPMA.

CRediT authorship contribution statement

Sven Hauff: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Nicole Franziska Richter: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Marko Sarstedt: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Conceptualization. Christian M. Ringle: Writing – review & editing, Writing – original draft, Validation, Software, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

This research uses the statistical software SmartPLS (https://www. smartpls.com). Christian M. Ringle acknowledges a financial interest in SmartPLS.

Data availability

Data available as part of a Data in Brief article (https://doi. org/10.1016/j.dib.2023.109190)

Appendix

Table A1
Data description
Latent variable (measurement adapted from) Indicator Mean Range [Min; Max] S.D. Excess kurtosis Skewness
Emotional value, reflective (Sweeney and Soutar, 2001) EMV_01 Enjoyment 3.902 [1; 5] 0.842 1.942 -1.036
EMV_02 Pleasure 3.724 [1; 5] 0.887 0.940 -0.675
EMV_03 Relaxation 3.799 [1; 5] 0.877 1.465 -0.675
Ease of use, reflective (Moore and Benbasat, 1991) EOU_01 Learning duration 4.011 [1; 5] 0.988 0.800 -0.996
EOU_02 Operation 4.092 [1; 5] 0.811 0.798 -0.822
EOU_03 Menu navigation 3.971 [1; 5] 0.867 1.201 -0.904
Perceived usefulness, reflective (Antón et al., 2013; Moore and PU_01 General advantage 3.397 [1; 5] 0.970 -0.176 -0.296
Benbasat, 1991) PU_02 Practical application 3.598 [1; 5] 1.055 -0.106 -0.585
PU_03 Improvement of reading 3.293 [1; 5] 1.109 -0.534 -0.474
Compatibility, reflective (Huang and Hsieh, 2012; Moore and Benbasat, 1991) CO_01 Reading behavior 3.299 [1; 5] 0.996 -0.238 -0.419
CO_02 Consumption pattern 3.427 [1; 5] 0.991 0.259 -0.646
CO_03 Reading needs 3.655 [1; 5] 0.992 0.430 -0.829
Adoption intention, reflective (Venkatesh et al., 2012) AD_01 Future usage 4.023 [1; 5] 0.928 1.210 -1.046
AD_02 Daily usage 3.776 [1; 5] 0.972 0.360 -0.712
AD_03 Frequent usage 3.845 [1; 5] 0.925 0.869 -0.785
Technology use, single item (Venkatesh et al., 2012) USE_01 e-books 3.983 [1; 7] 1.610 -0.894 -0.063
Table A2
Results summary of (reflective) measurement models
Latent variables Indicators Weights Loadings Indicator communality (squared loadings) AVE Composite reliability Cronbach’s HTMT 95% bootstrap confidence interval does not include 1
Emotional value EMV_01 0.338 0.891 0.794 0.853 0.946 0.914 0.917 Yes
EMV_02 0.375 0.950 0.903
EMV_03 0.368 0.929 0.863
Ease of use EOU_01 0.453 0.784 0.615 0.697 0.873 0.783 0.873 Yes
EOU_02 0.371 0.878 0.771
EOU_03 0.380 0.840 0.706
Perceived usefulness PU_01 0.319 0.722 0.521 0.642 0.842 0.723 0.753 Yes
PU_02 0.426 0.819 0.671
PU_03 0.491 0.856 0.737
Compatibility CO_01 0.396 0.901 0.812 0.779 0.914 0.858 0.859 Yes
CO_02 0.357 0.906 0.821
CO_03 0.381 0.840 0.706
Adoption intention AD_01 0.345 0.933 0.870 0.889 0.960 0.938 0.939
AD_02 0.347 0.935 0.874
AD_03 0.368 0.960 0.922
Table A3
Bottleneck table in different formats
Bottleneck technology use, standardized latent variable scores percentage ranges (as in Richter et al., 2020)
Adoption intention Emotional value Ease of use* Perceived usefulness Compatibility
0 NN NN NN NN NN
10 NN NN 10.2 NN NN
20 NN NN 10.2 NN NN
30 NN NN 10.2 NN NN
40 33.8 49.6 20.0 15.7 25.5
50 33.8 49.6 20.0 15.7 25.5
60 33.8 49.6 20.0 15.7 33.7
70 33.8 49.6 20.5 48.1 33.7
80 33.8 49.6 20.5 48.1 33.7
90 75.0 49.6 60.2 66.2 33.7
100 75.0 49.6 60.2 66.2 33.7
Bottleneck technology use, rescaled latent variable scores, percentage ranges
0 NN NN NN NN NN
10 NN NN 10.2 NN NN
20 NN NN 10.2 NN NN
30 NN NN 10.2 NN NN
40 33.8 49.6 20.0 15.7 25.5
50 33.8 49.6 20.0 15.7 25.5
60 33.8 49.6 20.0 15.7 33.7
70 33.8 49.6 20.5 48.1 33.7
80 33.8 49.6 20.5 48.1 33.7
90 75.0 49.6 60.2 66.2 33.7
100 75.0 49.6 60.2 66.2 33.7
Bottleneck technology use, rescaled latent variable scores, actual scores
0 NN NN NN NN NN
10 NN NN 25.4 NN NN
20 NN NN 25.4 NN NN
30 NN NN 25.4 NN NN
40 33.8 49.6 33.5 15.7 25.5
50 33.8 49.6 33.5 15.7 25.5
60 33.8 49.6 33.5 15.7 33.7
70 33.8 49.6 33.9 48.1 33.7
80 33.8 49.6 33.9 48.1 33.7
90 75.0 49.6 66.9 66.2 33.7
100 75.0 49.6 66.9 66.2 33.7
*The standardized and rescaled latent variable scores produce the same bottleneck results if presented as percentage ranges, but differ regarding the actual values of the construct ease of use. This difference is due to the minimum value of ease of use being 16.9. Consequently, ease of use’s potential values range from 16.9 to 100 ; that is, 83.1. Transferring the percentage ranges, for instance of 10.2 or 20.0 , to rescaled actual scores therefore yields or 33.5.
-axis: Adoption intention, -axis: Technology use
Fig. A1. NCA scatter plots with CE-FDH ceiling lines.

