Abstract
Individual differences in face identity recognition abilities are present across the lifespan but require developmentally differentiated methods of assessment. Here, we examine the empirical validity of a widely used face identity recognition measure, the Cambridge Face Memory Test for Children (CFMT-C). Logistic mixed-effects modelling of a large data set (607 children, 5-12 years) replicates and extends the findings of the only previous normative study of the CFMT-C (Croydon et al., Neuropsychologia, 62, 60-67, 2014). This novel, analytical approach enables us to take into account sources of variability typically overlooked in a classical analysis. We consider variability introduced by the task, alongside variability across children, to provide the first comprehensive characterisation of the interactive effects of factors inherent to participants (e.g. age, gender, and ethnicity), and the test (stage: face learning, simple recognition, harder recognition) on face memory performance. In line with past findings, we clearly observed age-related improvement in the task. Additionally, and for the first time, we report that this developmental effect is significantly more pronounced in the later, harder stages of the task; that there is an effect of gender, with females having better performance; and that consideration of participant ethnicity or testing context did not alter the best fitting model of these data. These results highlight the value of applying multilevel statistical models to characterise the factors driving performance variability, providing evidence of the divergence in recognition abilities across genders and confirming the stability of the CFMT-C in assessing face recognition abilities across variable experimental contexts and with diverse participant groups.