*3.2. Binary Logit Model*

To analyze the indicators affecting CSR adoption, we used the binary logit regression model. The dependent variable (CSR adoption) was a binary variable reaching the value of 1 (the company adopted CSR) or 0 (the company did not adopt CSR). The logit model, based on cumulative logistic probability functions, was computationally easier to use and could predict the probability of CSR adoption in the company. The results are presented in Table 5.

**Table 5.** The multivariate logistic regression model.


The results confirm a statistically significant positive impact of ROA on CSR adoption. From Table 5, we found that, with a 95% confidence level, only ROA and EV EBITDA ratios had a significant effect on the CSR adoption with a *p*-value of less than the significance level alpha = 0.05, although these effects were minor. In the logistic model, if the odds ratio was greater than 1, the higher the value of the variable, and the higher the odds were of implementing CSR. Only ROA had a positive effect on CSR adoption. That is, increasing the ROA ratio level increased the probability of adopting CSR. In terms of the Enterprise Value to EBITDA ratio, a statistically significant negative effect was observed. We did not confirm the influence on CSR adoption in terms of the other analyzed variables. Thus, only Hypothesis 1 was supported.

The area under the ROC curve was found to be 0.685 (Figure 1). Since the area under the curve was more than 0.5, and the closer the curve followed the left-hand border and then the top border of the ROC space, the more acceptable the model.

**Figure 1.** ROC curves for model results with AUC of 0.685.
