Secrets to Maximizing Success in the (Speed) Dating Experience

By Aaron Lee, Aden Chiu, Charlotte Huang, Grace Panos, Michelle Lee, and Teresa Rexin

Image for post
Image for post
Source: http://clipart-library.com/heart-png.html

Introduction

Now more than ever, the prospect of instant gratification has taken a hold of the dating world. Dating apps like Tinder and Bumble have capitalized on this phenomenon by allowing users to select potential matches based on appearance and single-sentence bios. Speed dating events have been going on for decades and are based on the idea that compatibility is pre-determined by certain attributes. Our group decided to analyze a speed dating dataset from Kaggle in order to make a conclusion about what factors generate attraction and influence “matches” between pairs of people. We hope you “super like” our interpretations!

Top Secret(s)

From our analysis, we were able to make the following conclusions about success in the (speed) dating experience:

  1. Race doesn’t matter much in finding a match but age and career do.
  2. Males and females look for different qualities in their partners.
  3. Participants that perceive themselves as more attractive, fun, and ambitious tend to get more matches, possibly due to self-confidence in their own attributes.
  4. First and last impressions are important.
  5. Keep an open mind when going into any dating experience and have fun!

Speed Data-ing

This Speed Dating dataset was compiled by Columbia Business School professors Ray Fisman and Sheena Iyengar, originally used for their paper “Gender Differences in Mate Selection: Evidence From a Speed Dating Experiment.” Over 500 people ages 18–55 participated in the experimental speed dating events from 2002–2004. During the events, attendees had a four minute “first date” with participants of the opposite sex. After each date, participants rated their date on six attributes: Attractiveness, Sincerity, Intelligence, Fun, Ambition, and Shared Interests and whether they expected to match with their partner. The dataset also includes other questionnaire data about the participants including: demographics, dating habits, self-perception across key attributes, beliefs on what others find valuable in a mate, and lifestyle information.

It’s important to acknowledge that the data is from 2002–2004 — nearly twenty years ago. These findings may have been reflective of different cultural values and preferences at the time. When looking at the distribution of ages, it is evident that most participants were under thirty years of age. Therefore, our findings are more applicable to younger people, and we can not necessarily consider them reflective of people of all ages looking for a partner.

Many of the variables in the dataset did not have an intuitive naming structure, so we examined the Speed Dating Data Key provided with the dataset and read excerpts from the paper to understand the methodology of the data collection. Then, we determined which factors we wanted to examine (such as race, career, and various ranking of attributes) and removed redundant columns in the dataset (such as “field” which measures the same information as “career”). Due to the nature of the experiment, some of the recorded attributes had rows of missing data since some attributes were only recorded for certain groups of people. Thus, we removed these columns with sparse data.

Another aspect of data cleaning was normalizing the attribute rankings. Some groups were told to rank the attributes out of 10 independent of one another, while other groups were given a total of 100 points to be distributed among the five attributes. We decided to normalize the data by converting the scores out of 10 to be out of 100. We accomplished this by summing the attribute scores for each participant, finding the proportion of each of the attributes, then multiplying this by 100.

(S)he’s All That

First, we investigated general characteristics including race, age, and career to determine what matters to participants in a speed dating experience.

We found that race is not a defining factor in terms of speed dating choices. Based on the data, the match rate for participants whose partners were the same race was 17.07%, and those of different races was 16.08% — this is only a 1% difference. But of the five race categories studied: African-American, Caucasian, Latino, and Asian (everyone else listed as other), there were differences in races getting a match.

Image for post
Image for post

African-Americans had the highest match percentage with a 20% chance of matching. Asians had the lowest match percentage with only 13%. However, more information may be required to see whether these results apply to long-term relationships as opposed to speed-dating.

In addition to race, we looked into another demographic factor: age. The total number of matches for both females and males follow an approximately normal curve. The total number of matches for females was highest at age 27, while total matches were highest for males at age 25. We compiled this with the match rate (proportion of matches a participant would get) among each gender and age.

Image for post
Image for post
Image for post
Image for post

For both females and males, the match rate was generally higher among younger participants. It is interesting to note that although the total match count for each gender follows similar distributions, there are nonetheless significantly more matches for males over the age of 30 compared to females. This supports the long-standing idea that younger women prefer to date older men, citing factors such as maturity, wisdom, and financial stability.

