Table 3 Multivariable regression for both the linear endpoint overall rating and binary endpoint negative perception. Both binary logistic and linear regression analyses were performed within each category. Backwards likelihood elimination method was deployed for both binary logistic and linear regression in each of the three categories. For linear univariable regression, parameter estimates ± their standard deviation are displayed (maximum likelihood estimates). For binary univariable regression, the odds ratio with its 95% confidence interval is shown (Wald). The intercept for multivariable linear regression was 15.745 ± 0.198 for teacher’s performance (p < 0.001), 15.292 ± 0.176 for student’s self-perception (p < 0.001) and 14.771 ± 0.114 for structural aspects (p < 0.001)

From: Regression analyses of questionnaires in bedside teaching