Okay, so I encountered this paper, and it fascinated me so much that I got my hands on the raw data, and decided to look more into the raw data myself using ChatGPT.

I realized the study used averages, not beta scores. When using averages, if an attractive person also smells good and is funny, those traits are all weighted equally. But in reality, it might have been attractiveness that was the true driver of interest — and the others just happened to be there.

To figure out which traits actually drove attraction, I used beta scores from regression models. These show which traits predict romantic interest when controlling for all the others. So if someone liked a person who was attractive, funny, and smelled good — but only attractiveness had a high beta — then that was the trait actually influencing the outcome.

Explanation of running a regression (only if you are curious)

When we run a regression, we take all the traits — like attractiveness, sense of humor, how good they smell — and put them into the same model. Then we ask: “Which of these traits best predicts how much someone liked the person, after accounting for the others?”

If attractiveness has a high beta score, and humor and scent have low ones, that means: Even when the person was funny or smelled nice, it was attractiveness that consistently explained why someone rated them as more romantically appealing.

The beta tells us: “If two people are equal in all other traits, the one who’s more attractive is more likely to be liked.”

Meanwhile, traits with low beta scores didn’t add much predictive power — they might have been present, but they weren’t moving the needle. That’s how we know it was attractiveness — not the other stuff — that was doing the heavy lifting.

Ranked Preferences by Gender and Their Beta Scores

Male Stated Trait Male Stated Beta Male Revealed Trait Male Revealed Beta Female Stated Trait Female Stated Beta Female Revealed Trait Female Revealed Beta
Patient 0.212173669 Attractive 0.574759969 Patient 0.113660962 Good in bed 0.601980522
Confident 0.212056899 Honest 0.565452295 Confident 0.107725158 Supportive 0.601826614
Good listener 0.131037657 Loyal 0.565404905 Good listener 0.087869533 Smells good 0.573880374
Understanding 0.12074947 Smells good 0.562928171 Honest 0.081773202 Loyal 0.566380275
Sporty 0.114067795 Supportive 0.522381891 Understanding 0.068348838 Attractive 0.564528293
Funny 0.106381052 Intelligent 0.51936867 Considerate 0.051444196 Intelligent 0.550994429
Adventurous 0.088572519 Considerate 0.518763386 Loyal 0.047010925 Fun 0.531500721
Honest 0.087511309 Good in bed 0.511471702 Sensitive 0.038461944 Honest 0.521540146
Considerate 0.084424005 Fun 0.506983123 Adventurous 0.026723917 Understanding 0.502375069
Open to new experiences, complex 0.076708431 Understanding 0.502449936 Supportive 0.023643173 Sexy 0.489527613
Ambitious 0.070906069 Nice body 0.492470414 Open to new experiences, complex 0.020731185 Funny 0.486668741
Good in bed 0.070124552 Sexy 0.480304622 Ambitious 0.000233478 Considerate 0.466217891
Financially secure 0.062355561 Sympathetic, warm 0.464677614 Sporty -0.007558407 Dependable, self-disciplined 0.418837337
Supportive 0.061663286 Calm, emotionally stable 0.457472946 Funny -0.021924471 Good listener 0.412850756
Has a good job 0.05649434 Good listener 0.454312332 Financially secure -0.022387504 Sympathetic, warm 0.404147772
Sensitive 0.054190188 Dresses well 0.451311947 Successful -0.024769891 Calm, emotionally stable 0.402537258
Intelligent 0.052889979 Sensitive 0.449284071 Fun -0.027062161 Nice body 0.38482939
Fun 0.051977077 Funny 0.439007227 Calm, emotionally stable -0.031772792 Successful 0.358151388
Dresses well 0.051858015 Extraverted, enthusiastic 0.411921714 Intelligent -0.037565277 Extraverted, enthusiastic 0.319009701
Loyal 0.04800184 Successful 0.396358645 Has a good job -0.038283445 Patient 0.318053254
Successful 0.047469296 Dependable, self-disciplined 0.351905724 Dependable, self-disciplined -0.039518715 Dresses well 0.317483453
Sexy 0.034127338 Has a good job 0.298756925 Attractive -0.041734524 Sensitive 0.291310174
Dependable, self-disciplined 0.033058184 Financially secure 0.290307017 Dresses well -0.042227562 Has a good job 0.270320926
Attractive 0.03001651 Patient 0.279352999 Religious -0.050178527 Ambitious 0.268224008
Sympathetic, warm 0.008540999 Ambitious 0.258587032 Nice body -0.054628739 Confident 0.251309235
Nice body 0.006530769 Confident 0.24715645 Sexy -0.055114198 Financially secure 0.244603972
Extraverted, enthusiastic 0.000903394 Open to new experiences, complex 0.245920988 Good in bed -0.069001698 Open to new experiences, complex 0.233535755
Calm, emotionally stable -0.001102494 Adventurous 0.231496231 Sympathetic, warm -0.070903984 Adventurous 0.190223756
Smells good -0.033591142 Religious 0.137954007 Extraverted, enthusiastic -0.07387324 Sporty 0.142753418
Religious -0.054177347 Sporty 0.123835273 Smells good -0.089107505 Religious 0.054874929

How to interpret beta scores:

  • 0.00–0.10 → Very weak effect
  • 0.10–0.30 → Modest influence
  • 0.30–0.50 → Strong predictor
  • 0.50+ → Very strong influence on attraction

Further Explanation (why stated preferences have low beta scores compared to revealed preferences)

Stated preferences have lower beta scores (max ~0.21) because everyone tends to rate all “ideal traits” highly, so there's not much variation. That makes it harder for any one trait to stand out as a strong predictor.

Revealed preferences have higher beta scores (up to ~0.6) because people vary more in how they rate actual partners. That variation lets us see which traits truly drive attraction.

Discussion: Which pill are you? Does this dataset challenge your views? When looking at the table, how would you describe what's important in dating (for each gender)?