Furthermore, males and females of different age ranges answered the survey question “What is your primary goal for participating in this event?” in the dataset. Our graphs below illustrate that as age increases, more males desire a serious relationship or to get a (second) date. Conversely, most females older than 35 years old still were interested in having a fun night out and meeting some new people.

Image for post
Image for post
Image for post
Image for post

Another factor we investigated was career. In the barchart below, the data was grouped into 17 career categories and the match rate is represented by the height of the bars. The color gradient is the overall proportion of matches normalized by the population of each career, since careers with more people will tend to have more matches by raw count.

Image for post
Image for post

Although athletes have the highest match rate at 30.00%, they actually make up a small proportion of total matches as indicated by the light green coloring. In comparison, people in business and academic/research related careers had match rates of 17.97% and 15.52% respectively, but they make up a larger proportion of the overall matches since there are more people in their careers as indicated by the dark green coloring. This illustrates that people working in the careers shaded in darker green may be more desirable than others.

Aside from the overall population of these careers, we can also examine how the ages of people in these careers impact their match rate.

Image for post
Image for post

As we can see in the scatterplot above, the careers with high match rates tended to have very low median ages, and those who were matched tended to have lower ages in general as discussed in the previous section. Thus, age may have been a contributing factor in certain careers having higher match rates than others.

♫That’s What I Like About You (na na nanana)♫

We also looked into how attributes affect a person’s likelihood of getting matches. In the dataset, every participant was heterosexual. The two donut graphs below depict what attributes men and women are initially attracted to.

Image for post
Image for post

The pink graph illustrates that most men notice attractiveness first in their female counterparts. In contrast, females regard the five attributes more evenly, with intelligence being highest at 24%, shown in the blue graph.

We also investigated how participants who expected to have a high match proportion rated themselves across the attributes.

Image for post
Image for post

The radar graph above shows that people who expected to get a high match rate (expecting to match with over 50% of dates) rated themselves as less sincere, but more attractive, fun, and ambitious. Those people probably do not take speed dating as seriously as others and are looking for chances to make friends instead of a serious relationship. Looking closer in terms of just attractiveness, we plotted the graph below.

Image for post
Image for post

This dot plot displays how many matches participants of different levels of attractiveness ended up having. The data shows that people who rated themselves as more attractive received a higher proportion of matches. Most participants rated themselves as 7 or 8, so there is less data for people that rated themselves below a 6. This may explain the trend of less attractive people receiving a lower match rate.

The Day of Your Date

Finally, we analyzed some factors that may come into play during your actual (speed) date, such as your expectations of receiving matches. The results suggest that participants should come in with reasonable expectations.

Image for post
Image for post

This visual represents the difference in how many matches a person expects vs. how many matches a person actually ends up with. A negative value means that the number of matches a participant received was below their expectations, and a positive value means that the number of matches a participant received exceeded their expectations. The average difference is -0.05327, meaning in general participants would get slightly less matches than they expected.

We also looked at how the timing of your speed date influenced your probability of receiving a match. The bar graph represents different match rates when participants meet with each other at different times.

Image for post
Image for post

It is most effective to be one of a person’s first or last dates.This result corresponds to a psychological phenomenon: primacy and recency effect that people tend to remember the first (primacy) and last (recency) items in a series. This can be applied in our scenario. People will have a clearer memory of the first several partners because they can pay more attention to them and have longer time to digest the information, which has a higher probability to be transferred to Long-Term Memory. People also can remember the last few partners because these information are still in people’s Short-Term Memory. Therefore, if you would like to make a memorable impression on your crush, try to be his or her first or last date!

Conclusion

From our analysis of the data, we conclude that participants should enter the dating scene with an open mind to find a friend, not a partner. Never try to change yourself, but keep in mind the important qualities each gender is looking for and be sure to highlight those qualities about yourself. Try to stay calm and confident during the date. Race is not an important factor in finding a match, but you may be at a disadvantage due to your age or career choice. Make sure you remember that there is no “algorithm” to finding a perfect match; these tips are just meant to improve your chances of getting someone’s attention and securing a second date!

Overall, we believe that compatibility plays a large role, not just in long-term relationships, but also in speed-dating. Understanding a potential partner’s priorities on personal attributes will play a larger role than most other factors. Most importantly, first impressions do matter, so make sure yours counts!

If you are interested in seeing our code, please visit our Github here!

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store