The method worked better than simple aggregation for validation words such as "fifty-fifty chance", "never" or "always" (indicated in the plot by the black intervals compared to the gray interval areas).

Just submitted a minor bugfix of RRreg to CRAN:
github.com/danheck/RRre...

The R package focuses on analyzing univariate & multivariate data collected with the Randomized Response Technique (RRT), which protects anonymity in surveys by randomly scrambling responses.

I also added a new logo 🚀

of course, we also need a new badge!

New preprint by my PhD student @semihaktepe.bsky.social:

Revisiting the Effect of Discrepant Perceptual Fluency on Truth Judgments
osf.io/preprints/ps...

We could not replicate the effect that higher color contrast results in higher truth judgments in two online experiments (N=98 and N=294).

New pulication in JPSP with Isabel Thielmann, Benjamin Hilbig & @schildchristoph.bsky.social

Cheat, cheat, repeat: On the consistency of dishonest behavior in structurally comparable situations
doi.org/10.31234/osf...

--> incentivized dishonest behavior is consistent across 5 weeks & 3 years

New publication 🥳

"Personality traits (un)related to dishonest behavior"
(with Benjamin Hilbig & Isabel Thielmann)

--> HEXACO Honesty-Humility is a robust predictor of incentivized, dishonest behavior which is not the case for Big-5 Agreeableness.

journals.sagepub.com/doi/full/10....

New preprint with Maren Mayer:

Modeling Dependent Group Judgments: A Computational Model of Sequential Collaboration
osf.io/preprints/ps...

The model is consistent with previous findings and makes a novel prediction about long sequential chains of dependent judgments.

New preprint by my PhD student Semih Aktepe 🥳

Modeling the Link between the Plausibility of Statements and the Truth Effect
osf.io/preprints/ps...

- formalization of simulation-based model
- nonlinear scale transformations & interactions
- Bayesian modeling in Stan

New preprint by my PhD student @matzekloft.bsky.social:

Discriminant validity of interval response formats: Investigating the dimensional structure of interval widths

osf.io/preprints/ps...

In my lecture on statistical modeling, I illustrated the different types of true/observed/fitted/hypothesized covariance matrices that occur in estimation & testing of SEM/CFA models via hexagons and pentagons.

I like the analogy, but probabably it is not valid in all aspects.

This sounds like General Recognition Theory:

Ashby, F. G., & Perrin, N. A. (1988). Toward a unified theory of similarity and recognition. Psychological Review, 95(1), 124–150. doi.org/10.1037/0033...

This reminds me of a paper of mine about the most extreme correlation matrices of individual (per-person) effect sizes that are mathematically possible: psyarxiv.com/ca8z4/

a) All correlations are as large as possible (1-factor model)

b) All correlations are as negative as possible (weird model)

Great paper: "Impossible Hypotheses & Effect-Size Limits"
journals.sagepub.com/doi/10.1177/...

1) The correlation matrix of all population effect sizes must be positive definite.

2) Hence, there is a lower/upper limit to each effect size that is possible (even if we study only a subset of effects)

The website has changed once again....

A version control system would render changes in arguments more transparent.

Silently changing one's arguments on a website called "Innocence" to defend oneself does not make the arguments more convincing or credible...

Source: francesca-v-harvard.org/innocence