Plenty Of Fish Experiment: Study 1


Badoo Study

In this first test, conducted on the dating site Badoo, I published the results of a subset of dummy profiles. I begin by reviewing contact and reciprocation parameters. I tested what is the relationship between the attractiveness of initiators and recipients of the initial messages and reply behaviors. How the reply probability of a message correlates with attractiveness of the sender and receiver. And how the reply probability depends on the extent to which the sender’s physical attributes match the receiver’s stated preferences. These variables allowed me to measure the mating skew that quantifies the degree of unequal partitioning of mating output among individuals, the female mating biases that generate higher mating success for a subset of males. And I could partially quantify the set of mating options.

Okcupid Study

In this quick second experiment conducted on the dating site OKCupid, my purpose was to test a new study to assess the mating pool of some dummy profiles estimated through compilation of incoming messages and potential matches’ offers. This matching feature works as a unimodal or binary “like” function (“yes”/”no” rating) and it’s a depth and single dimension for defining mating choice decisions.

Lovoo Study Part 1 / Lovoo Study. Part2

In this third study, my task was to provide a broader topography using a wide-ranging variety of dummy profiles. I have omitted the compilation of incoming messages and I focused on the quantification of the dating pool just through the matching system feature. In this experiment, I’ve inserted a wide range of profiles on the dating site Lovoo.


It is important to understand users’ mating preferences. The message sending and replying actions of a user are strong indicators for what an individual is looking for in a potential date and reflect the user’s actual dating preferences.

In this serie of new studies I want to find out again how users’ online dating behaviors correlate with physical attributes using dummies within a other online dating site: Plenty of Fish. Since the dating sites to which I have consulted don’t want to provide me their data, such a log files and users’ profiles, and compiled it in a data warehouse, I’d need again work with fictitious profiles within a dating system. Athough I’m aware that a small sample size means smaller power and the positive predictive value for a 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 with big data, such as studies analyzing data sets obtained through collaboration with online dating web sites.

Now I’m searching for mate acceptance or response curves that give the probability that different male individuals (with a given phenotype within a attractiveness scale) will be accepted as mate by different potential female mates (along the attractiveness scale). When a female subject classifies a male prospective mate as acceptable (i.e above or below her thresholds of acceptance)?.

I am trying to find out which phenotypic stimulus of attractiveness that evokes a response exceeding this threshold. It is surprising, however, the paucity of theoretical and empirical work in humans (though there are a variety of zoological studies) addressing these probability densities functions for each phenotypic stimulus (i.e. attractiveness) that evokes a response exceeding the mate threshold by each male/female.


I will collect and analyze the same variables studied in my first experient on Badoo. The dummy profiles will be introduced on the same mating system, the dating site “Plenty of Fish”, over the same 1-week period. They were placed into the same geographical place, a big-sized city.

I wanted place my focus mainly in response rates, since last studies on OkCupid and Lovoo I could not research this parameter. Anyway I am going to gather several types of information about messaging statistics:


Messages received: Number of messages received per time active on site. (7-days).

Contacts initiated:  Number of individuals I’ve contacted through each dummy profile. I chose a sample of 150 users for this first experiment.

These messaged users are not randomly chosen. I chose a filtered sample of 150 female users for each dummy through advanced search. They were to meet the following parameters:

Age range: 18-35 years old.

City or Postal Code / Miles: 100 miles

Body type: “thin”, “athletic” and “average”. I discarded profiles self-identified as “a few extra pounds, “big & tall / bbw” and “prefer not to say”.

Within the subset of browsed female profiles, I chose as target those profiles that contained pictures from women that could be included within the upper half of facial attractiveness spectrum. This is, from average or medium-attractiveness until highly-attractiveness.

Contacts received: Number of people who contacted this person per time active on site.

Reply proportion at first message: Proportion of initial contacts from this person to which others replied. (reply rate).

Localitation: big-sized city A.

Data-collection period: 7- days.

Dummy profiles:

This first study is composed of 4 male profiles, 2 of them highly attractive (males Z and Y), 1 moderately attractive (male F, r= 6.93) and 1 medium attractive or average attractive (male B, r=5.18).

Recall that to ease attractiveness comparisons, I’ve sorted the ratings into five equal categories of attractiveness (highly-attractive: 8-10, moderately-attractive:6-8, medium-attractive:4-6, and below medium-attractive(< 4).

