Thread by @mcmullarkey

New preprint from @JSchleiderPhD & me: Emotion and anxiety mindsets share little unique variance with internalizing problems in adults once you account for hopelessness (Ns = 200, 430)

Open code & data + interpretations in this thread!

psyarxiv.com/qtrxs/

We used commonality analysis (CA), which allows us to directly examine how much predictive variance is unique & shared among predictors

This technique can help us identify important individual predictors even when they're highly correlated (A no-no in traditional linear models)

If you want to try it out for yourself, the code and the data are part of this OSF project!

osf.io/wrc2m/

You can also apply the code to your own data. Would be great to see more CA papers out there given how often relevant predictors are highly correlated

By bootstrapping the unique estimates in CA, we can also determine which predictors share significantly more unique variance with the outcome of interest

In every case, hopelessness predicted more unique variance than emotion and anxiety mindsets

These differences weren't small either. The only time a mindset shared reliably non-negligible, unique variance with an outcome (anxiety mindsets with sum score anxiety in Study 2) hopelessness shared ~7 times more variance with the same outcome (22.79% vs. 3.16%)

Since we can directly estimate the shared variance among the predictors, we also show ≥ 91% of the variance mindsets predict in internalizing problems is shared with hopelessness

In other words, nearly all variance that mindsets shared with depression and anxiety were accounted for by concurrent hopelessness levels

Does this mean we should abandon mindset style interventions for internalizing? We don't think so

These formats are brief, scalable, and far more accessible than most interventions for internalizing problems. We should just put more effort into finding their active ingredients

This is a cross-sectional, correlational analysis, so cause is *peers into the distance* nowhere in sight

But these analyses allow us to expand our ideas of what the active ingredients for these brief interventions might be

We traditionally think of hopelessness as something that takes a long time to change

Based on these analyses, running experiments to determine whether that's the case could be worthwhile

If you're interested in commonality analysis, you can also check out these other papers, including another one where hopelessness shares a lot of unique variance with internalizing (Full disclosure: Igor and I collaborate)

sciencedirect.com/science/articl…

journals.plos.org/plosone/articl…