References

Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50, 179-211.
Alyahya, M., Agag, G., Aliedan, M., Abdelmoety, Z.H., Daher, M.M., 2023. A sustainable step forward: understanding factors affecting customers’ behaviour to purchase remanufactured products. J. Retailing Consum. Serv. 70, 103172.
American Customer Satisfaction Index, 2022. Unparalleled Customer Intelligence. Retrieved from. https://www.theacsi.org/our-industries/.
Anderson, E.W., Mittal, V., 2000. Strengthening the satisfaction-profit chain. J. Serv. Res. 3 (2), 107-186.
Antón, C., Camarero, C., Rodríguez, J., 2013. Usefulness, enjoyment, and self-image congruence: the adoption of e-book readers. Psychol. Market. 30 (4), 372-384.
Bergh, D.D., Boyd, B.K., Byron, K., Gove, S., Ketchen, D.J., 2022. What constitutes a methodological contribution? J. Manag. 48 (7), 1835-1848.
Bolívar, L.M., Roldán, J.L., Castro-Abancéns, I., Casanueva, C., 2022. Speed of international expansion: the mediating role of network resources mobilisation. Manag. Int. Rev. 62, 541-568.
Cai, X., Cebollada, J., Cortiñas, M., 2023. Impact of seller- and buyer-created content on product sales in the electronic commerce platform: the role of informativeness, readability, multimedia richness, and extreme valence. J. Retailing Consum. Serv. 70, 103141.
Cheah, J.-H., Magno, F., Cassia, F., 2023. Reviewing the smartPLS 4 software: the latest features and enhancements. Journal of Marketing Analytics. Advance online publication.
Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13 (3), 319-340.
Davis, F.D., Bagozzi, R.P., Warshaw, P.R., 1989. User acceptance of computer technology: a comparison of two theoretical models. Manag. Sci. 35 (8), 982-1003.
Dul, J., 2016. Necessary condition analysis (NCA): logic and methodology of “necessary but not sufficient” causality. Organ. Res. Methods 19 (1), 10-52.
Dul, J., 2020. Conducting Necessary Condition Analysis. Sage, London.
Dul, J., 2021a. Advances in Necessary Condition Analysis. Online book retrieved from:, Version 0.1. https://bookdown.org/ncabook/advanced_nca2/.
Dul, J., 2021b. Necessary Condition Analysis (NCA) with R. A Quick Start Guide: 2 March 2021. Online document retreived from:, Version 3.1.0. https://www.erim.eur.nl
/fileadmin/user_upload/_generated/download/Quick_Start_Guide_NCA_3.1.0_Ma rch_2_2021.pdf.
Dul, J., 2022. R Package NCA: Necessary Condition Analysis. Version:, 3.2.0. https://cr an.r-project.org/web/packages/NCA/index.html.
Dul, J., Hauff, S., Bouncken, R.B., 2023. Necessary condition analysis (NCA): review of research topics and guidelines for good practice. Rev. Manag. Sci. 17 (2), 683-714.
Dul, J., Hauff, S., Tóth, Z., 2021. Necessary condition analysis in marketing research. In: Nunkoo, R., Teeroovengadum, V., Ringle, C.M. (Eds.), Handbook of Research Methods for Marketing Management. Edward Elgar, pp. 51-72.
Fornell, C., 1992. A national customer satisfaction barometer: the Swedish experience. J. Market. 56 (1), 6-21.
Fornell, C., Johnson, M.D., Anderson, E.W., Cha, J., Bryant, B.E., 1996. The American satisfaction index: nature, purpose, and findings. J. Market. 60 (4), 7-18.
Guenther, P., Guenther, M., Ringle, C.M., Zaefarian, G., Cartwright, S., 2023. Improving PLS-SEM use for business marketing research. Ind. Market. Manag. 111, 127-142.
Hair, J.F., Hult, G.T.M., Ringle, C.M., Sarstedt, M., 2022. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), 3 ed. Sage, Thousand Oaks, CA.
Hair, J.F., Hult, G.T.M., Ringle, C.M., Sarstedt, M., Thiele, K.O., 2017. Mirror, mirror on the wall: a comparative evaluation of composite-based structural equation modeling methods. J. Acad. Market. Sci. 45 (5), 616-632.
Hair, J.F., Risher, J.J., Sarstedt, M., Ringle, C.M., 2019. When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 31 (1), 2-24.
Hair, J.F., Sarstedt, M., Ringle, C.M., Gudergan, S.P., 2024. Advanced Issues in Partial Least Squares Structural Equation Modeling (PLS-SEM), 2 ed. Sage, Thousand Oaks, CA.
