iiinova — where your digital self keeps evolving

Piotr Koc

@pkoc.bsky.social

Postdoc at GESIS | Measurement, Public Opinion & Political Behaviour

Piotr Koc @pkoc.bsky.social · 17/09/2026

@betanalpha.bsky.social has a nice piece on how to handle the problem of empty categories in the Bayesian context: betanalpha.github.io/assets/chapt...

betanalpha.github.io

Ordinal Modeling

141

Reposted by Piotr Koc

GESIS - Leibniz-Institut für Sozialwissenschaften @gesis.org · 22/05/2025

#jobs #stellenangebote #GESISjobs #jobfairy Our Team Digital Society Observatory in #Cologne is looking for a #SeniorResearcher & #TeamLeader in Computational Social Science #CSS (Salary group 15 TV-L, working time 100%, initially for four years with possible tenure): gesis.jobs.personio....

01113

Reposted by Piotr Koc

Michael "Shapes Dude" Betancourt @betanalpha.bsky.social · 17/03/2025

5 stars is better than 4 stars, but can we even define how much better it might be? Modeling ordinal outcomes like ratings is a subtle topic; fortunately I have a new chapter that dives directly into that nuance. HTML: betanalpha.github.io/assets/chapt... PDF: betanalpha.github.io/assets/chapt...

betanalpha.github.io

Ordinal Modeling

2276

Reposted by Piotr Koc

Paul Smaldino @psmaldino.bsky.social · 09/01/2025

New paper led by my former student Peter Steiglechner. We introduced a model of dynamic opinion perception that accounts for social identity, and use data on German climate opinions to show how subject perceptions of polarization can differ from opinion polls. osf.io/preprints/so...

Illustrations of the model with fabricated opinion data. Panel A shows opinions in the two-dimensional space and the corresponding representation of the opinion space (arrows) used by individuals without identities. The ellipses indicate opinions that would be one, two, three, . . . unit distances from the center of this distribution. Panel B shows the same opinion data, but for individuals that affiliate with either a red or a blue identity group. There is a significant difference between the subjective representation of the opinion space—the lens—used by the red group, which is more homogeneous and, thus, characterized by a smaller variance, and the subjective representation used by the blue group, which exhibits a larger variance, especially along the issue represented on the y-axis. Panel C shows how the distance between two opinions, r and b, is perceived asymmetrically by two individuals with different identities. The distance between r and b perceived by the individual holding opinion r (with red identity) is twice as large as the distance perceived by the individual holding opinion b (with blue identity).

Distribution of climate-related opinions in 2021 seen from different viewpoints. The sizes of the gray circles indicate the relative frequencies of the answers. Panel A depicts these frequencies on the objective 5 × 5 grid of possible responses to the two survey questions. Our model assumes that individuals perceive opinions in a subjective representation of that space, i.e., they view the opinion landscape through a lens shaped by the opinion variance within their in-group. The lenses that individuals considered in 2021 are, of course, not known and we assume that the individuals adjust their lenses over time. Panels B and C show the same opinion distribution of 2021 (as Panel A), but viewed by an individual affiliating with the Green party and using the lens of 2016/17 Greens, i.e., based on the opinion variance exhibited by the Greens in 2016/17 (Panel B), and by the same individual using the lens of 2021 Greens, i.e., based on the opinion variance exhibited by the Greens in 2021 (Panel C). A Green using the ‘new’ 2021 lens would perceive a substantially higher level of disagreement than the same Green using the ‘old’ 2016/17 lens (compare the green arrows in panels B and C, which correspond to the perceived distances between the responses (4, 4) and (1, 2), panel A, seen with the two different lenses). This suggests a high degree of instantaneous lens-specific polarization P2∗ for the Greens.

Mean disagreement index for opinions on climate change among Germans (relative to opinions in 2016/17) obtained from objective pairwise opinion distances for all German respondents (gray dots) and from subjectively perceived opinion distances calculated using our model with either fixed or (instantaneously) adjusted lenses (blue/purple dots). The blue arrows indicate pure polarization, P1, which represents the perceived increase in disagreement between 2016/17 and 2021 and between 2016/17 and 2023, assuming lenses fixed to the opinion distribution in 2016/17. Disagreement measured in this way has increased by roughly 10% (similar to the objective polarization). The purple dots represent the disagreement perceived by the individuals when they update their subjective lenses to the current in-group opinion distributions instantaneously. This comprises pure polarization, P1 (blue arrows), and instantaneous, lens-specific polarization, P2∗ (red arrows), which either further amplifies (2021) or shrinks (2023) the polarization perceived by individuals relative to opinions in 2016/17.

Effects of dynamically adjusting lenses across the different political groups. The dots show the mean perceived disagreement for each of the political identity groups relative to 2016/17 assuming that individuals use fixed subjective lenses calibrated to in-group opinions of 2016/17 (A) and assuming that individuals instantaneously adjust those lenses to the current in-group opinions (B). For comparison, the dashed blue and purple lines mirror the perceived disagreement from Fig. 4, i.e., the (weighted) average across groups. The difference between the two scenarios, i.e., the difference between the dots in panels B and A, is the instantaneous lens-specific polarization, P2∗. Panels C and D show P2∗ as arrows for each group separately in 2021 and 2023 (see also Fig. 4).

13511

Reposted by Piotr Koc

Denis Cohen @denis-cohen.bsky.social · 20/11/2024

🚨 CfA: MZES-DVPW Conference "Methods of Political Science" 🗓️ Mannheim, Mar 27-28, 2025 ⏰ Deadline: Dec 16, 2024 🔗Apply via sosci.sowi.uni-mannheim.de/mzes-dvpw-me... @gessler.bsky.social, @aleininger.bsky.social, @hannahrajski.bsky.social, Oliver Rittmann & I look forward to your submissions!

A screenshot of the call for applications.

04637