# Does playing hard to get work? AB testing for romance

post by amplemaple · 2020-10-26T15:29:21.960Z · LW · GW · 26 comments

## Contents

  Motivation
Intuition

interest
the prior
Experiment design
control
treatment
I will do in both cases

plan
Ethics
None


# Motivation

I have long wondered if I would be more romantically successful if I played hard to get. In a majority of my relationships I pretend to be less interested in my (female) to increase their interest in me. Because I am unusually affectionate, excitable and gregarious, playing hard to get (PHTG) moves me toward the modal behavior.

I recently moved to a coastal US city and experimented with texting strategies on dating apps. I switched to low effort, easy-to-respond-to texts, and greatly increased my dates/week rate. I now go on about two dates a week. I worry that when I meet someone that I really like, I get too excited and signal "desperate" or "low status". So only less exciting women attend the second date. This could extend the girlfriend search and force me to settle for a bad match.

## The Intuition

In a date, women have to assess lots of behavioral factors about a partner. Unfortunately, dates do not offer real-time tests of problem-solving, social, physical and mental health skills. You just learn if the person is an interesting talker. So women want to infer about partner's behavioral traits from limited information.

One mechanism that female primates (including human women) have evolved is to prefer men who are high in status. Becoming high status requires those social and mental health attributes that women want, so status is correlated with effectiveness (presumably it also had direct benefits in the evolutionary environment). Women therefore evolved to be more attracted to high status men. My favorite dating blog puts it

Men can easily create [the impression that they are lower status than their partner] by playing out too many low status behaviors, making women feel like they're on a pedestal. Women are evolutionarily programmed to lose attraction for men who make them feel this way.

But women need honest signals of status (signals which are difficult to fake). A great signal would be how my peers treat me, but this signal is unavailable on a date. Instead women can only judge my status from surface signals and my reactivity.

## Reactivity

To avoid reinventing the wheel, I'll quote Grotesman's practical definition here in full.

When you feel reactive in an interaction, your brain is saying, “this person is interesting and important; give them your undivided attention”. Your behaviors will have the characteristic signs of low status: anxiety, discomfort, excitement, eagerness, anger, curiosity.

The higher a person’s status is relative to yours, the more you react in an interaction with them. The lower the status, the less you react – or maybe you don’t react at all.

High-status, unreactive behaviors

• Holding eye contact
• Extending limbs, taking up a lot of space
• Exposing vulnerable body parts: throat, abdomen and groin
• Succinct and monotone speech
• Disclosing little information
• Comfortable and relaxed body language
• Emotionally composed
• Ignoring questions or requests
• Interrupting
• Breaking rapport
• Still body positions
• Slow movements

Low-status, reactive behaviors

• Obeying demands
• Speaking verbosely
• Contorting the body to take up little space
• Darting eyes
• Disclosing a lot of information
• Overly loud or quiet voice
• Avoiding eye contact
• Being the first to laugh after making a joke
• Trying to impress
• Being defensive in disagreements

(Hugh Ristik's take [LW(p) · GW(p)], A related video )

Intentionally reducing my reactivity helps me play hard to get.

## Reducing interest

Furthermore, expressing too much interest is hard to recover from on a first date. Expressing too little interest is relatively easy to recover from. Waiting to express interest in the partner until you're confident they will reciprocate is a better strategy. Evidence

• Most of my relationships began when I intentionally hid my interest
• My relationships have a strong tendency to end shortly after I make an unreciprocated affection signal
• Women often complain that men seem "desperate", which probably means both low status and too interested.
• Folk wisdom that "when it rains it pours"

## Setting the prior

Because I am highly reactive and affectionate relative to the average man, I expect I can gain by reducing my reactivity and hiding interest in the first three dates. However, there is also a folk wisdom that "being yourself is always most attractive" or "being authentic is best". I am doubtful but cannot disprove it. Furthermore, I might be too bad an actor to hide my feelings. Both theories imply intentionally reducing reactivity and affection will have negative results. I will test the claim that "intentionally reducing reactivity and affection for the first three dates will increase attraction in partners" (H1)

My prior is 70% that the H1 is true (conditional on null being false).

# Experiment design

Fortunately now that I have frequent dates, I can test the theory. For the next ~20 dates I will flip a coin before arriving at the venue. If the coin lands heads, I will intentionally reducing my reactivity and interest signals. If the coin lands tails, I will make no effort in either direction. I talk a lot and loudly. I tell the other person tons about my life. I riff a lot on their statements.