Hair, J.F., Sarstedt, M., Ringle, C.M., Mena, J.A., 2012. An assessment of the use of partial least squares structural equation modeling in marketing research. J. Acad. Market. Sci. 40 (3), 414-433.
Hauff, S., Guerci, M., Dul, J., van Rhee, H., 2021. Exploring necessary conditions in HRM research: fundamental issues and methodological implications. Hum. Resour. Manag. J. 31 (1), 18-36.
Hauff, S., Richter, N.F., Tressin, T., 2015. Situational job characteristics and job satisfaction: the moderating role of culture. Int. Bus. Rev. 24 (4), 710-723.
Huang, L.-Y., Hsieh, Y.-J., 2012. Consumer electronics acceptance based on innovation attributes and switching costs: the case of e-book readers. Electron. Commer. Res. Appl. 11 (3), 218-228.
Huete-Alcocer, N., Hernandez-Rojas, R.D., 2022. Do SARS-CoV-2 safety measures affect visitors experience of traditional gastronomy, destination image and loyalty to a world heritage city? J. Retailing Consum. Serv. 69, 103095.
Kristensen, K., Martensen, A., Grønholdt, L., 2000. Customer satisfaction measurement at post Denmark: results of application of the European customer satisfaction index methodology. Total Qual. Manag. 11 (7), S1007-S1015.
Le, H.T.P.M., Kim, D., Park, J., 2024. The way to generate customer citizenship behavior with customer experience. J. Retailing Consum. Serv. 76, 103608.
Lohmöller, J.-B., 1989. Latent Variable Path Modeling with Partial Least Squares. Physica, Heidelberg.
Martilla, J.A., James, J.C., 1977. Importance-performance analysis. J. Market. 41 (1), 77-79.
Moore, G.C., Benbasat, I., 1991. Development of an instrument to measure the perceptions of adopting an information technology innovation. Inf. Syst. Res. 2 (3), 192-222.
Ng, F.Z.-X., Yap, H.-Y., Tan, G.W.-H., Lo, P.-S., Ooi, K.-B., 2022. Fashion shopping on the go: a dual-stage predictive-analytics SEM-ANN analysis on usage behaviour, experience response and cross-category usage. J. Retailing Consum. Serv. 65, 102851.
Oh, H., 2001. Revisiting importance-performance analysis. Tourism Manag. 22 (6), 617-627.
Pappas, N., 2023. Came and gone? A longitudinal study of the effects of COVID-19 on tourism purchasing intentions. J. Retailing Consum. Serv. 72, 103269.
Perez-Aranda, J., Gonzáles Robles, E.M., Alarcón Urbistondo, P., 2023. Understanding antecedents of continuance and revisit intentions: the case of sport apps. J. Retailing Consum. Serv. 72, 103288.
Richter, N.F., Hauff, S., 2022. Necessary conditions in international business research: advancing the field with a new perspective on causality and data analysis. J. World Bus. 57 (5), 101310.
Richter, N.F., Hauff, S., Gudergan, S.P., Ringle, C.M., 2022. The use of partial least squares structural equation modeling and complementary methods in international management research. Manag. Int. Rev. 62 (4), 449-470.
Richter, N.F., Hauff, S., Kolev, A.E., Schubring, S., 2023a. Dataset on an extended technology acceptance model: a combined application of PLS-SEM and NCA. Data Brief 48, 109190.
Richter, N.F., Hauff, S., Ringle, C.M., Sarstedt, M., Kolev, A.E., Schubring, S., 2023b. How to apply necessary condition analysis in PLS-SEM. In: Latan, H., Hair, J.F., Noonan, R. (Eds.), Partial Least Squares Path Modeling: Basic Concepts, Methodological Issues and Applications. Springer, pp. 267-297.
Richter, N.F., Martin, J., Hansen, S.V., Taras, V., Alon, I., 2021. Motivational configurations of cultural intelligence, social integration, and performance in global virtual teams. J. Bus. Res. 129, 351-367.
Richter, N.F., Schubring, S., Hauff, S., Ringle, C.M., Sarstedt, M., 2020. When predictors of outcomes are necessary: guidelines for the combined use of PLS-SEM and NCA. Ind. Manag. Data Syst. 120 (12), 2243-2267.
Rigdon, E.E., Ringle, C.M., Sarstedt, M., 2010. Structural modeling of heterogeneous data with partial least squares. In: Malhotra, N.K. (Ed.), Review of Marketing Research, vol. 7. Sharpe, Armonk, NY, pp. 255-296.
Ringle, C.M., Sarstedt, M., 2016. Gain more insight from your PLS-SEM results: the importance-performance map analysis. Ind. Manag. Data Syst. 116 (9), 1865-1886.
