Magnus Johansson
@pgmj.bsky.social
PhD & lic. psychologist. Research specialist at Karolinska Institutet. R package for Rasch psychometrics: pgmj.github.io/easyRasch/
#openscience, #prevention, #psychometrics, #rstats, #photo
#openscience, #prevention, #psychometrics, #rstats, #photo
More tinkering with relative measurement uncertainty proposed by @bignardi.bsky.social et al, using plausible values to estimate conditional reliability (inspired by @dmcneish.bsky.social) for Rasch models with WL estimation of latent scores. Code here: github.com/pgmj/easyRas...
November 21, 2025 at 8:29 AM
More tinkering with relative measurement uncertainty proposed by @bignardi.bsky.social et al, using plausible values to estimate conditional reliability (inspired by @dmcneish.bsky.social) for Rasch models with WL estimation of latent scores. Code here: github.com/pgmj/easyRas...
I've done some more work on the relative measurement uncertainty, comparing `brms` posterior draws to "plausible values" in Rasch models, and some other reliability metrics. Estimating RMU from draws adds some variation, as shown in the figure.
pgmj.github.io/reliability....
#rstats #psychometrics
pgmj.github.io/reliability....
#rstats #psychometrics
October 24, 2025 at 4:49 PM
I've done some more work on the relative measurement uncertainty, comparing `brms` posterior draws to "plausible values" in Rasch models, and some other reliability metrics. Estimating RMU from draws adds some variation, as shown in the figure.
pgmj.github.io/reliability....
#rstats #psychometrics
pgmj.github.io/reliability....
#rstats #psychometrics
The RelRep.R function "just works" with a dataframe with items as columns and produces pretty neat output for alpha or omega. Point estimate and CI for alpha is almost identical to RMU for my example with `eRm::raschdat1[,1:20]` data. github.com/melissagwolf...
October 10, 2025 at 9:20 PM
The RelRep.R function "just works" with a dataframe with items as columns and produces pretty neat output for alpha or omega. Point estimate and CI for alpha is almost identical to RMU for my example with `eRm::raschdat1[,1:20]` data. github.com/melissagwolf...
For instance, yesterday I read a paper with a table describing participants' sickness absence days with a mean of 71 and SD = 88. Generating a random (gaussian) sample using these values produces ~20% participants with less than zero sick days.
September 22, 2025 at 1:24 PM
For instance, yesterday I read a paper with a table describing participants' sickness absence days with a mean of 71 and SD = 88. Generating a random (gaussian) sample using these values produces ~20% participants with less than zero sick days.
Grateful to have this view a few minutes walk from home.
August 15, 2025 at 10:35 PM
Grateful to have this view a few minutes walk from home.
Inspired by a paper (linked) that added a response category to the PHQ-2 and GAD-2 screener questionnaires we did the same in Swedish with good results in a sample of 15-18 year olds. The uppermost categories performed worse, though (q25-26=GAD,27-28=PHQ). #psychometrics osf.io/preprints/ps...
August 7, 2025 at 2:32 PM
Inspired by a paper (linked) that added a response category to the PHQ-2 and GAD-2 screener questionnaires we did the same in Swedish with good results in a sample of 15-18 year olds. The uppermost categories performed worse, though (q25-26=GAD,27-28=PHQ). #psychometrics osf.io/preprints/ps...
Reading the preface to "Thinking through statistics" and this section already made me like the book.
August 4, 2025 at 9:25 PM
Reading the preface to "Thinking through statistics" and this section already made me like the book.
Might be the perfect guitar string gauge? Seems ridiculous, but .005 inches thinner than the .011 I've used forever on my solid body seems really nice.
July 29, 2025 at 8:44 PM
Might be the perfect guitar string gauge? Seems ridiculous, but .005 inches thinner than the .011 I've used forever on my solid body seems really nice.
...since it needs revising (ongoing) to incorporate correct information about critical values for GOF metrics: osf.io/preprints/os... We basically use the 4 criteria as presented by Kreiner (2007), who refers to Rosenbaum (1989).
May 22, 2025 at 2:02 PM
...since it needs revising (ongoing) to incorporate correct information about critical values for GOF metrics: osf.io/preprints/os... We basically use the 4 criteria as presented by Kreiner (2007), who refers to Rosenbaum (1989).
Really looking forward to this. It's been six years since his last book.
April 29, 2025 at 8:43 PM
Really looking forward to this. It's been six years since his last book.
Great to see conditional reliability and the "targeting" of test/sample discussed. I've incorporated a similar approach in a TIF curve figure in the `easyRasch` R package. Note that IRT/Rasch also provides information about the reliability of the test/questionnaire itself, independent of the sample.
March 12, 2025 at 4:25 PM
Great to see conditional reliability and the "targeting" of test/sample discussed. I've incorporated a similar approach in a TIF curve figure in the `easyRasch` R package. Note that IRT/Rasch also provides information about the reliability of the test/questionnaire itself, independent of the sample.
Such style and grace. Have one of those too
January 15, 2025 at 9:16 PM
Such style and grace. Have one of those too
Made with a Makina 67 a few years ago. Rollei Retro 400S overexposed a step and developed in Adox Silvermax 1+19.
#filmphotography #blackandwhite #believeinfilm
#filmphotography #blackandwhite #believeinfilm
January 11, 2025 at 7:52 AM
Made with a Makina 67 a few years ago. Rollei Retro 400S overexposed a step and developed in Adox Silvermax 1+19.
#filmphotography #blackandwhite #believeinfilm
#filmphotography #blackandwhite #believeinfilm
Sunset at 13.30 has some benefits.
December 28, 2024 at 12:46 PM
Sunset at 13.30 has some benefits.
Stunning clouds, which my old phone struggles to reproduce.
December 25, 2024 at 1:29 PM
Stunning clouds, which my old phone struggles to reproduce.
Colors and sunlight this time of year is just so nice.
September 21, 2024 at 8:28 PM
Colors and sunlight this time of year is just so nice.
The TAM author does not seem very enthusiastic about that setting though
September 4, 2024 at 8:46 PM
The TAM author does not seem very enthusiastic about that setting though
I also added multicore processing and introduced a bug. Here are proper results, table and figure.
September 4, 2024 at 10:57 AM
I also added multicore processing and introduced a bug. Here are proper results, table and figure.
I added `pairwise` and reran the same 250 datasets (9 items, 4 thresholds each, using PCM with 720 well-targeted simulated responses for each iteration). PAIR seems to perform worst. Summary table of absolute deviations from input parameters.
September 4, 2024 at 8:56 AM
I added `pairwise` and reran the same 250 datasets (9 items, 4 thresholds each, using PCM with 720 well-targeted simulated responses for each iteration). PAIR seems to perform worst. Summary table of absolute deviations from input parameters.
I messed up earlier, assuming the output of tam.threshold() to be comparable to similar functions in other packages. But useful item thresholds were found in the tam() output object and I could rerun the simulation. Less bad for TAM, but still worse than eRm and mirt. #rasch #rstats #psychometrics
September 3, 2024 at 7:18 PM
I messed up earlier, assuming the output of tam.threshold() to be comparable to similar functions in other packages. But useful item thresholds were found in the tam() output object and I could rerun the simulation. Less bad for TAM, but still worse than eRm and mirt. #rasch #rstats #psychometrics
