*3.3. Data Collection and Analysis Procedure*

The survey questionnaires in SurveyMonkey (Momentive Inc., San Mateo, CA, USA) ran in September and October 2021. The data was inserted into a data matrix in SPSS 23.0 (https://www.ibm.com/support/pages/downloading-ibm-spss-statistics-23, accessed on 1 November 2021) software for further analysis. Initially, the data were screened for normality. Then an EFA was carried out to identify the number of factors that explain green marketing, as this technique helps to understand and clarify new scales [36]. EFA aims to reduce the scale dimensionality, pin down the subjacent dimensions, and guide the subsequent CFA. It is a transformative statistical tool to find out the underlying dimensions and convert them into new variables to use. As in performing it, some scale's variables are deleted. Therefore, it is a valuable tool to enhance the reliability of the final obtained scale [37].

The CFA using Amos 23.0 (https://www.ibm.com/support/pages/downloadingibm-spss-amos-23, accessed on 5 November 2021) was then performed to validate the scales empirically. CFA is used to uphold a theory and is loosely based on the EFA since it starts with testing the obtained experimental dimensions structure [38]. For this reason, it goes beyond refining and validating the original scale [39] because it attempts to test the underlying dimensions derived from both the EFA and the supporting theories. Consequently, CFA aims to test the convergent validity of the scale inasmuch as it demonstrates that several items are rooted in the same factor. Similarly, it can indicate the discriminant validity if some things do not belong to the same element [37]. A CFA aimed at assessing construct validity, i.e., discriminant validity and convergent validity. The goodness of fit between the factor models was measured as discriminant validity. Convergent validity was tested using standardized factor loadings that indicated how acceptably latent variables explained each observed variable. The chosen combination of EFA and CFA was in agreement with Otaye-Ebede [40], who confirmed that EFA and CFA together may ensure higher accuracy and provide more robust evidence for a more valid instrument.

## **4. Results**

## *4.1. Results of Study 1*

First, we used a data set collected from 102 marketing managers (Study 1). The endorsement rates and variance were checked for each item. According to the scaling metric, the ideal values for the means are between two and four, while for standard deviation (SD), it is ≥ 0.80 [41]. Needless to say, while means describe the concentration of responses, standard deviation shows how dispersed they are. Each item fits into the suggested intervals, thus assuring an appropriate distribution. The normality check resulted in an absolute value of univariate skewness between −0.895 and 0.672, which fits the standards for absolute skewness below two [42]. Univariate kurtosis was between −0.990 and 2.099. It means that no item exhibited a severe deviation from a normal distribution. Therefore, no items were removed after this initial check. The Kaiser–Meyer–Olkin measure of sampling adequacy (0.872) greater than 0.6 and the Bartlett test of sphericity (5112.456 [df = 1326], *p* < 0.001) that was less than 0.05 [43] indicated that the application of factor analysis was appropriate.

We have examined the data matrix for underlying factors applying EFA with principal axis factoring and Promax rotation. The first EFA analysis resulted in a 12-factor solution that explained 78.03% of the variance. Aiming for a more meaningful solution, we have deleted the items if the load was equally heavy on more than one factor. Considering the relatively small sample size, the loadings smaller than 0.55 were deleted [44]. After every removal of items, we have rerun factor analysis and reestimated coefficients until we have received a satisfactory result. Revisions continued until every item remained factor loaded onto one factor with a loading value greater than 0.55. After deletions, the number of items was reduced to 29 (Table 3).

**Table 3.** Results of the exploratory factor analysis (EFA) for the retained Green Marketing Scale (GMaS) items (Study 1).

