5.5.2. Multinomial logit
In line with Judge et al. (2010) and Ramanna and Sletten (2011), we also apply a multinomial
logit regression analysis based on our full sample. The dependent variable (IFRS_SME) describes
countries’ adoption decisions regarding IFRS for SMEs pursuant to the statutory accounting laws and takes three different values (0 ¼ ‘prohibited’, 1 ¼ ‘optional’, and 2 ¼ ‘required’). The multinomial
logit model does not assume any ordering among the adoption levels. Instead, it compares
the voluntary adoption and mandatory adoption categories to the baseline decision, that is, nonadoption.
Our inferences remain unchanged. The results (untabulated) show that the coefficients
of our independent variables from the binary logit model in Table 6 remain statistically significant
with the expected signs for the group of Voluntary-Adopters and Mandatory-Adopters. However,
note that this regression analysis should be interpreted with caution, since the category Mandatory-
Adopters (required adoption) only contains six countries.