Investigating measurement invariance in computer-based personality testing: The impact of using anchor items on effect size indices