If you are completing an experiment, measuring how behaviour outcomes and psychological resources change from time one (before an intervention) to time two (after an intervention), and you want to now complete regression analysis to look at whether any of the psychological resources predicted change in behaviour, should you only conduct this regression with psychological resource that significantly changed from time one to time two and why?
Also if I'm conducting a regression is it better to perform multiple simple regressions in this care or one multiple regression?
please help! thank you
You shouldn’t limit your predictors to only those that changed significantly over time, a variable can still help predict "behavioral change" even if it didn't change itself. Also, a multiple regression is preferable here because it allows you to assess the unique contribution of each "psychological resource" while controlling for the others. Multiple simple regressions are generally unwise as you can get spurious trends and issues with type I errors.
Hi, I think you should not conduct this regression with psychological resource that significantly changed between t1 and t2 because of the risk of introducing biases. Multiple regression should be more suitable to your data than simple regression. But it could be possible that SCED fits better to your aims, see https://sced-for-clinicians.netlify.app/resources/shiny-apps/ and the related paper by R. Manolov). Another tip : use JASP software and its useful guide to regression analysis !
En termes simple, on peut bien utiliser l'analyse de la regression lors de la prédiction ou quand on évalue l'effet d'une variable sur une autre.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication