Jenny Howell

The SPHN list of advisors accepting students to work at the intersection of Social/Personalty & Health Psych is now updated for '23-'24 admissions: tinyurl.com/SPHNPositions Faculty can add themselves here: goo.gl/forms/BdJVz6zx… Please RT!

If you're moving over, I'm at jennyhowell dot bsky dot social.

@BThinkful @profcikara They're the standard metrics by which submissions are evaluated every year plus the EIA scores, which were designed and refined to best capture what the committees would want in decision-making: Whether or not EIA applied, and if the submission was particularly outstanding on EIA

@profcikara How does a scientist construct metrics as weak as these? Did you at least pre register what tests you would do?

Remember how SPSP asked at submission how the proposed work advanced equity, inclusion, and anti racism? And some people expressed concern at how that would affect the ultimate program? Pilot results are in:

@BThinkful @profcikara The bullet point is misleading—each single scientific metric predicted above EIA scores. You're right about the halo effect and we thought about that. It wouldn't've been ethical to introduce an experiment into actual decision-making.

@profcikara This seems like a meaningless evaluation. 1) It would be impossible for the full score not to be more predictive than a sub score. 2) Halo effects are common in evaluations. When people like one aspect it makes them look more favorably on others.

@dreddiewaldrep @profcikara You do have access to the data, just not anything that can identify people: osf.io/r27kz/ Would love to see your approach. Let me know if you need anything from me (e.g., some coding of statements) in that pursuit.

@profcikara Perhaps. I don't have access to the data. But a good Bayesian analysis would need to accout for prior probabilities of political bias in the field, ideological capture, activist "scholars", Anointed mindset impervious to evidence, etc. Example: unsafescience.substack.com/p/deconstructi…

@RogertheGS @profcikara Sorry, I mean "EIA." "D" is being added for this next round of the pilot.

@RogertheGS @profcikara DEIA scores did predict acceptance. It's just that NOT contributing to DEIA did not get you rejected out of hand and science is still essentail to acceptance. The decision-making committees liked having them. Worth expanding the pilot a third year, IMO.

@PaoloAPalma @RogertheGS @profcikara The goal is to allow the committee to consider DEIA when building the convention program, but to allow people to personally define how they contribute to that. 2023 was the 2nd year of the pilot; people seem to have missed year 1. spsp.org/events/annual-… <-more info from SPSP.

@datepsych Some insight from the "inside": SPSP does take the concerns seriously. That's why the pilot is being extended a year with an attempt to address some major concerns and reevaluation after another round. The original task-force suggestion was to implement these without any pilot.

@datepsych Thanks for the thread and the summary. Just want to clarify here: Saying you did not contribute to EIA predicted lower EIA scores (right column)—which makes sense, reviewers rated those statements as "not applicable" to EIA—but not lower acceptance (left column).

@Chrismartin76 @profcikara They then use DEIA scores alongside other things (e.g., topic area, teacher-scholar representation, interest value) to try to build a program that fulfills SPSP’s broader goals and represents the full spectrum of solid personality/social psychological science. 2/2

@Chrismartin76 @profcikara With the caveat that I'm no longer convention committee chair: Decision-makers use the DEIA scores as part of a holistic review of submissions. Top-rated science will essentially always get in: Score 4's on all scientific metrics, you’re practically guaranteed acceptance. 1/2

@colinsmithpsych I feel like… at least ruminating on my own inadequacy has a catchy tune now 🤷‍♀️

@Chops310 @WilliamMeese @pdakean “It’s technically worse than you think: A memoir about learning things, getting a degree, and being on my uniquely academic bullshit.”

@WilliamMeese @pdakean Technically, they often sell subscription bundles back to universities so...it's technically worse than you think

New study in @socscimed by @JennyLeeHowell et al. on parents' implicit and explicit beliefs with respect to vaccine refusal sciencedirect.com/science/articl…

@LaurietobrienO Interesting question! I have seen it on t-shirts, fwiw. I have also noticed the bias toward treating correlation as merely subservient to causal inference in conversations with folks trained in other experimental disciplines (e.g., chemistry).

@JennyLeeHowell So true! I have often wondered if "the correlational does not equal causation" refrain is echoed in other subdisciplines of psychology as much as social psych. The field has such a strong experimental bias, although I think it is changing.

Just because correlation ≠ causation doesn't mean correlation isn't important. Understanding when something is most likely to happen can be just as useful as understanding why it happens.

@artistexyz @j_colomb I agree that casual inference should be thoughtfully taught, but my original point was that correlation has utility outside of implying cause (like predicting weather) and we shouldn’t pigeonhole it to be only an indicator of cause. When can be important independent of why.

@JennyLeeHowell @j_colomb Why not teach all 3 rungs of Pearl's ladder of causal inference? Correlation is used for prediction (like predicting the weather, which nobody will deny is useful). Prediction is Rung 1. Rungs 2 and 3 deal with interventions and counterfactuals, also very useful to consider.

@LaurietobrienO I think this is great! But, also, I think we often teach correlation *only* as a means of understanding cause. Knowing that something is likely to happen, even if one can’t control it, can be very useful (e.g., for understanding risk and preemptively mitigating the consequences)

@JennyLeeHowell Then, I try to point out how longitudinal research can help to get at temporal precedence and that correlational research can control for spurious relationships.