2.4. Statistical Analysis

Differences between patients with schizophrenia with moderate and severe impairments and between environment accessibility with and without barriers were assessed using an independent t-test for continuous variables and Pearson’s chi-squared test or Fisher’s exact test for categorical variables. Logistic regression was used to test the explanatory power of the factors and compute the odds ratio (OR) of each factor included in the Data Bank of Certification of Disability and Care Needs. Access barriers with a lower bound of 95% confidence interval (CI) of c-statistic >0.7 were considered adequate fit and acceptable to draw inferences from the dataset [28]. We applied a receiver operating characteristic (ROC) curve to determine the best probabilistic cutoff value for the SI performance score to classify patients experiencing an access barrier to e1 categories. We compared the RD in each category between patients with and without an access barrier to e1 categories, stratified for moderate and severe patients using a Mann–Whitney U test. According to Chang et al. [17], the mean RD values were estimated to be 0.06 and 0.04 in the patients with environmental access barrier group and the open access group, respectively, and the SDs were estimated to be 0.06 and 0.04, respectively. The effect size d was 0.39. For a 95% power at a 5% two-tailed significance level with the Mann–Whitney test (two groups), we required the complete data from at least 356 cases (178 in each study group) in order to detect significant differences in the RD with power analysis using G*Power 3.1 [29]. Data were analyzed using the Statistical Package for the Social Sciences (ver. 9.4, SAS, Institute Inc., Cary, NC, USA). Differences between groups were considered significant if the two-tailed p values were <0.05.