home.social

#factoranalysis β€” Public Fediverse posts

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

  1. Possibly the worst #crime perpetuated by a #mathematics and #computerscience university degree is teaching that #statistics is boring.

    Studying #psychology taught me that #factoranalysis and #ANOVA are super neat.

    Today, I've been looking at and implemented rank-biased overlap (and distance) and #PERMANOVA (permutational ANOVA) to compare groups' (partial) rankings of preferences. The computer scientist in me is horrified, no longer at the statistics, but at the computational complexity.
  2. An adversarial collaboration reports that mathematical reflection tests assess #math ability about as much as anything else. 😬

    And what is the evidence that the paper's β€œindependent” measure of reflection actually measured reflection? 🧐

    doi.org/10.1073/pnas.240919112

    #psychometrics #psychology #cogPsych #JDM #decisionScience #economics #factorAnalysis #stats #rStats

  3. TL;DR: It's not a psychometric scale and it's not weebly

    This "Weebly racism scale" has been posted a couple of times on the Fediverse, and I finally decided to look it up. It seems like a multidimensional construct reduced to one dimension, so I wanted to see the items, hopefully an exploratory #FactorAnalysis, evidence of #reliability and (dare I hope?) #validity.

    Yeah, no. It's not a scale at all, in the psychometric sense. No data collection or #analysis (AFAICT) was part of this, so there is no (and, at this point, can be no) validity information.

    This is one person's ideas about racism. This is absolutely the kind of thought work that should be done when one is at the very beginning of scale creation, but it's not a scale like that.

    The scale is hosted on weebly dot com, a web host kind of like squarespace.

    The fact that hundreds (or thousands) of other people resonate with this scale is a good sign for potential validity. However, that is not sufficient psychometric evidence to call this a "racism scale" alongside things like the Modern Racism Scale, etc.

    This seems like a useful activity to get you thinking about your experiences with racism (on either side of that line), but because it's not a psychometric scale, there's no *scientific* reason to believe

    * The implied or explicit categories map onto actual racist thinking/behavior patterns

    * The order of the categories is valid--e.g., there's no evidence "I'm not racist but..." is more racist than "'Funny' Black Face", etc.

    * The categories even belong on the same continuum

    As I said above, it really feels like a decent start, but with several dimensions squashed into one. I'd personally love to see a racism researcher use this to develop an actual scale, or try to. I suspect the result would be something vaguely resembling this scale but with significant differences.

    #psychometrics #psychology #racism #prejudice #scale #discrimination #scaleconstruction

    *

  4. PCA focuses on maximizing variance and transforming variables into principal components, while Factor Analysis seeks to model the underlying structure and relationships among variables through latent factors.

    PCA in R programming online course starting on April 02: statisticsglobe.com/online-cou

    #DataScience #Statistics #PCA #FactorAnalysis #RStats

  5. Quote from the popular Tucker & Lewis paper on a reliability coefficient for maximum likelihood #FactorAnalysis.

    #Psychometrics #History