I find numbers very sexy. This has led me to two behaviors with OkCupid. One -fairly innocuous- idea is that I find people with a higher match percentage more attractive. The other -kinda peculiar- urge I have is to keep a record of my match percentages with people I meet from the website.
Yeah, it’s weird, but I started doing it when I was unemployed and needed something to fill my days.
This is why we need to build a better economy, it’ll prevent guys like me from having the time to build spreadsheets about weird shit. If you think I’m bad, you should read some of the studies they do at OkTrends…
Anyway…over time, I’ve collected a sample size of 35 percentages. These are the match percentages of people I’ve met on OkC and with whom I’ve had relations. I tried to be good about recording these. Given that 35 is a
nearly complete fair size to draw conclusions about my preferences with online dating, here’s what I’ve found with OkCupid dates:
- Average Percentage Match: 89%
- Median Match: 93%
- Highest Match: 99%
- Lowest Match: 63%
What can I draw from this data? Firstly, I can see that a low percentage match will not automatically disqualify a connection. 63% is low. Like…catfish profile low. But even then, there’s hope. So if you ever see someone that seems like your type, but may not have the numbers to push you all the way to send that message, maybe send it anyway!
Outside of that, it tells me that I generally end up in the bedroom with people in or around the 90% match range.
Makes sense. When I’m browsing profiles and see a high percent match; it’s just as sexy as seeing a photo of them at the beach, a number of different languages spoken, or listing Digimon as one of their favorite shows. But this provokes a whole different series of questions about whether or not we would have slept together if we didn’t have such a high match.
Did we sleep together because we were a good match or did we sleep together because we had a high match?
OkCupid programmers actually lied to their users one time in an effort to test the persuasive ability of the algorithm. It’s a very interesting study in consumption of statistics and also involuntary experimentation if you want to read that here. The final numbers on their study were as such:
The end results of this experiment did confirm my suspicions, but not by that much. Was I being swayed by a big, strong, match percent? Maybe…Marginally, at most.
While that study yielded results based on comparing their real algorithm to a fake algorithm, I wanted a more comprehensive comparative sample. I collected the data from people that I have found on OkCupid after we’d already met somewhere else.
Maybe we met at a party, on tinder, or in line at the Planned Parenthood and had a liaisons before they showed up in my Quickmatch later on and I scribbled down our numbers. Obviously this sample size is smaller (20 entries), but as a tool for comparison, that’s a substantial value. This is a real test of the OkCupid algorithm:
How do OkCupid matches compare to real life matches?
If we met on OkCupid…
If we met outside Okcupid…
|High Match %||
|Low Match %||
These numbers barely have a deviation. It’s almost a little disturbing…
To be extra sciency, I had my most skillful statistician buddy run the two sets of data through a T-Test Calculator which yielded a p-value of .844416. After I had him explain to me what that meant, I learned that OkCupid matches are the same percent as my existing romantic life 84% of the time.
84 whole percents? That’s nothing to scoff at. OkCupid gets a B- at my dating life! Or as my aforementioned stat-nerd-bro puts it, “It means their algorithm really is quite good”.
While the OkCupid study was able to illustrate how fake their numbers can be, I think my data indicates just how real things can be. While this study is based on a limited sample size (35 Okc dates Vs. 20 IRL dates), I believe this to be a representative sample of my dating life as a whole.
The one outlier comes from the significantly lower lowest I found in my non-OkC dates. At nearly half the percentage match of my lowest OkC, my lowest matches from real life might have come from a time when I didn’t have a computer program to tell me it was a bad idea.
Higher match percentage may not always make an online date more attractive, but a very low match percentage is apparently a bigger deal for electronic affections.
The study needs more data. If you’d like to help, feel free to come over and sleep with me, then we can share match percentages and I could add them to my spreadsheet. All my grant requests have been denied for me to conduct this study, however. If you would like to take part, you’ll have to test pro-bono…
Anyway…Do you keep track of your percentages? How do you think your results vary? Have you found the OkC algorithm to be a fair decider for a good match? Post in the comments!