Note: This is a guest post originally posted back in 2013/2014. With the introduction of ELO in the algorithm changes, the swiping experiment can no longer be performed without adaptations to new variables. However, the findings still are insightful.
So the question of the week is: Is Tinder Racist?
I won’t tease you with this one. We already know that minorities receive less replies via OKCupid’s analysis and probably the rest of the online dating world. The short answer is “Online dating is influenced strongly by race preferences”.
*Note, some people in the comments have noted that its just preference and not racism. So what we can say is that based on this experiment, Tinder results are heavily race-influenced (to a statistically significant degree). So is online dating. So is society. But the data may surprise you as far as how much race influences really are vs. what we expect.
Luckily for us, I’m Asian and super smart so I devised a super cool way to test Tinder’s racism HAHA
For those of you that don’t know, Tinder is a fast growing app that allows users to “like” or “reject” a profile based on that person’s picture. If both users “like” each other, a match is made and you can then message the person. People praise it for weeding out the “creepy” factor because girls get to choose which guys they prefer based on their Tinder pictures. Other people complain that it is based on superficial factors like one’s photo and physical attributes.
Hypothesis: Tinder is racist*
*Fix: Tinder results are significant influenced by race (statistically significant <5% deviation)
Methodology to test hypothesis:
- I created 6 profiles of different types of people
- I ran a hacked automated tool that likes everyone in your area on Tinder
- Each profile had multiple pics and distinct tag lines, and I only included below the primary profile pics
- We recorded the number of return likes in 24 hours, and the number of messages each profile received
So, without further ado here are our contestants:
Profile #1: King Booth
King is relatively normal white guy who we consider good looking. He does’t have the coolest picture, just pics with his friends hanging out
Profile #2: Prince Caspian
You may recognize Cas from the PUA community – we needed his pic to test out the “girls like guys with muscle” theory. I have another friend who’s latino and has a lot of success with a shirtless pic on Tinder. Not all of Cas’s pics are muscle pics, only his primary photo.
Profile #3: AlphaWolf
You’ll have to believe me when I say I’m a 8.5. I’m Asian too. Instead of the muscle route, we went with the “high class” business photo:
Profile #4: Vince Lin
One of our cherished Asian model / actors, Vince Lin was supposed to represent the epitomize of Asian looks and an “Asian 10”. You may argue that his pic is less friendly and the black and white distills emotions a bit more than colored ones, so this is worth another test with a different Asian model who’s working full-time in America. We should also test half Asian models like Daniel Henney.
Profile #5: Harry Goodwins
One of the top rising male models of our time, Harry is the traditional good looking all American model with a European bone structure. We picked him to contrast out model vs. mortal data and also to compare him to Vince Lin, our Asian model counterpart.
Profile #6: Esmeralda HB9
As a control, we setup a pretty hot girl Tinder profile, to see what type of responses we would get. We wanted a solid 9 girl but not a perfect 10 because most women, if they take care of themselves can reach this level of physical attractiveness.
Yes, all search conditions being equal, white guys performed better than Asians, with the exception of the Asian model vs. normal white guy.
When we arrange the data to reflect top conversions:
1. You will notice like all online dating, girls get a disproportionate amount of attention from guys. EVERYONE wants to meet her.
2. Our male model got the highest conversion rate, but even then, it is only 10% of what the HB9 girl got! Women just have much more streaming inbox messages than men. This is an important mindset. At all levels of the game, women get bombarded with interest. If they ignore you, it is usually not because what you said was wrong, but rather, you didn’t stand out among the hundreds of guys messaging / liking her.
3. With normal guys me and Cas competing with King, King got a slightly higher percentage conversion than we did, but not as much as we expected.
Let’s arrange the data by race only:
If you look at the difference between 0.39 (me) vs. King Booth 0.52, the difference is only 0.13%. In the grand scheme of 1000+ likes, this is not that big of a deal. However, it is 34% higher than me (14 vs 13), but I had a larger like pool of 3335 vs. his 2684. Given that King’s picture is just an everyday facebook pic, and mine was properly taken with photographers and all, his profile pic still outperformed mine. Basically, at the “normal guys” level, being white gives you a slightly higher likelihood of being “liked”. So we confirmed what we already knew for a long time: Asian, black,etc, minorities have to stand out an compete more fiercely with their white counterparts to get the same results.
Also, I was surprised that Cas did not receive more likes. I’m not sure why, and another profile pic is worth a test.
At the model level, Harry got over 162% more likes than Vince Lin. 69 vs 14. This requires further testing. First, his picture was in color, and he had more pics. Vince only had 2-3 pics. Nonetheless, Harry was bombarded with messages mostly from fat girls and medium-looking white girls. The really hot white girls still did not message him initially.
1. Yes, Tinder match rates and success is at least racially influenced. Normal Asian guys need to break stereotype quickly and stand out to compete for white girls
2. Most ok-looking white girls are open to matching, the real hotties never message first, and are usually on Tinder for vanity and will occasionally go out with a hunk of their choice at their time of choosing. Most of them get overwhelmed with likes and messages on other sites, so Tinder’s app allows them to really distill based on looks
3. A normal looking white guy can outperform a muscle Asian guy or Asian guy in a suit
4. Even at guy model level, response rates were around 2.6% vs. HB9’s 20%. Girls just don’t ‘like’ everybody and get bombarded with matches and messages. The supply and demand is inverted.
5. If you are a minority, you can compete at a high level with most normal guys in society by working out and having an edge. However, anyone who thinks that everything is fair and that racism doesn’t exist is ignoring reality.
6.Overall, the percentages were not as high as I thought they would be for racism. It is more likely that girls are overwhelmed online, and also with Tinder, they are more selective about the types of guys they like. The whole demand-supply issue is more of a factor than race.
7. My other white friends have reported getting no matches and quitting, so it is possible that the picture has a huge factor to do with results.
In Scottsdale AZ, I ran Tinder auto app and got tons of likes from pretty hot girls @ about 5%. It all depends on the male-female ratio and hotness level of girls in your city.
1. Tinder seems to have some issues on their servers and sometimes a bunch of likes come in at once, it is possible our data was corrupted by some of their scaling issues. Most tests needed over a longer period of time, perhaps 1 week.
2. We can test response rate on Tinder by sending messages and see how down the girl is. I didn’t do this because it was too deceiving and I didn’t want to lead anyone on.
3. If you put a white guy in a suit vs. Asian guy in a suit, will percentages change?
4. Have a variety of male models with different “looks” emo. vs. classy vs. hunk and see which ones perform better. Have white and Asian models
5. Test Asian girls vs. black and white girls. Do they have an “Asian fetish” advantage, or does racism apply to Asian girls too?