Three outs away from losing and you just give an out away? It bothered me in the moment and it’s still bothering me. I get the satisfaction from small ball and a well-executed bunt — but that doesn’t make it the right call.

2/2 Suppose we want to quantify the association between A and B. One study on this association + a direct replication is a nice start — nothing more. To quantify it with confidence requires *dozens* of studies on the association using both overlapping and different methodologies. It takes a village.

FYI I found your point clear. (Unless I misunderstood it, and I’m good at misunderstanding things. But I think I got it).

I see Aaron’s point being mainly about interpretation. If they reported the same stats but interpreted the effects as minimal to negligible, without overinterpretation of model fit or low p-values, the remaining problems would be technical rather than fundamentally misleading about findings.

Even if effect estimation is the goal, sample size matters. The former is the effect observed in the sample, the latter impacts confidence that the sample effect approximates the population effect. Interpretation for the study in question should have been: effects seem tiny, with high confidence.

Also relevant that in suicide research NU robustly predicts ideation but not attempts among ideators: more evidence NU’s explanatory power is mostly about having more frequent and stronger negative affect, which in turn causes more impulsive acts — rather than an altered affect-impulse association.

Re this study, really nice. My own guess is that self-appraised NU will be mainly determined by frequency/intensity of NA — because more NA will then lead to more impulsive acts and consequences even if the affect-impulse association were perfectly constant across folks.

Appreciate this thread. Some ‘easy’ underutilized alternatives for contextualizing an effect size might include comparing it to other ‘benchmark’ effects using diff predictors: a) within the same dataset and analytic approach & b) from earlier studies using parallel methods/analytical approach.

realignforpalestine.org#what-we-stan...

PS 4/3 Other people can also make claims to having deep connections to the land of Israel. More than one people can live on the land with autonomy and self-determination.

3/3 What's prejudiced is: making Israel and Israelis the only country and people in history to deserve blanket cancellation, appropriating "Zionist" (a Jewish value of self-determination) into a political slur, and denying/pathologizing the deep and Indigenous Jewish connection to Israel. 🙏💕🕊️

2/3 Of course you can advocate for Palestinian safety and autonomy, of course you can criticize Israel's policies, and of course you can give voice to tragic histories. My two countries (Canada and the US) also have tragic, if different, histories that require ongoing reconciliation.

Hahahaha .. what's funny is that in my memory I tricked you into doing it. Apparently the joke was on both of us.

In the clinical psych world this allows me to do my psychic party trick: “Oh, you’re doing a cluster analysis … [initiate Spooky Voice] I already see two of your groouuppps… one high on everything… and one looowwwwwwwww…”

Resonates. For me 1) Thoughtful robustness checks should be part of every paper, AND 2) It should become normal for reviewers to suggest robustness checks they think are missing, BONUS AND 3) It should be normal for post-pub commentaries to pursue robustness checks.