I also wanted to test if there could be some slight influence of occupational and educational level in the choosing pattern. To rule out unobserved factors correlated with professional occupation as the basis of attraction, I assigned higher occupational and educational levels to less physically attractive profiles. The highly attractive profiles were assigned with jobs of less professional category, while the rest of the profiles were configured with white collar professions (higher education level required and higher paying jobs.).

Some scholars might expect that it was expected that women would find that men of moderate and average physical attractiveness but with more resources, finances and education, such as finances and education, more attractive than men with more physical attractiveness and less financial resources. Or at least make up the shortfall in looks and get a similar dating success than the most appealing males.

Male Z:  Fitness/Kickboxing Instructor; Male Y: Nightclub Promoter. Male F: Economist; Male B: Doctor Ophthalmologist.


a) High-attractiveness dummy profiles (>8 points on a 1-10 scale):

  • Male Z :
Highly attractive male Z

Figure 1. Photos of male Z.

Male Z profile

Figure 2. Headboard of Male Z Profile


  • Male Y:
Highly attractive male Y

Figure 3. Photos of male Y.

Male Y Profile

Figure 4. Headboard of Male Y Profile


b) Moderately-high attractiveness dummy profile:

  • Male F:
Male F photos

Figure 5. Photos of male F.

Male F Profile

Figure 6. Headboard of Male F Profile

c) Medium-attractiveness dummy profile:

Male B photos

Figure 7. Photos of male B.

Male B Profile

Figure 8. Headboard of Male B Profile.




a) High-attractiveness dummy profiles (>8 points on a 1-10 scale):

  • Male Z :

Figure 9. Screenshot displaying Mail Inbox after the 7-days period; “Meet me” section: Number of people who wants to meet this user; Number of visitors.



Figure 10. Screenshot displaying Contact History: Contacts Received.

Sin título

Figure 11. Screenshot displaying Contact History: Contacts Initiated. The number of people contacted for all dummies were 150. Note that the screenshot after the 7 day period shows 147 users, because 3 of them have removed their accounts.


Mail inbox = 122

Meet me = 52

Contacts received = 36

Reply proportion =86 out of 150.

Reply rate = 57.3%


  • Male Y

Figure 12. Mail Inbox after 7-days period; “Meet me” section; Number of visitors.


Figure 13. Screenshot showing Contact History: Contacts Received.


Figure 14. Screenshot showing Contact History: Contacts Initiated.

Mail inbox = 121

Meet me = 50

Contacts received = 48

Reply proportion = 73 out of 150.

Reply rate = 48.7%

b) Moderately-high attractiveness dummy profile:

 Mail Inbox after 7-days period for Male F: Mail Inbox after 7-days period; "Meet me" section: Number of people who wants to meet this user; Number of visitors.

Figure 15. Mail Inbox after 7-days period for Male F; “Meet me” section: Number of people who wants to meet this user; Number of visitors.


Figure 16. Screenshot showing Contact History: Contacts Received.


Figure 17. Screenshot showing Contact History: Contacts Initiated.

Mail inbox = 27

Meet me = 11

Contacts received = 1

Reply proportion = 26 out of 150.

Reply rate = 17.3%


c) Medium- attractiveness dummy profile:


Figure 18. Mail Inbox after 7-days period; “Meet me” section: Number of people who wants to meet this user; Number of visitors.


Figure 19. Screenshot showing Contact History: Contacts Received.


Figure 20. Screenshot showing Contact History: Contacts Initiated.

Mail inbox = 9

Meet me = 3

Contacts received = 2

Reply proportion = 7 out of 150.

Reply rate = 4.7%


Table of results Study 1:

Sin título

Dating pool in meet me: The highly-attractive men has as mean 51 women interested (potential matches in “meet me” section) per week, 4.6 times more women interested than the moderately-attractive male (11 potential matches), and 17 times more women than the medium-attractive man (3 potential matches).

Contacts received: The four accounts combined received 87 unsolicited messages, and two best looking men monopolized 96.6 % of these contacts.

Reply rate at first message: The mean response rate for the two most attractive men is of 53%, quite higher than the response rate for moderately-attractive profile, (17.3%) and much higher than the average-attractive profile (4.7%). Or in other words, best attractive guys are reciprocated by more than 1 out of 2 women. While the moderately-attractive F gets around 1 out of 6 females, and the average-attractive around 1 out of 21 women.


I must admit that response rates for the highly-attractive men were lower than I had expected. I had hypothesized before running this first test that this response proportion would be close to 75% or higher.

