Estimation and Validation of the “c” Factor for Overall Cerebral Functioning in the Philadelphia Neurodevelopmental Cohort

Psychopathology and cognitive ability indicators correlate, within each other, in a way that is well-captured by hierarchical factor models, but integrating both into a single hierarchical framework remains a challenge. Because both aspects of behavior presumably reflect aberrant cerebral function, the highest-order latent variable in such a model would be the “c” (for cerebral) factor. We estimated a tri-factor model of “c” in N=9,494 children and young adults from the Philadelphia Neurodevelopmental Cohort using the Penn Computerized Neurocognitive Battery and the GOASSESS clinical interview. We tested the validity of factor scores resulting from this model by relating them to external criteria, including global functioning, parent education, socioeconomic status, intracranial volume, and longitudinal clinical outcome. Fit of the structural model was acceptable (CFI = 0.98; SRMR = 0.030), and scores correlated with external criteria as expected (functioning = 0.27; parent education = 0.43; socioeconomic status = 0.47; intracranial volume = 0.39; clinical outcome Cohen d = 0.30 and 0.57). For most criteria, the effect for the “c” factor was larger than either the “p” or “g” factors alone. These results provide evidence for the feasibility and potential utility of modeling the “c” factor when cognitive and clinical data are both available.