## The control

My control behavior is my regular date behavior, but people vary wildly in their courtship behavior so that deserves explanation. I have ADHD, am extroverted and am prone to ranting. I fidget a lot. I make lots of eye contact. When I am interested in a partner, I give a lot of reinforcement. I laugh hard at their jokes. I validate their statements (sometimes even bad ones). I do tease a lot, which is high-status.

It's difficult to compare my behavior with others because I do not date men. From observing other couples I am louder, more talkative and more reactive, but I cannot distinguish first from 30th dates at the table over. These are my best guesses of comparative behavior.

## The treatment

In the treatment group I will intentionally reduce the interest signals. In general, I will attempt to observe the dates reactivity and interest and exactly match it. That could include:

• Thinking to myself "Yes I am anxious that I will seem unattractive. That anxiety is normal and healthy. A failure is an opportunity to learn, and you only need one success."
• Occasionally feigning losing interest in a line of conversation if the other person is also losing interest.
• Allowing longer silences into the conversation.
• Open hand gestures (hands on the table or expansive gestures)
• Disclose aspects of my life more slowly than my partner
• If the other person is contributing less to the conversation, asking direct questions about their life like "who are you closest to?", "What did you do after college?" or "Are you friends with your coworkers". Ideally these questions are short but open up space for her to overshare
• Between venues walking slightly ahead of or alongside her
• Teasing her more than I normally do
• Generally repressing the excitement I feel after each conversational move
• Never ever ever talk about meta dating

## Behaviors I will do in both cases

• Make lots of direct eye contact (unless feigning losing interest)
• If asked "What I'm looking for" I will dodge the question. It has been my experience that if you say "If I meet the right girl, I would like a girlfriend", she will assume that you want to date her. This is too much early interest, and an instant K.O. for the romance. I have not decided how I will solve this problem. Whichever solution I choose, will be applied evenly.

## Data

In the treatment category I will continue the altered behavior for three dates. However, I will only gather data from the first two, since I might lose interest after the first date and skew my own data. I'll collect three signals of interest

• Shows up to second date
• Invites herself over for sex
• Eye contact (subjectively reported)

## Analysis plan

I reject the null "that PHTG does not matter". In an academic sense, the null is perfectly possible. In a practical sense, I must choose to PHTG or not. If the null is true the decision does not matter, so getting it wrong has no practical consequences.

I set my prior at .65 that PHTG is more effective. PHTG is more consistent with the presented evidence, but not overwhelmingly so.

After 10 dates in each group, I will do a t-test on both second dates and eye contact to produce a likelihood ratio. I will post that likelihood ratio in the final analysis. Then I just multiply the prior by my LR to find my posterior.

If my posterior belief is between .4 and .6, I will just not PHTG. It's too much of a hassle.

# Ethics

All posts about dating get ethical responses because they inevitably violate our vague but intense norms about dating. Ultimately, I must present myself somehow, and I choose the best light possible.

Because this is an AB test, ethical complainants face an additional hurdle. I'm going on these dates anyway, and must apply either the treatment or the control. Therefore, if one wishes to state that the AB test is unethical, one must further assert that either "all treatment" or "all control" or "do not go on dates" would be more ethical (non-consequentialist ethicists please start your own post to debate philosophy).

comment by Liron · 2020-10-26T13:59:59.248Z · LW(p) · GW(p)

Cofounder of Relationship Hero here. There's a sound underlying principle of courtship that PHTG taps into: That if your partner models you as someone with a high standard that they need to put in effort to meet, then they'll be more attracted to you.

The problem with trying to apply any dating tactic, even PHTG, is that courtship is a complex game with a lot of state and context. It's very common to be uncalibrated and apply a tactic that backfires on you because you weren't aware of the overall mental model that your partner had of you. So I'd have to observe your interactions and confirm that "being too easy" is a sufficiently accurate one-dimensional projection of where you currently stand with your partner, before recommending this one particular tactic.

Instead of relying on a toolbag of one-dimensional tactics, my recommended approach is to focus on understanding your partner's mental model of you, and of their relationship with you, and of the relationship they'd want. Then you can strategize how to get the relationship you want, assuming it's compatible with a kind of relationship they'd also want.

comment by amplemaple · 2020-10-26T15:20:05.050Z · LW(p) · GW(p)

Instead of relying on a toolbag of one-dimensional tactics, my recommended approach is to focus on understanding your partner's mental model of you, and of their relationship with you, and of the relationship they'd want.

That's a bit vague and difficult to enact. I want to get better at "Understanding your partner's mental model of you". That's the point of the AB test. I'm very open to alternative strategies for building this skill. I hope that responding to interest signals with more reactivity and disinterest signal with less reactivity will provide leverage on that question.

