Research Article

A Simulation Study to Assess the Effect of the Number of Response Categories on the Power of Ordinal Logistic Regression for Differential Item Functioning Analysis in Rating Scales

Table 3

The type I error of OLR model under different combinations when .

ConditionsRatioDIF = 0.5DIF = 1

10.030.030.030.050.070.07
0.030.030.030.050.050.06
0.030.030.030.040.050.05

20.030.030.030.050.060.06
0.030.040.040.050.050.07
0.030.040.040.050.040.05

30.030.030.030.040.040.05
0.030.030.030.040.040.05
0.030.030.030.040.040.04

40.030.030.030.050.050.05
0.030.030.030.040.050.05
0.030.030.030.040.040.04

50.040.040.040.060.070.08
0.040.040.040.060.070.07
0.040.040.040.060.060.07

60.030.030.030.070.050.04
0.030.030.030.040.050.04
0.030.030.030.040.040.04

70.040.050.050.070.080.09
0.040.040.050.070.080.08
0.040.040.040.060.070.07

80.040.050.050.080.090.10
0.040.060.060.080.090.10
0.040.050.050.070.080.09

90.060.060.060.070.080.10
0.050.050.060.060.080.09
0.040.050.050.060.060.08

: number of response categories.
: and , : and , and : and .
The conditions are described in Table 1.