home.social

#rasch — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #rasch, aggregated by home.social.

  1. A recent tutorial paper on Bayesian #ItemResponseTheory with brms called for the creation of an R package make it easier for researchers to run this type of analysis. While my package is focused on #Rasch models, I hope it contributes towards this end doi.org/10.1016/j.jsp.2024.101

  2. A post by Julia Rohrer on 🟦 had me revisit one of my earlier studies in which we used Mixed #Rasch Models to identify response styles in a personality questionnaire and then investigate differential item functioning across gender* groups.
    econtent.hogrefe.com/doi/10.10

    Overall, the DIF results were consistent and gender-DIF and response styles likely independently influenced item responses.

    #Psychometrics

    *Published 2013, i.e. we would probably approach that part with a bit more nuance today.

  3. Right now, a #Rasch's world premiere and #Beethoven's Ninth with Mesak, d'Angelo, Sokolowski and Fischesser from #Berlin buff.ly/4gQmsX2 #wch

  4. In 20 minutes, a #Rasch's world premiere and #Beethoven's Ninth with Mesak, d'Angelo, Sokolowski and Fischesser from #Berlin buff.ly/4gQmsX2 #wch

  5. Today, a #Rasch's world premiere and #Beethoven's Ninth with Mesak, d'Angelo, Sokolowski and Fischesser from #Berlin buff.ly/4gQmsX2 #wch

  6. Rule of thumb cutoff values for various model fit metrics are ubiquitous in #psychometrics, despite evidence of the pitfalls with using them. I’ve added functions to my #rstats package to easily get correct simulation based cutoffs for some #Rasch model aspects. A brief example: pgmj.github.io/simcutoffs.html

  7. Moin ☕ Dieser Artikel aus dem Ruhrpott motiviert vielleicht ein bisschen. Zu einer Suche in #Bremen ganz oben: Seit 2 Jahren hier im Norden im Bau, der Radschnellweg #RASCH von Verden her.

    rasch-achim.de/ #verkehrswende #fahrrad

  8. I will be giving a pre-conference workshop on #Rasch analysis using R and Quarto, focused on the package I developed, RISEkbmRasch, at the 1st Scandinavian Applied #Measurement Conference on the 12th of June.

    hkr.se/en/collaboration/samc20

    pgmj.github.io/raschrvignette/

    #psychometrics

  9. New paper out on the Perceived Stress Scale! The PSS is a very commonly (mis)used measure, and our analysis clearly shows the need for changing how it is used.
    doi.org/10.1186/s12888-023-051

    #openaccess #openscience #opendata #psychometrics #rasch #measurement #stress

    First, only the negative items actually assess stress and work together as a unidimensional scale. This means a shorter scale, 7 items from the PSS-14 (6 items from PSS-10).

    1/2

  10. @jrboehnke Yes, these are great. I make use of the catR::thetaEst() function in my #Rasch #rstats package.

    I currently do a lot of work with public health (youth/school) surveys and really recognize the disconnect you mention. The norm seems to be dichotomization of single items and arbitrary "index" creation based on multiple items, scaled to 0-100.

    I've documented (in Swedish, unfortunately) extensive psychometrics for a school survey (with R-code but no data): pgmj.github.io/sthlmsenk/

  11. I've made some updates to my #rstats package for #Rasch #psychometrics. For one, you can get this figure, illustrating the relationship between ordinal sum scores and logit interval scores (with measurement uncertainties). Logit scores are now estimated using WL to minimize bias. More info in the vignette: pgmj.github.io/raschrvignette/

  12. Last week "Quality of Life Research" #ISOQOL published 6 papers:
    link.springer.com/journal/1113

    For example,
    the use of #EyeTracking technology to investigate use of the #EQ5D
    link.springer.com/article/10.1

    the Norwegian Fatigue Characteristics and Interference Measure for #stroke survivors
    link.springer.com/article/10.1 #Rasch #Psychometrics

    and social factors of health-related quality of life in older adults
    link.springer.com/article/10.1

    #HRQL

  13. purrr -> furrr is such a wonderful thing for parallel processing in #rstats. I’m running #Rasch #psychometric models for 52500 simulated datasets, comparing packages/estimators, and the speed-up is very nice when going from single core to 8 parallel cores. And so simple to implement!