Correlating acceptance thresholds with the strength of choice, or choosiness, can make intuitive sense as individuals with higher acceptance thresholds (females, mainly more attractive ones) are selecting a smaller elite portion of potential male mates. So individuals with higher acceptance thresholds will also need to expend greater mate search effort to find a sufficient number of acceptable mates.

It would not be difficult to infer that these two males profiles within the top tier of attractiveness would be above the mate preference thresholds of the most of the sampled women.  So maybe there are a certain percentage of women that are sure that they are talking to real profile when they are contacted by a highly-attractive guy, but maybe other amount of girls think these two male accounts are catfishes.

On the other hand, it can not be observed an association between occupational status and messaging statistics since I’ve  not controlled for this variable. There could be a factor that affect slightly messaging statistics, although according to the results of having effect it should be quite irrelevant. In a forthcoming study I will study this factor in one of the male dummies used on this first research, but I’m going to change this factor and holding the physical attractive variable constant, we’ll be able to come closer to understanding the true effect of the ocupational status.

I will not discuss the results of this first study in depth, since I prefer to wait for a more comprehensive data base. In future studies I will continue conducting tests using different male dummies,varying the location, analyzing more exhaustively response rates (reply rates at second message), and also researching on women’s profiles. So I would urge readers to waiting for succeeding studies, which I hope all of them will be more clarifiers.


About Sir Tyrion Lannister

I am not associated with any institution (which seems still necesary for get invitations to participate in writing review papers) but I am doing some theoretical unpaid research on my own. I want to work/publish some Paper but I am not affiliated with an Institution and I have not heard anything about selling research (paper) outcomes to an institution.
This entry was posted in mate choice, online dating, pof experiment, reply rates and tagged . Bookmark the permalink.

8 Responses to Plenty Of Fish Experiment: Study 1

  1. thegreatshebang says:

    As usual, fascinating results and well done. I particularly agree with your analysis of acceptance thresholds. I will wait for the future studies.

    • sirtyrionlannister says:

      Thank you, very kind. Surely I’ll be posting more dating studies in the next days. And probably some theoretical essay too.

      • jess says:

        please can you do a study on tinder. Find out what your match rate and reply rate is of each of the males you used. Then find out what percentage of girls would agree to actually meet for dates and which percentage are dtf

        • sirtyrionlannister says:

          I’d like it but that is not feasible for me due to the following constraints:

          1) Tinder requires to link a facebook account. So I would have to first create a bunch of fake accounts on Facebook, with all what this would entail.

          2) Tinder has no website support, it just works as App format for smartphones.

          Anyway I performed a study on Lovoo, a dating site similar to Tinder. Lovoo works both as App and website format (which enabled me to conduct that study). While Tinder is more widely used in US, Lovoo is better established in Europe.

  2. pluvialpremonitions says:

    What do you think about the idea that women aren’t visual? It seems like your site has established that this is not the case. Anyway, it’d be nice to hear your input.

    • sirtyrionlannister says:

      That idea is completely false, since female mate choice is focused on male visual traits. Furthermore evidence proves that women as compared to men state more restrictive preferences for these same features.

  3. Gvc79 says:

    Women aren’t less visual, but they are more attracted by personality traits. If a guy is in control and confident her attraction will go shooting up; If he’s weak,overly nice and needy it will plummet. And women regularly test a man. I remember reading a dating coach article who said 80 percent of women said they fell for a guy they initially had little attraction for,whereas 80 percent of men said that had never happened.

    For men we are very visual.. We can fall for a pretty women and hardly know her. But we all have different tastes.

    What does this mean? Paradoxilly it means a man has to have top 20 percent looks to win her on his looks alone… Because it’s only a small piece of the pie so it has to be a good piece…. Whereas men assess women accurately on their looks and will be attracted to a greater percentage of women on that alone.

    How does that play out for men online dating? Very badly for most of us… We can’t get the personality in there.. So we’re all competing on our looks.. So naturally women are all going for the top 20 percent. Because they’re the men who are able to do that. In fact okcupid had women rate 80 percent of men below average in looks. These are decent looking guys if you see the photos of those rates poorly.

    Try making the less attractive male photos showing him engaged in activities and situations that would raise his status. Im not talking about flashy car.. Etc. But with lots of friends.. Show him as football capt. Or in control.. Show he’s a valued member of the community and has lot of friends… See if that affects the results. Keep the attractive ones the same generic photos. The least attractive guy had an awful response rate and I’d say he was average Joe.

  4. Travis says:


    Why would you think that when this experiment proves otherwise?

What do you think? Let me know in the comments below!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s