In my last post I began to assess mating options and dating pool in a online dating context for the couple B and 2 control profiles rated as highly attractive as a comparative sample.
We find that, consistent with predictions:
• The highly-attractive man has 93 women interested in him, almost 19 times more dating pool in a week than the male B (5 offers). Anyway we can conclude that the dating pool (number of women willing to have a relationship with him) for an medium man is almost ridiculously small.
To address our research questions about the mating size of all those individuals that were rated in the pool, I’ll start in this post by addressing several data sets obtained through a online dating called Lovoo for the male dummy profiles. So the data I use in this work derive from male profiles on this dating site in the 1-week period.
This dating site allows users to browse and search the profiles of others, and it also provides a matching mechanism. Its feature works much like quickmatch from Okcupid or winks from Match.
This is how the match feature from Lovoo works:
If people likes a user’s profile picture, they click on a heart symbol. If people don’t like the picture, they click on the X symbol. Each time a user gets a positive rating for his/her profile photo, the corresponding user will be displayed under the tab “who likes me”. People can choice and accept someone playing “match” or browsing profiles using the “search function”
The primary quantity I consider in computing mating statistics on here is not the number of messages received but the number of distinct people whom a person is selected on match feature (number of “like you”). This places the focus not on how many messages a user obtained or your reply rate to initial contacts from this person, but rather on mate choice decisions or number of user’s choice an individual on the match feature. I consider it’s a valid track to capture the dimension of its popularity or desirability, and this is main key variable in my analysis.
Using these data analysis from this research, someone could correctly modelling and measuring dating pool, choosiness and preference functions, response curves that give the probability (y axis) that a focal individual will mate with a potential mate with a given phenotype (x axis). Moreover it would be interesting to know if online daters could base decisions either on a sample of mates (sampled-based decisions) or on a threshold of comparison (threshold-based decisions).
Measurement of the size of the mating pool:
As I said, in an online dating context, the number of mating opportunities can be computed as the sum of unsolicited messages recollected quickmatch, winks or “likes you” and answered messages.
The primary quantity I consider in computing mating statistics on here is not the number of messages received or answered, but the number of distinct people whom a person is selected on match feature (number of “likes you”).
This places the focus not on how many messages a user obtained or his/her reply rate to initial contacts from this person, but rather on mate choice decisions or number of user’s choice an individual on the match feature. I consider it’s a valid track to capture the dimension of its popularity or desirability, and this is main key variable in my analysis.
Lovoo offers a sending kiss feature, which cost some credits, and a Fans/followers feature, where is displayed the number of users that have marked as “favourite” to that profile. This has little relevance to our research, but it will be shown to readers because it’s part of the image displayed on each profile.
I have thought about omitting those two variables for different reasons:
a) Received Messages
This unimodal or binary “Like” function (“yes”/”no” rating) acts as a depth and single dimension for defining mating choice decisions. Furthermore there is usually an small degree of overlap of this two different features (mainly en male daters when some of them are viewing a female profile through the browsing tool). Some of the daters choosing a profile and clicking them as “like”, then contact him/her by sending a message.
b) Reciprocated messages.
In all online dating sites there are a substantial amount of fake profiles, which can interfere somehow to explore the relation of the some data to the underlying population. By example, I figured out that it could affect reply rates decreasing likelihood of a response, mainly in those plots of sent messages directed towards top tier of attractiveness groupings, where is most concentrated the fake profiles.
The unsolicted messages, quickmatches/wink/liks and other parameters are little affected by the introduction of these fake accounts, because most of these fake profiles are virtually generated for the website itself for luring customers, and these fakes profiles dont send messages or “likes you”. They are just visual baits. But these fakes will not respond to messages sent by real users, decreasing their response rates.
Moreover, experimenting with this parameter involves systematically send a lot of messages from each dummy profile, which would take a lot of work. Furthermore, on this website the messaging function is limited, since each user can only send 5 messages per day, so for sending, for example, 100 messages, it would take 20 days.
Measurement of quality of the mating pool:
I’d need to define women’s social desirability on the basis of the evaluations of other users in the dating site (ratings). In this way I would be able to analyze the results plotted in evenly-spaced “attractiveness groups” But I found several technical problems to address this issue. Studies of online dating, building our confidence that the rated attractiveness measures provide adequate proxies of dating market value.
Unlike okcupid, where you must upgrade to A-list to acces to your quickmatch list, Lovoo shows the full list of people who cliked “like you” but it is stored as blurred portraits. It’s required sign up for a vip membership to find out to reveal the profiles list. Although the “like you” list shows the users as blurred portraits, this would not hamper my analysis, while each one of these profiles was assigned with a rating. This is a limiting statistical requirement for a proper rendering of the quality of each available mate.
I realized that I could not quantify mate desirability with profile ratings from opposite-gender daters. This website does not enable a system-generated rating tool that presents users with a series of profiles for rating them a scale of attractiveness. The former okcupid version had this tool available, but the ratings were hidden from the public.
I’m aware that Badoo, website where I did my first study, displays these attractiveness scores. But this site asks its both sex users to rank one another based on a 10-points rating system. That greatly affects the accuracy of the ratings.
Besides the inherent overall lack of control the medium seems to inextricably exert on ratings, the evidence such as this has lead me to realize that opposite-sex ratings are better than both-sex ratings.
For example, when I’ve submitted photographs of objectively highly attractive males (homemade pictures drawn from social networks, since professional photos would be identified as belonging to a fake profile), I noted that generally their ratings were much lower than expected, yielding a cumulative middle ground. Either a lot of man gives to gorgeous male pictures a low rating of punishment or sanction. While some girls vote them really high and others girls vote them into a ground slightly above-middle.
Paradoxically, my highly-handsome male dummies were perceived as a highly interesting catch for females (their inboxes were packed full of messages with unsolicited messages and replies from women diary), but their ratings was not appropriate to their aesthetic level.
I found an great obstacle to use again Badoo, it was that to sign up to the new website version , each user have to enter a phone number. This preclude to create multiple dummy accounts within this site. So I decided work with Lovoo, because it’s a free website and I could sign up there multiple dummy accounts quickly.
Anyhow I must apologize to my readers for being unable to perform the assessment of quality of these mating pools. However I have tried to give an output to this question.
I’ve decided to submit as screenshots all prortraits belongs to users stored into the “likes you” list. This will not solve the inability to perform a statistical analysis of physical attractiveness of these lists of users. Firstly, it was required to reveal blurred profiles by playing myself into the match tool, selecting positively all users that appeared in the selection set. When there is a match, the system notifies it and reveals the images of those users. Thus I could view the profiles withing the “like you” List.
However this does not mean that readers are going to be able to judge accurately these groups of photographs. If we define photographic accuracy as the degree to which an observer would consider the photograph to provide a good enough approximation of the person in it. Keep in mind that:
1) There are possible incongruities between photography and reality. Anyway we already know that.
2) Physical appearance in one only profile photo differ from their appearance perceived on an overall basis viewing a set of photos of that same person. For example, note that I always show a collage of photos set for defining each one of our studied individuals, where readers can quite accurately assess their facial and body attractiveness (for obtaining a current representation of their physical appearance). This will not happen here. Although as author of the study I was able to take a look at many of these profiles, it was not feasible to collect all the account photos, and then create photo collages. Then we will have discrepancies concerned with the physical characteristics of the person portrayed in their profile photos. In a user with facial portraits will be appreciated slightly only her facial appearance, and in user with body picture only her body figure.
3) Furthermore, profile portraits will be displayed in reduced format, for technical reasons. Note that the “like you” listings are exposed as of photos that automatically scroll vertically through a long screen background. And those dummies with a bigger size pool (high-attractiveness), contains a great number of female profiles in their matches list, making it difficult to capture in one only snapshot all those users. The solution was to decrease the screen size and assemble different screenshots.
