Congrats Joy Tong for her single-presenter talk acceptance at #SPSP2026: “Intrapersonal Factors Trump Contact Frequency in Cross-Racial Face Recognition.”

Network analysis shows that quality interactions (not mere exposure) are what matter for improving other-race recognition.

Congrats Finneaz @firdausmoner.bsky.social for his accepted symposium “The Architecture of Status Perception: Economic Status Stereotype Awareness Shapes Clothing-Based Competence Perceptions” at #SPSP2026!

He’ll present “Dressing the Part,” and Ming Huang (Ben) Teo will share “Empowering Attire."

[11/11] Thanks to PhD students Anqi Mao & Yuqing Shi for their work. And to @erichehman.bsky.social and Rebecca Neel @travislim.bsky.social for organizing this novel competitive collaborative approach to advancing prejudice research. Excited to see where the future research will take this research.

[10/11] The consistency in outcome is remarkable considering a wide range of groups considered, from historically marginalized groups (Black/gay/transgender people, immigrants) to high-status groups (extremely wealthy people). This suggests a "base recipe" for prejudice, regardless of target group.

[9/11] Pretty pleased to share that MAP Team placed 3rd on bias prediction and 6th on outgroup attitudes. The models explain more than half the variance in prejudice across very different groups with similar predictors. Good work, team!

[8/n] The winning approach among our internal candidates kept it focused: just 4 key stable predictors in each model, including symbolic threat, contact quality & group identification. These focused models actually outperformed more complex versions by reducing statistical noise.

[7/n] Our final models used Random Forest for bias & PLS regression for outgroup attitudes. Using the full dataset with careful variable selection based on model performance proved more effective at finding universal predictors (e.g., rather than separating data by prejudice target group or for CV).

[6/n] To predict universal prejudice patterns, our team balanced competing goals: learning from all data vs. avoiding overfitting. We tested multiple approaches - Random Forest, Elastic Net regression, PLS regression, & Principal Component Regression.

[5/n] The competition drew 90+ researchers across 40+ teams (!). Goal was to Create models predicting prejudiced attitudes toward different social groups. Team Cluster Busters won by achieving 55%+ variance explained, big improvement over the baseline. Kudos to Team Cluster Busters.

[4/n] As in the original Prejudice 1.0 model, the competition targeted two aspects of explicit prejudice: 'Bias' (ingroup vs outgroup rating difference) and 'Outgroup Attitudes' (direct outgroup ratings). Different psychological processes might drive these two outcomes.

[3/n] The approach used explicit item-level measures of key constructs (e.g., contact quality, perceived threat, group identification) rather than pre-defined latent factors. We tested these predictors across different target groups to identify the most consistent, universal prejudice predictors.

[2/n] Why important? While psychology has produced many verbal theories of prejudice over 80+ years, they rarely made precise predictions or competed head-to-head. According to the authors (and I agree) this mathematical approach lets us make specific, numerical predictions about prejudice levels.

<>
📜 1130 Alyssa Goh
📜 1300 Ben (Huang Ming) Teo

#apcv2024
apcv2024.com

<>
🗣️ 1030 Joy Tong
📜 1300 Firdaus Moner
🗣️ 1400 DongWon Oh

#apcv2024
apcv2024.com

Kudos to Jothiga Baskaran, yet another thesis student in the lab who got the Dean’s list award 🔥🏅