David Feldon

Moving on. The new service invokes a certain nostalgia.

The USU College of Education and Human Services is hiring 1 postdoctoral fellow to participate in a 2-year training funded by NSF. Successful completion of the postdoctoral training yields an automatic tenure-line faculty offer at USU. cehs.usu.edu/faculty-employ…

Utah State University is hiring 3 postdocs with interdisciplinary research agendas related to access and accessibility in STEM. Postdocs who meet performance expectations over two years will receive tenure-track assistant professorships. Learn more at careers-usu.icims.com/jobs/8708/job

@rtraborn @JasonSynaptic @pastramimachine @stgoldst As the population of graduate students becomes more diverse in every way, standardized test scores are less consistent. ETS is clear that there are persistent differences in scoring by race and SES that don’t reflect differences in ability. You might also want to read the article

@dffeldon @JasonSynaptic @pastramimachine @stgoldst Why do you think the population of students matters in this instance? (I assume you mean demographics, but please correct me if I am mistaken)

@rtraborn @JasonSynaptic @pastramimachine @stgoldst The approach to range restriction is the next generation of the same approach Kuncel used. Also, Kuncel treats all data equally across time but the population of students in the 1950s looks a lot different than today. The 2023 meta-analysis shows decrease over time.

@JasonSynaptic @pastramimachine @stgoldst Is this meta-analysis based on a bunch of restricted range studies? If so, the review is close to useless.

@rtraborn @JasonSynaptic @pastramimachine @stgoldst Interestingly, all of those are from the same corpus of data and core analyses, so they are not actually replication.

@JasonSynaptic @pastramimachine @stgoldst Thankfully I can continue to do my job serving this country, because studies have already been done! Here’s a good place to start: science.org/doi/10.1126/sc…

@rtraborn @JasonSynaptic @pastramimachine @stgoldst The problem of restricted range is addressed in multiple ways in the main article and the addendum. The larger studies in the sample also represent nearly the full range of possible scores.

JUST OUT! A fusion of learning sciences, queer theory, and mixed methods, as well as a labor of great love with Colby Tofel-Grehl, Andrea Hawkman, @FTMathteacher, and others! tandfonline.com/doi/full/10.10…

Bayes’ theorem, clearly explained:

Banger of a paper

🚨CALL FOR PROPOSALS! 🚨 Super excited about co-editing a special issue on effective feedback to teachers for Theory into Practice w/ Mary Lynne Derrington! Share widely and submit your proposals here by Dec. 1st: osu.az1.qualtrics.com/jfe/form/SV_8B…

✨WE'RE HIRING✨We have 2 TT appointments at the rank of assistant or associate professor! Apply⬇️ Research specialty related to gender and migration: usccareers.usc.edu/job/los-angele… Research specialty related to political sociology, culture, and social theory: usccareers.usc.edu/job/los-angele…

Register to attend the @NASEM webinar on October 10 at 5 PM ET featuring Dr. Shuchi Grover discussing K-12 Education in the Age of AI: events.nationalacademies.org/43705_10-2024_… #EdResearch #AIEd #AI #NSFSTEM

🚨Review of Edu-Research in 2023/24 → 11 standout papers → 5 emerging themes → My finding/sharing process (this is my first crack at a vid—would love any feedback, esp how it could be better next time)

Measurement Malpractice: @lrj417 reports how Oklahoma lowered standards but still reported trends, leading to nonsensical gains. (Researchers, think effect sizes approx. +.11 to +.63). Lesson: OK, be like other states and listen to your TAC (many @NCME38 members serve on TACs).

Nice study! In our work (eg. psycnet.apa.org/doi/10.1037/ed…) we find across replications that time and cognitive load type matter. Self-efficacy is harmed by higher levels of extraneous load within time and enhanced by intrinsic load across time. SE may differ from general affect?

@DegenRolf Nice study! In our work (eg. psycnet.apa.org/doi/10.1037/ed…) we find across replications that time and cognitive load type matter. Self-efficacy is harmed by higher levels of extraneous load within time and enhanced by intrinsic load across time. SE may differ from general affect?

Meta-analysis: Thinking hard feels bad. apa.org/pubs/journals/… This meta-analysis provides a new, central piece of evidence for models that assume that mental effort is costly. The main finding from this meta-analysis is that mental effort is strongly associated with negative affect. We found this association in all types of tasks that we studied, including tasks that have motivating features (e.g., tasks in which people have autonomy, tasks in which people receive feedback, and tasks in which performance has real world consequences). Furthermore, we found this association in all types of populations that we studied, including populations in which mental effort likely had been rewarded in the past (e.g., experienced professionals, university-educated people). Thus, our findings showed that when a sample of participants felt more effort on average, that sample also tended to feel more negative affect on average. Finally, it is important to note that people may justify their effort expenditure after the fact. That is, when people do something unpleasant to attain some goal (e.g., exert effort, endure pain, undergo humiliation), they may later infer that that goal must have been very valuable to them. After all, why else would they have carried out the aversive action? This notion of effort justification [is related to] the so-called Ikea effect.

“All pedagogies I don’t like are evil because they are always applied in the way in which I mischaracterize them.” —EduTwitter

@KathiFisler Krippendorff’s alpha seems to be the current favorite. Fleiss’s kappa is also solid). However, the combinatorial explosion isnt an issue bc agreement should be calculated separately for each code. That said, modifiers can be subsumed within a category as values of the variable

We're doing qualitative coding. For each top-level code in a category, we also select multiple modifier tags (think "pos/neg sentiment", "examples yes/no", etc.). The end result is exponential blowup of code combinations. What is an appropriate measure of inter-coder reliability?

@guzdial Are you presenting today? It would be great to see you!

So fun to see so many old friends at #CSTA2024