It's very common to be uncalibrated and apply a tactic that backfires on you because you weren't aware of the overall mental model that your partner had of you.

That seems like it would be true for many man-strategy dyads. It might be true for this dyad. Unfortunately without doing the experiment, there's no way to know. However, I do know that I am generally more reactive and become interested earlier in relationships than most men. So this seems worth trying.

I'll lay out some alternative strategies you might be advocating for:

## Building practical skills for theory of mind

The alternative strategy, which I think you're suggesting, is to build general social intelligence, like "reading the mind in the eyes test". I could improve this by having lots of conversations with friends. I could constantly try to guess what they are thinking, and respond to it. That should build my ability to "undstand my partner's mental model of me".

One problem is COVID, so I can't really get face-to-face interaction outside of dates (major frustration in my life atm). The other problem is that I still need to behave some way on dates. So I can either do the AB test, all treatment, all control, or not go on dates. So I need to answer that question in a practical sense.

## Just do control

Maybe you're suggesting I just do control? My prior is still 60% that PHTG would be more effective on dates on average. To do just control, I would forego all the value of finding out. That could only be worth it with an unjustifiably low prior (10%), such that checking the PHTG strategy is no longer worth it.

## Do the AB test, but remember there are other factors

Maybe you are suggesting I "have an open mind" about the different causal factors involved.

This actually updates me toward the "all treatment" strategy. The reason is that PHTG requires I read a lot more signals than I normally do on a date. So if the set of causes is large, I'll only identify them from micro facial cues. So any strategy that forces me to practice theory of mind more will help me identify those factors in the long run.

comment by Ericf · 2020-10-26T17:17:02.924Z · LW(p) · GW(p)
1. Acting on Level 2, rather than Level 1, adds additional fail conditions and stress. http://benjaminrosshoffman.com/simulacra-subjectivity/
2. If your date realizes that you are "PHTG" that is, itself, a signal and will select for/against certain types of partners.
3. Other than the "teasing her more than I normally do" and "walking in a specific place relative to her" everything in your treatment group could also be called "being a good date."

I would recommend, in your own head, think of it not in terms of "playing hard to get" but in terms of "treating your partner with respect." If you focus on helping your partner have a good time (ie, you leave the date thinking "we both had fun, but she put in way less work than I did"). Then you can see if the extra effort is producing more of what you want.

On that note, are you clear in your own mind about what your goal is? "What does it profit a man to gain the whole [girlfriend] but lose his [happiness]" and all that.

comment by amplemaple · 2020-10-26T18:01:54.652Z · LW(p) · GW(p)

Thanks for the response!

Other than the "teasing her more than I normally do" and "walking in a specific place relative to her" everything in your treatment group could also be called "being a good date."

Sounds like I'm on the right track :)

I would recommend, in your own head, think of it not in terms of "playing hard to get" but in terms of "treating your partner with respect."

"Treating your partner with respect" is a poor heuristic. It includes some great behaviors like listening but also some terrible ones that "put the woman on a pedestal". If you think respect has a special unusual definition on first dates, just taboo the word for better communication.

One could think about "being coy" or "respecting myself" or "withholding judgement until I know the whole woman" if they find "PHTG" distasteful. I have no preference.

Acting on Level 2, rather than Level 1, adds additional fail conditions and stress. http://benjaminrosshoffman.com/simulacra-subjectivity/

This concern I hear a lot. I originally applied it to my dating-app-texting and send texts that felt natural to me. Those majorly underperformed texting strategies which relied on a few heuristic rules (short, easy to respond to, more teasing, etc.). We also know that most male animals do not use "be yourself" - see here. Also, the nature of system 1 is that it learns increasingly complex behavior with habit and reinforcement. So treatment could easily underperform at the start but eventually overperform.

Generally, I put some weight on this argument but not enough to switch to control only.

If your date realizes that you are "PHTG" that is, itself, a signal and will select for/against certain types of partners.

Half the probability mass in the "control is better" hypothesis comes from me being a bad actor, yes. On the other hand, looking desperate extends the number of dates needed to find an equivalent compatibility partner.

On that note, are you clear in your own mind about what your goal is? "What does it profit a man to gain the whole [girlfriend] but lose his [happiness]" and all that.

After the experiment if I hated PHTG but to outperformed the control, I can choose to do control anyway. That's unlikely but possible.

comment by Ericf · 2020-10-26T21:52:45.530Z · LW(p) · GW(p)

I would recommend getting your (cis het male) dating advice primarily from women who you know and would like to date, but who already have husbands/boyfriends. Rather than solely from other cis het males on the internet.

