**4. Analysis**

Data collection involved employees working in the administrative, academic, managerial, and strategic sectors in the cities of Rio de Janeiro and São Paulo. After debugging and compiling the database, a total of 86 valid responses were obtained. The average age of respondents was 38 years. Of the total, 46.5% are male and 53.5% female. Regarding the level of education, the majority (64.70%) of respondents are postgraduates, then totaled 25.88% who have a degree. In Table 4, the data are presented in more detail.

Regarding the age of respondents, most are in the range between 30 and 35 years (21 individuals). FGV operates in various areas, from consulting to educational services. This segmentation is perceived by analyzing the position of respondents in the survey. Most of them, 26 individuals, were working in administrative positions. Next, consultants (13), coordinators (11), and managers (11). Academics and technicians totaling eight individuals in each category. In the last positions were strategic positions (five) and analysts (four).

Regarding working time, the majority (37,2%) of employees are in the organization between 1 and 5 years. Then, 25.6% of respondents are individuals who are between 5 and 10 years in the organization. In a smaller percentage, 18.6% of employees had worked in the organization for between 10 and 15 years. The extreme points (less than one year and more than 15 years) presented the lowest percentages, with 9.3% in each group. The data were analyzed using SPSS statistical analysis software (version 23) and SmartPLS (version 3.2.8). Additionally, Excel was used to tabulate the database.



The validation of the instrument began in the pretest phase in which 33 answers were obtained from employees of the institution. Items that presented corrected item-total correlation indices below 0.300 were removed, following the recommendations of Pedhazur and Schmelkin [53]. In the final version of the questionnaire, nine of the 38 items did not present the recommended indexes so that it was possible to proceed with the instrument validity and reliability tests [54,55] and, therefore, were withdrawn.

The KMO test resulted in an index of 0.837. Coefficients above 0.6 indicate the appropriateness of the sample for factor analysis [56]. The presence of correlation between the items was analyzed by Bartlett's sphericity test. The test was significant at a confidence level of 99% and an index of 1,813,873 was obtained. After debugging the instrument, a new collection was performed. The final number of valid answers was 86 questionnaires, so above the minimum number of 68 respondents.

Minimum sample estimation was performed using GPower software [57] using the parameters recommended by Cohen (1992) for the multiple linear regression method. Due to the predictive and exploratory character of the research, PLS-SEM was adopted as a statistical technique to test the hypotheses proposed in the model. For the implementation of technical statistics, the recommendations of Hair et al. have been implemented [58].

Additionally, the sample normality test indicated that the data are not normalized. This second finding reiterates the choice made by the data analysis method, whereas the literature largely recommends the choice of the PLS-SEM statistical technique over the covariance analysis (CB-SEM) technique when data are not normally distributed [59].
