Bias in student survey findings from active parental consent procedures.


Increasingly, researchers are required to obtain active (explicit) parental consent prior to surveying children and adolescents in schools. This study assessed the potential bias present in a sample of actively consented students, and in the estimates of associations between variables obtained from this sample. Students (n = 3496) from 36 non-government metropolitan schools in Perth, Western Australia completed an online survey in 2010 as part of the Cyber Friendly Schools Project. Students with active (35%) and passive (65%) parental consent were compared on a range of variables including demographic, bullying and social-emotional outcomes. The moderating effects of consent status were also tested. Comparisons of the two consent groups showed that older students and students involved in problem behaviours such as bullying others, with lower pro-social scores, who lived with one parent and reported doing less well academically than their peers, were underrepresented in the sample with active parental consent. Additionally, consent status was a significant moderator of the associations between bullying victimisation and certain social-emotional variables. Active only parental consent leads to biased samples and biased estimates of associations between outcomes of interest, which could lead to miss-targeted behavioural policies and interventions. Strategies to boost response rates to levels sufficient to warrant the conduct of the research are labour-intensive and costly, and the obtained samples are still likely to be biased. For low risk research, such as bullying surveys, rigorous active-passive consent procedures which result in higher participation rates, lower costs and reduced burden on teachers and schools, are recommended.