Research Article

A Machine Learning Approach to Assess Differential Item Functioning in Psychometric Questionnaires Using the Elastic Net Regularized Ordinal Logistic Regression in Small Sample Size Groups

Table 2

The type I error rates of the regularized (elastic net) and non-regularized OLR models in detecting moderate uniform DIF (DIF=0.4) when J=5.

IRatioNOLRRidgeElastic net OLRLASSO
w=0w=0.01w=0.02w=0.03w=0.04w=0.05w=0.06w=0.07w=0.1w=0.5w=1

5nr=nf1000.0370.0070.0070.0200.0280.0330.0360.0400.0420.0480.0660.068
1500.0400.0090.0080.0210.0290.0380.0420.0460.0500.0560.0750.076
2000.0420.0080.0070.0180.0280.0380.0460.0500.0560.0630.0800.081
3000.0520.0110.0100.0230.0380.0450.0510.0570.0600.0700.0880.089
4000.0560.0090.0080.0230.0370.0460.0530.0610.0660.0770.0980.100

5nr=2nf1000.0380.0120.0110.0180.0270.0340.0400.0420.0440.0510.0670.068
1500.0380.0060.0060.0180.0260.0340.0400.0450.0490.0580.0710.073
2000.0350.0080.0060.0200.0300.0350.0400.0430.0460.0540.0670.067
3000.0530.0100.0100.0290.0370.0450.0530.0590.0610.0700.0870.089
4000.0540.0120.0100.0210.0340.0460.0520.0560.0620.0700.0910.093

5nr=3nf1000.0360.0090.0090.0190.0290.0350.0390.0440.0460.0520.0690.071
1500.0390.0080.0070.0180.0300.0340.0400.0480.0510.0580.0720.074
2000.0360.0100.0080.0210.0300.0370.0410.0460.0500.0550.0690.070
3000.0410.0100.0090.0200.0300.0380.0440.0480.0530.0610.0750.077
4000.0520.0110.0080.0250.0360.0430.0490.0550.0590.0680.0880.090

λBIC-0.3800.3810.1900.1300.0950.0760.0630.0540.0380.0080.004

10nr=nf1000.0290.0090.0080.0210.0290.0340.0370.0400.0410.0460.0560.056
1500.0300.0100.0090.0200.0300.0350.0380.0420.0450.0500.0580.059
2000.0300.0120.0100.0220.0300.0360.0400.0430.0450.0500.0580.059
3000.0310.0100.0080.0220.0280.0340.0380.0410.0440.0480.0580.059
4000.0310.0100.0090.0200.0270.0320.0350.0400.0410.0460.0550.056

10nr=2nf1000.0290.0090.0090.0200.0270.0310.0360.0380.0400.0450.0550.055
1500.0310.0120.0110.0220.0320.0380.0430.0450.0470.0500.0590.059
2000.0270.0090.0080.0190.0270.0340.0370.0400.0420.0460.0560.057
3000.0270.0110.0090.0200.0280.0330.0370.0390.0410.0460.0550.056
4000.0310.0100.0090.0210.0290.0350.0390.0410.0430.0470.0570.058

10nr=3nf1000.0290.0100.0090.0210.0280.0310.0350.0370.0400.0440.0530.053
1500.0300.0090.0090.0220.0300.0340.0380.0410.0430.0480.0580.059
2000.0240.0090.0080.0170.0240.0300.0330.0360.0380.0420.0500.051
3000.0340.0110.0100.0230.0310.0360.0400.0430.0450.0490.0580.059
4000.0280.0090.0080.0210.0270.0320.0360.0380.0410.0450.0540.055

λBIC-0.3150.3150.1600.1050.0800.0630.0520.0450.0320.0060.003

Note: DIF: differential item functioning; I: number of items in the scale; J: number of response categories; LASSO: least absolute shrinkage and selection operator; OLR: ordinal logistic regression; w: weighting parameter; Ratio: sample size ratio between the focal and reference groups; nf and nr indicate sample sizes in the focal and reference groups, respectively; N: total sample size (N=nf +nr). These λ values were obtained according to the Bayesian information criterion (BIC).