Quantitative lunch -- Dr. Alex Christensen (Penn State).

Scales are used in almost every psychological study. Sum (or average) scores of scales are a common approach yet computing sum scores when factor scores are more appropriate can reduce validity and stifle substantive theory. Recently, psychometric network analysis has emerged as another model, rapidly rising in use. The addition of network modeling raises further questions about the effects of when an inappropriate model is used to score scales. In this talk, I introduce network models, demonstrate the relationship between their parameters (e.g., communities, centralities) and factor model parameters (e.g., dimensions, loadings), and present a novel neural network algorithm to determine whether data are generated from a factor or network model. Using an empirical example, I demonstrate the potential consequences of applying the (in)appropriate model. I close with discussion on the importance of using the appropriate model (even when the stakes are low)