Data Dictionary: Speed Dating Experiment

Fisman & Iyengar, Columbia University

Overview

This dataset comes from a speed dating experiment conducted at Columbia University by Ray Fisman and Sheena Iyengar. Participants attended a series of 4-minute speed dates. After each date, both participants rated each other on several dimensions and indicated whether they would like to see the other person again.

Each row is one speed date — a directed observation of person i rating person j.

Variable Groups

Demographics

Variable	Type	Range	Description
gender	Binary	0/1	Gender of self (0 = female, 1 = male)
age	Numeric	18–55	Age of self
age_o	Numeric	18–55	Age of partner
d_age	Numeric	continuous	Difference in age (self minus partner)
race	Factor	1–6	Race of self
race_o	Factor	1–6	Race of partner
samerace	Binary	0/1	Whether both persons share the same race
field	String	—	Field of study

Preferences & Values

These variables capture what participants say they value in a partner, and how important same-race/same-religion matching is to them.

Variable	Type	Range	Description
importance_same_race	Numeric	1–10	How important is it that partner is of same race?
importance_same_religion	Numeric	1–10	How important is it that partner has same religion?
attractive_important	Numeric	1–10	What you look for in a partner — attractiveness
sincere_important	Numeric	1–10	What you look for in a partner — sincerity
intellicence_important	Numeric	1–10	What you look for in a partner — intelligence
funny_important	Numeric	1–10	What you look for in a partner — being funny
ambtition_important	Numeric	1–10	What you look for in a partner — ambition
shared_interests_important	Numeric	1–10	What you look for in a partner — shared interests

Partner’s Stated Preferences (about the partner)

How the partner weights each attribute when evaluating potential matches.

Variable	Type	Range	Description
pref_o_attractive	Numeric	1–10	How important partner rates attractiveness
pref_o_sinsere	Numeric	1–10	How important partner rates sincerity
pref_o_intelligence	Numeric	1–10	How important partner rates intelligence
pref_o_funny	Numeric	1–10	How important partner rates being funny
pref_o_ambitious	Numeric	1–10	How important partner rates ambition
pref_o_shared_interests	Numeric	1–10	How important partner rates shared interests

Ratings by Partner (how partner rated me)

These are the ratings the partner gave to self on the night of the event.

Variable	Type	Range	Description
attractive_o	Numeric	1–10	Partner's rating of self — attractiveness
sincere_o	Numeric	1–10	Partner's rating of self — sincerity
intelligence_o	Numeric	1–10	Partner's rating of self — intelligence
funny_o	Numeric	1–10	Partner's rating of self — being funny
ambitous_o	Numeric	1–10	Partner's rating of self — ambition
shared_interests_o	Numeric	1–10	Partner's rating of self — shared interests

Self-Ratings

How participants rated themselves on each attribute.

Variable	Type	Range	Description
attractive	Numeric	1–10	Self-rating — attractiveness
sincere	Numeric	1–10	Self-rating — sincerity
intelligence	Numeric	1–10	Self-rating — intelligence
funny	Numeric	1–10	Self-rating — being funny
ambition	Numeric	1–10	Self-rating — ambition

Ratings of Partner (how self rated the partner)

How the participant rated their partner on the night of the event.

Variable	Type	Range	Description
attractive_partner	Numeric	1–10	Your rating of partner — attractiveness
sincere_partner	Numeric	1–10	Your rating of partner — sincerity
intelligence_partner	Numeric	1–10	Your rating of partner — intelligence
funny_partner	Numeric	1–10	Your rating of partner — being funny
ambition_partner	Numeric	1–10	Your rating of partner — ambition
shared_interests_partner	Numeric	1–10	Your rating of partner — shared interests

Personal Interests

Self-reported interest levels in various activities, rated 1–10.

Variable	Type	Range	Description
sports	Numeric	1–10	Interest in sports
tvsports	Numeric	1–10	Interest in watching sports on TV
exercise	Numeric	1–10	Interest in exercise
dining	Numeric	1–10	Interest in dining out
museums	Numeric	1–10	Interest in museums
art	Numeric	1–10	Interest in art
hiking	Numeric	1–10	Interest in hiking
gaming	Numeric	1–10	Interest in gaming
clubbing	Numeric	1–10	Interest in clubbing
reading	Numeric	1–10	Interest in reading
tv	Numeric	1–10	Interest in watching TV
theater	Numeric	1–10	Interest in theater
movies	Numeric	1–10	Interest in movies
concerts	Numeric	1–10	Interest in concerts
music	Numeric	1–10	Interest in music
shopping	Numeric	1–10	Interest in shopping
yoga	Numeric	1–10	Interest in yoga

Expectations & Compatibility

Variable	Type	Range	Description
interests_correlate	Numeric	−1 to 1	Correlation between self's and partner's interest ratings
expected_happy_with_sd_people	Numeric	1–10	Expected happiness with people met at the event
expected_num_interested_in_me	Numeric	0–20	Expected number of people who will want to date you
expected_num_matches	Numeric	0–20	Expected number of mutual matches

Outcomes

These are the key dependent variables of interest.

Variable	Type	Range	Description
like	Numeric	1–10	Overall liking of partner ⟵ primary continuous Y variable
guess_prob_liked	Numeric	1–10	How likely you think your partner likes you
met	Binary	0/1	Whether you had met your partner before the event
decision	Binary	0/1	Your decision to see partner again (1 = yes)
decision_o	Binary	0/1	Partner's decision to see you again (1 = yes)
match	Binary	0/1	Mutual match — both said yes ⟵ primary binary Y variable

Notes

Unit of observation: one directed speed date (person i rating person j). The same two people appear twice in the data — once from each perspective.
Variable naming convention: suffix _o indicates the variable is from the partner’s perspective; suffix _partner indicates the participant’s rating of their partner.
Recommended Y for OLS: like (continuous, 1–10). Use match for a logit/probit or Linear Probability Model demonstration.
Source: Fisman, R., Iyengar, S. S., Kamenica, E., & Simonson, I. (2006). Gender Differences in Mate Selection: Evidence from a Speed Dating Experiment. Quarterly Journal of Economics.