Social Desirability Bias in Child-Report Social Well-Being: Evaluation of the Children’s Social Desirability Short Scale Using Item Response Theory and Examination of Its Impact on Self-Report Family and Peer Relationships

Research on child well-being largely relies on children’s self-report data, potentially biased by social desirability (SD). In this study, we aim to (1) evaluate the psychometric properties of the Children’s Social Desirability Short (CSD-S) scale, and (2) examine if and, if so, how SD systematically biases child-report family and peer relationships as indicators of social well-being. In spring 2015, 843 elementary school children (aged 10) and their parents were surveyed on well-being indicators and SD measured with the 14-items Children’s Social Desirability Short (CSD-S) scale. The CSD-S was evaluated using nonparametric Item Response Theory (NIRT). Linear mixed-effects regression models based on multiple imputations of multilevel missing data were run to examine the role of SD in self-report social well-being in addition to socio-demographic characteristics, accounting for the nested structure of the data (students were sampled at class level). Applying NIRT, we identified a 13-items subset of the CSD-S with double monotonicity. Cronbach’s alpha was.82. When controlling for children’s socio-demographic characteristics, SD significantly positively predicted subjective evaluations of family relationships (B = 0.04, t(49272) = 7.45, p <.001), whereas it significantly negatively predicted self-report deviant behavior performed towards peers (B = −0.03, t(39927) = −14.40, p <.001) and experienced from peers (B= −.0.01, t(39028) = −2.86, p =.002). SD bias explained additional 22 percent of variance in self-report deviant behavior performed towards peers. Since SD impacts the validity of self-report well-being, child indicators research should include age-specific SD scales, e.g., the CSD-S, and control for the bias in statistical analyses.