Ringle, C.M., Wende, S., Becker, J.-M., 2022. SmartPLS 4. Oststeinbek: SmartPLS. Retrieved from. https://www.smartpls.com/.
Rodríguez, P.G., Villarreal, R., Valiño, P.C., Blozis, S., 2020. A PLS-SEM approach to understanding e-SQ, e-satisfaction and e-loyalty for fashion e-retailers in Spain. J. Retailing Consum. Serv. 57, 102201.
Sarstedt, M., Hair, J.F., Pick, M., Liengaard, B.D., Radomir, L., Ringle, C.M., 2022. Progress in partial least squares structural equation modeling use in marketing research in the last decade. Psychol. Market. 39 (5), 1035-1064.
Sarstedt, M., Hair, J.F., Ringle, C.M., Thiele, K.O., Gudergan, S.P., 2016. Estimation issues with PLS and CBSEM: where the bias lies. J. Bus. Res. 69 (10), 3998-4010.
Sever, I., 2015. Importance-performance analysis: a valid management tool? Tourism Manag. 48, 43-53.
Shahidi, N., Tossan, V., Bourliataux-Lajoinie, S., Cacho-Elizondo, S., 2022. Behavioral intention to use a contact tracing application: the case of StopCovid in France. J. Retailing Consum. Serv. 68, 102998.
Sheppard, B.H., Hartwick, J., Warshaw, P.R., 1988. The theory of reasoned action: a meta-analysis of past research with recommendations for modifications and future research. J. Consum. Res. 15 (3), 325-343.
Sheth, J.N., Newman, B.I., Gross, B.L., 1991. Why we buy what we buy: theory of consumption values. J. Bus. Res. 22 (2), 159-170.
Siyal, A.W., Chen, H., Jamal Shah, S., Shahzad, F., Bano, S., 2024. Customization at a glance: investigating consumer experiences in mobile commerce applications. J. Retailing Consum. Serv. 76, 103602.
Skok, W., Kophamel, A., Richardson, I., 2001. Diagnosing information systems success: importance-performance maps in the health club industry. Inf. Manag. 38 (7), 409-419.
Slack, N., 1994. The importance-performance matrix as a determinant of improvement Priority. Int. J. Oper. Prod. Manag. 44 (5), 59-75.
Streukens, S., Leroi-Erelds, S., Willems, K., 2017. Dealing with nonlinearity in importance-performance map analysis (IPMA): an integrative framework in a PLSSEM context. In: Latan, H., Noonan, R. (Eds.), Partial Least Squares Path Modeling. Springer, Cham, pp. 367-403.
Sukhov, A., Olsson, L.E., Friman, M., 2022. Necessary and sufficient conditions for attractive public transport: combined use of PLS-SEM and NCA. Transport. Res. Pol. Pract. 158, 239-250.
Sweeney, J.C., Soutar, G.N., 2001. Consumer perceived value: the development of a multiple item scale. J. Retailing 77 (2), 203-220.
Tiwari, P., Kaurav, R.P.S., Koay, K.Y., 2023. Understanding travel apps usage intention: findings from PLS and NCA. Journal of Marketing Analytics. advance online publication.
Turner, M., Kitchenham, B., Brereton, P., Charters, S., Budgen, D., 2010. Does the technology acceptance model predict actual use? A systematic literature review. Inf. Software Technol. 52 (5), 463-479.
Venkatesh, V., Thong, J.Y.L., Xu, X., 2012. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Quarterly 36 (1), 157-178.
Wold, H., 1982. Soft modeling: the basic design and some extensions. In: Jöreskog, K.G., Wold, H. (Eds.), Systems under Indirect Observations: Part II. North-Holland, Amsterdam, pp. 1-54.
Zhang, P., Chao, C.-W., Chiong, R., Hasan, N., Aljaroodi, H.M., Tian, F., 2023. Effects of in-store live stream on consumers’ offline purchase intention. J. Retailing Consum. Serv. 72, 103262.
Bokrantz, J., Dul, J., 2023. Building and testing necessity theories in supply chain management. J. Supply Chain Manag. 59 (1), 48-65.
Sarstedt, M., Mooi, E.A., 2019. A Concise Guide to Market Research: The Process, Data, and Methods Using IBM SPSS Statistics. 3, ed. Springer, Berlin, Heidelberg.
Sharma, P.N., Liengaard, B.D., Hair, J.F., Sarstedt, M., Ringle, C.M., 2023. Predictive model assessment and selection in composite-based modeling using PLS-SEM: extensions and guidelines for using CVPAT. Eur. J. Market. 57 (6), 1662-1677.
Shmueli, G., Sarstedt, M., Hair, J.F., Cheah, J., Ting, H., Vaithilingam, S., Ringle, C.M., 2019. Predictive model assessment in PLS-SEM: guidelines for using PLSpredict. Eur. J. Market. 53 (11), 2322-2347.

