Abstract
The question whether positive and negative affect ratings are indicators of a single bipolar dimension or whether they are markers of two dimensions, is usually explored by analysing associations between these ratings. Bipolarity is indicated when correlations between items of positive affect are of similar absolute height as those between positive and negative affect ratings (leading to a one-factor solution). However, the widely used product-moment correlation is susceptible to underestimate associations between opposite variables. The authors show that this underestimation bias is particularly strong for opposite affect items with strictly unipolar response formats, because these items tend to form L-shaped bivariate distributions. For this reason, it is recommended using polychoric correlations instead, which are shown to give unbiased estimates of association.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call