  14. New preprint: Our psychometric analysis of the Swedish version of the Perceived Stress Scale shows significant issues with dimensionality (for all versions), which affects how the scale should be used. doi.org/10.21203/rs.3.rs-26992

    We have also made the dataset, analysis code (in R) and full analysis documentation freely available: doi.org/10.17605/osf.io/txvw2

    #openscience #opendata #research #stress #psychometrics #rasch

  15. I've added a function to estimate person scores based on Rasch analysis item parameters, utilizing the thetaEst function from the catR package. Really impressed with how easily furrr makes parallel processing available! #rstats #psychometrics #rasch

    pgmj.github.io/raschrvignette/

  16. Neues zum #Radschnellweg #Rasch in Achim: teurer, schmaler, dunkler … der übliche Projektablauf zeichnet sich ab - numbers go 📈 features go 📉.
    is.gd/eABfJT

    Kann man sich mit ein paar Grundstücksbesitzer*innen nicht einigen, muss der Rasch schmaler werden als vorgeschrieben. Für max. 10% des Projektes aber erlaubt.

    Auch nicht so toll: im Industriegebiet wird nur der Rasch gebaut, Fußgehende sollen den alten, längeren Fußweg nehmen. Sollen. :blobcatthink:

  17. New preprint on #psychometrics criteria for "valid & reliable" measures, and reporting guidelines doi.org/10.31219/osf.io/3htzc

    Also, we present an #rstats package to simplify psychometric analysis with #Rasch and combine analysis code and documentation with #quartopub pgmj.github.io/raschrvignette/

  18. @JoranJongerling Better late than never, here is a paper that may be interesting on the topic of #Rasch #psychometrics and item hierarchy: ncbi.nlm.nih.gov/pmc/articles/

    Boone, W. J. (2016). Rasch Analysis for Instrument Development: Why, When, and How? CBE Life Sciences Education, 15(4). doi.org/10.1187/cbe.16-04-0148

  19. I've finally managed to put together a sort of vignette for the #rstats package for conducting #Rasch #psychometric analysis documented with #quartopub : pgmj.github.io/raschrvignette/

    Link to the package (not available on CRAN): github.com/pgmj/RISEkbmRasch

  20. One of the most underused ways to work with validity in #psychometrics is item hierarchy. Researchers should be able to hypothesize (roughly) how item locations will be arranged. This could even be pre-registered. Here is an item hierarchy figure made with a function from the #rstats #rasch package I’m working on (for #quartopub use).

    Also, note the disordered item thresholds for q9, and distances between T2 and T3 for many items. Response categories will be a topic for another Friday! :)

  21. I've made an #rstats package to simplify #Rasch #psychometrics with #quartopub reporting. This is the "targeting" figure (output from a single function), a sort of Wright Map that shows person and item threshold locations on the same logit scale.

    The top part shows person location histogram, the middle part a histogram of item threshold locations, and the bottom individual item threshold locations. The histograms also show means and standard deviations.

  22. @lakens I’m wrapping up a paper where we propose new reporting guidelines for #psychometric papers that clarify basic criteria for measurement. I’ve developed an #Rstats package that vastly simplifies conducting #Rasch analysis with #quartopub that follows our reporting guidelines. We also include a brief checklist for intervention scientists to assess psychometric quality of measures. Hopefully this is useful to improve quality of measurement. Preprint available in December.

  23. 1973;

    #Exponentiell​es #Wachstum: In der #soziale​n #Wirklichkeit muss eines Tages zu #Landung angesetzt werden. #Belastung durch #Erfahrung, #vorbeugend​es #Handeln führt noch #rasch​er zu #neu​en #Problem​en, als #wissenschaft​liche und #politisch​e #Potenzial​e #anwachsen, mit ihnen #fertig zu werden.

    #Evolution #Lernen #Lerntheorie

    Hermann #Lübbe - Der #lang​e #Marsch durch die #Bildungsinstitution​en; #Ursache​n und #Ziel​e #aktuell​er #ideologisch​er #Aggression, S. 15