    • Corresponding author. Ludwig-Maximilians-University Munich, LMU Munich School of Management, Ludwigstraße 28 RG/III, 80539, Munich, Germany.
    E-mail addresses: hauff@hsu-hh.de (S. Hauff), nicole@sam.sdu.dk (N.F. Richter), sarstedt@lmu.de (M. Sarstedt), c.ringle@tuhh.de (C.M. Ringle).
  1. We refer to a trichotomous illustration here to facilitate interpretation; however, the NCA is not limited to specific measurement levels. For more information, see Dul (2020, Chapter 3). Also note that the symbols used in the NCA context have a different meaning than in the PLS-SEM context. While the latter usually uses to denote latent constructs and to denote indicators, the former uses Y for outcomes and X for conditions.
  2. Note that the bottleneck levels differ due to the different scales involved. Also note that this statement refers to the standardized, unstandardized or rescaled scores calculated in the IPMA context on the basis of a linear transformation. Other forms of transformation, particularly non-linear transformations, should be carefully considered, since they could affect the NCA’s results.
  3. Under https://www.pls-sem.net/downloads, we provide a Microsoft Excel template for downloading in order to produce this kind of indication.
  4. In the Appendix, we offer a comparison of bottleneck tables for the standardized latent variable scores based on percentage ranges, the rescaled latent variable scores based on percentage ranges, and the bottleneck table for rescaled latent variable scores based on the actual values. We also comment on the differences (see Table A3 in the Appendix).