4) Some of online users do not upload photos, but they can choose people in the match game. So some of the profile portraits are blank.
According our previous poll, we have 4 guys (A, B, D and E) sorted into the medium category (ranging from 4 to 6 on a 10 points scale):
and 2 guys (C and F) into the medium-high niche (ranging from 6 to 8 on a 10 points scale):
I will introduce into the system 2 profiles within the “high”category:
and 1 profile unattractive guy, below of medium category (< 4 points):
Low attractive male:
1) Medium males:
Profile pictures of his potential mates (1 woman clicked on the heart symbol):
Profile pictures of his potential mates (10 women clicked on the heart symbol):
Profile pictures of his potential mates (2 women clicked on the heart symbol):
2) Medium-high males:
Profile pictures of his potential mates (6 women clicked on the heart symbol):
Profile pictures of his potential mates (12 women clicked on heart symbol):
3) High-attractive males:
Profile pictures of his potential mates (137 women clicked on heart symbol):
Profile pictures of his potential mates (123 women clicked on heart symbol):
4) Low-attractiveness male:
Note: For proving (or disproving) the validity of the dating pool of each user, we would have to involve collecting of interactions of every single dater included in our sample of potential partners. So we should send messages to each one of them for scrutinizing the likelihood of reciprocated exchange (not just at point of a first reciprocated message). I guess most readers who are familiar with online dating, will know that frequently they are unable to communicate with the matches they thought they were getting.
• Each man received at least one “like you”, except the unattractive guy.
• The mean of the mating pool size for the high group is of 130 women.
• The mean of the mating pool for the medium-high group is of 9 women.
• The dating pool mean for the medium group is of 3.75. Note that male D, rated as 5.5 points (10 girls), received 71.4 % of all matches of his group. His mating pool resemble to the male F (6.93 points; 12 girls), and overcomes the mating pool of the male C (6.34 points; 6 girls). It seems that according to his mating size, its valuation was underrated on the previous survey.
• The mating size for the best looking guys is almost 15 times larger, in a week, than the mating size for the medim-high group.
If we try to explain a modal tendencies in the online population, the mating system for men seems hopelessly skewed (ie. it becomes almost a lottery in order to find a receptive female for those males who are not a 9 or 10 in terms of physical characteristics (high-attractiveness group). Since guys below of top tier of attractiveness have a narrow range of mating opportunities. Women set up a highly restrictive preferences for their ideal daters.
Although we do not know how many women have viewed every male profile in the matching tool, ie we can not establish the acetance rate. We know the number of women who have marked as potential partner to these male suitors, but not how many women have rejected them, rendering a skew which hinders assumptions of pair-matching.
Assuming that female attractiveness on Lovoo gets close to a “Normal Distribution”, where most women tends to be around a central value, at all levels of attractiveness, primarily sought out the most attractive male daters as potential partners. The modal category of sent “likes you”, regardless of the choosers’ level of attractiveness, was to the highest attractiveness male group. Although I could not run a qualitative analysis for each dating pool, taking a quick glance to their profile portraits, I’d say that male prospects have become so skewed against all but the most select cohort, that all that is left for the typical male (medium and high-medium guys) are just a few mating opportunities , ranging mainly, from medium and low-medium attractiveness , lower than their statistical equivalents. While the attractive man has no potential mate. Which makes me keep asking myself as these lucky guys have gotten such a mating success in their real lives.
The dating pool for two best looking guys is quite varied; I’d dare to say that the vast most of their girls are located within the medium and high-medium spectrum, and few of hotties. The presence of women below-medium is scarce. So this tendency to aim for the most desirable partners declined somewhat with one’s own desirability, resulting in tempered vertical preferences as one moves down the desirability scale.
In the next post, I will address the patterns of mating options for the female dummy profiles, in this same real online dating system.
TO BE CONTINUED.