There are a new academic online dating study (Ben Seefeldt, University of Illinois) working on Tinder with ficticious profiles (one medium-atractive male and other medium-attractive female):http://benseefeldt.com/content/sites/Tinder/tinder_paper.pdf
This research carries out a bisexual setting, selecting both same sex and opposite-sex partners for both dummy profiles. The author was swiping right for 2020 (both sex) profiles on each account.
Results for opposite-sex matches:
Male profile: 173 female matches ( number of female matches who sent a message =6)
Female profile: 1.011 male matches (number of guys who sent a message = 429).
Anyway it is a small sample size (just one male profile and one female profile) and he's not controlling for attractiveness, which means smaller power and, for all functions, the positive predictive value for a true research finding decreases as power samples decreases. Thus, other factors being equal, research findings are more likely high in predictive value in scientific fields that undertake large studies, such as studies analyzing data sets obtained through collaboration with online dating web sites. Best online dating datasets used to be extracted of data from log files and users’ profiles, anonymized it, and compiled it in a data warehouse.
These research working with log files (Taylor et al, Kreager et al, Hitsch et al.) provides a higher ecologically valid context, but investigators use online data inefficiently, just logistic regressions for some parameters without appropriate graphics and data analisis or fails to notice statistically significant relationships).
And some dating sites as Okcupid marketed their own study on mating distributions vs attractiveness (unsolicited messages and response rates):http://blog.okcupid.com/index.php/your- ... ne-dating/
but their response rates differ significantly from those provided by other dating sites, like AYI for example:http://www.businessinsider.com/likeliho ... men-2013-7
But you know that the greater the financial and other interests and prejudices in a scientific field, the less likely the research findings are to be true. Conflicts of interest and prejudice may increase bias. Such nonfinancial conflicts may also lead to distorted reported results and interpretations.
Anyway most controlled testing reveals of male/female mating preferences, and the resultant asymmetry. But this new author should apply the results from his Tinder experiment to more general populations. On the other hand, the controls used in a new experiment should be more meaningful and strict. And the sample size should be large enough to predict any meaningful relationships between the variables being studied.
Hence I want to suggest you could make a cross-sectional study of population variation in mating success (i.e. with more human means and logistical resources). There are simple models to quantify the potential contributions to such variation in mating success (equivalent to the number of individual’s matings per user).
My suggestions for future study on online dating:
1. The impact of experimental design on estimates of the strength of mate choice preferences. Seefeldt is computing just number of individuals who perceive that 2 targets as a potential prospect for mating, but not the quality of the pool of their potential mates. He should use these two approaches if he wish estimate the accurate index of mating opportunities. Since physical attractiveness is the limiting factor, he must create multiple dummy profiles varying in attractiveness. Also some method should be designed to get ratings of attractiveness of their matches.
2. Not all matches mean real mating interest. Getting a match on a dating site is half the battle — but it isn't everything. The likelihood of any given match or first reply resulting in a reciprocated exchange and eventual date (exchanging contact info) is extremely small (see Kreager et all, where at first, 79% of men's sent messages, and 58% of women's sent messages, went unreciprocated. As number of reciprocated responses increased, the percentage of messages in each category declined, so that only 3% of men's, and 7% of women's, sent messages resulted in more than five exchanges).
There’s only a certain of spurious matches chance that a match will turn into an actual conversation — a correspondence that lasts for three exchanges or longer. For example, Hitsch et al. (2010a) consider that 6 is the mean number of exchanges required until the dyadic relationship result in an offline date.
My idea for computing the actual matches for a given user, it would require to sending messages to every initial match and collect how many users would be willing to exchange phone numbers for set a date offline