Some implications of unamalgamating scale items

David Trafimow (New Mexico State University, USA)

The generally small but touted as ‘statistically significant’ correlation coeIicients in the social sciences jeopardize theory testing and prediction. To investigate these small coeIicients’ underlying causes, I consider traditional equations such as Spearman’s (1904) classic attenuation formula, Cronbach’s (1951) α, and Guilford and Fruchter’s (1973) equation for the eIect of additional items on a scale’s predictive power. These equations’ implications diIer regarding large interitem correlation coeIicients enhancing or diminishing predictive power. Contrary to conventional practice, such correlation coeIicients decrease predictive power when treating items as multi-item scale components but can increase predictive power when treating items separately. The implications are wide-ranging.

In-person: Lumsden Suite (Edinburgh Business School Building), Heriot-Watt University, Edinburgh; Online: click ‘Join via MS Teams’.