Page 127 - SUSTAINABILITY ISSUES & COVID-19

3.5 Control Variable
Unlike organization types, the number of employees’ information is relatively challenging to search. Employees’ number
changes over time, and organizations are not always transparent about employees’ data. Therefore, respondents were
asked to classify organizations size into six categories to minimize the number of employees’ volatility and uncertainty.

3.6 Statistical Model
Five statistical tests are performed in this research: statistic descriptive, correlation test, assumption test, Kernel-Based
Regularized Least Squares (KRLS), and robustness test. KRLS is used if the data are not normally distributed. KRLS allows
researchers to regress and classify problems without assumption tests (Ferwerda, Hainmueller, & Hazlett, 2015). The
result provides predicted values, variances, and the pointwise partial derivatives that characterize each independent
variable’s marginal effects at each data point in the covariate space. Therefore, this method accommodates a nonlinear
relationship with heterogeneous marginal effects that usually exist in social science research.

4. RESULT AND ANALYSIS

4.1 Descriptive results
The respondent consisted of 171 participants, as shown in Table 4. These respondents represent 51 organizations
classified into three types and six sizes, as shown in Table 5. Also, 266.609 tweets were collected to represent those 51
organizations. Then, tweets are analyzed using sentiment analysis and transformed into 51 sentiment indexes with the
distribution shown in Figure 5. Figure 5 shows that the distribution of sentiment score is skewed to the left toward zero,
which indicates the tendency of positive sentiment in sample organizations. Sentiment scores range between -0.9543
to 0, with the average is -0.3415. All sample organizations have sentiment scores more than -1, which indicates that
all organizations have more positive tweets than negative tweets. Table 6 shows the correlation between all variables,
including the positive relationship (0.0678) between SMG and organizational legitimacy.

Table 4 Participant demographic data (n=171)
Description Group Frequencies Percentages
Gender Male 84 49.12
Female 87 50.88
Age 20-30 years old 102 59.6
31-40 years old 57 33.33
41-50 years old 11 6.43
More than 51 years old 1 0.58
Educational Background High School 3 1.75
Vocation 13 7.6
Bachelor 124 71.51
Engineer 1 0.58
Master Degree 30 17.54
Working Experience Less than 3 years 72 42.11
3-10 years 74 43.27
10-20 years 25 14.62

126 International Conference on Sustainability
(5 Sustainability Practitioner Conference)
Th