"Respect" in this context means treating your date as an autonomous human with an internal narrative, desires, thoughts, history, and everything else that makes you a unique person. Rather than as an inscrutable piece of software or machinery that you are trying to figure out and/or get to act in a specific way.

And the final warning isn't about experimental design, its about not tuning yourself into a paperclip maximizer (especially when what you really want is a staple)

comment by gjm · 2020-10-27T01:29:40.659Z · LW(p) · GW(p)

especially when what you really want is a staple

An amplemaplestaple?

comment by amplemaple · 2020-10-26T23:56:26.577Z · LW(p) · GW(p)

I would recommend getting your (cis het male) dating advice primarily from women who you know and would like to date, but who already have husbands/boyfriends. Rather than solely from other cis het males on the internet.

Attraction is a subconscious process. If women or men could just introspect and output the attraction function lots of psychologists would be out of a job. Sadly we cannot.

"Respect" in this context means treating your date as an autonomous human with an internal narrative, desires, thoughts, history, and everything else that makes you a unique person. Rather than as an inscrutable piece of software or machinery that you are trying to figure out and/or get to act in a specific way.

This is my perspective

comment by Bucky · 2020-10-27T15:57:34.808Z · LW(p) · GW(p)

If PHTG is successful, do you expect more or less eye contact from your dates? If PHTG raises your status this implies it puts your date at relatively lower status and according to your list they would make less eye contact. However your post suggests that you'll take it as a sign of interest as is usually the case.

comment by amplemaple · 2020-10-27T17:41:30.772Z · LW(p) · GW(p)

Interesting question Bucky!

I think on a date eye contact is usually a signal of intimacy and interest. If you were in a meeting with your boss you might look at their chin or your shoes as a sign of submission, but people would never do this is in a date. On a date eye contact is more about intimacy, interest, rewarding good convesation, etc. Therefore I expect more eye contact in the successful case.

What else does eye contact signal? It also says something about interest in the topic. I do observe that when people lose interest they look away a lot on dates. But this could just mean I'm talking to much, or picked a bad topic. I think conversation style choices should be guided by a "host" mindset rather than a status mindset. Making the other person comfortable is your prime concern. But when your doing posture, speech speed, pacing personal information, or defining the relationship it might be the dominant concern.

What do you think Bucky?

comment by Bucky · 2020-10-28T10:11:42.066Z · LW(p) · GW(p)

I think that both effects are likely and that this will add noise to the measure.

Noise is my main concern about your experiment in general - with only 10 samples in each treatment any effect would have to be large to reliably show up. If you were doing a t-test with p<0.05 then you would need Cohen's d of 1.3 to get a significant result. This would be the equivalent of PHTG having the effect of moving a median date to a 90th percentile date which feels unlikely.

Obviously you're being sensible and not being frequentist but the underlying problem is still there - even if PHTG has a decent sized effect the experiment might not show it, or, worse, if PHTG makes things slightly worse it could show up as being good in your experiment.

I would suggest that you try to work out a power calculation (even just by setting up your planned calculation and plugging in some fake numbers to see what happens) - if PHTG is slightly harmful to your chances (say 20% decreased chance of getting a second date) what are the odds that the experiment will lead you to accept PHTG?

As an aside, have you read this on putanumonit? It presents an alternative to PHTG which you might find interesting.

comment by Nacruno96 · 2020-10-26T19:38:56.620Z · LW(p) · GW(p)

I actually agree with many of your ideas and I think the things you are doing and your approach will show results and work, but there is one huge problem that is making me never go so far to actually change my behavior which makes me end up behaving as I am. I just don’t feel comfortable if I act. Before long I will end up having no respect for the girl I am dating and if she shows interest in me I would know she does not like me but the image I present her. Also implicitly you end up showing more regard for a stranger you don’t know than for yourself, because you basically end up fighting for someones affection instead of giving someone the choice to like you or not like you. I tested your strategy for myself, but I wasn’t happy with it.  And often the underlying problem to I think the  top 10% of the smartest men in society ending up with no partner is not about things they do wrong it’s about the distribution disparity of smart men to smart women. There are far more smart men than women. As there are far more stupid men than stupid women. there are 5 times or 4 times as many men with iq >140 and with iq >120 there is still a big sex difference. So I think that is the real problem. Which is why I prefer hookers as long I am not really dating someone who is about as smart as I am. By the way I never have taken an iq test so I am not sure how smart that would be.

comment by Lukas_Gloor · 2020-10-29T12:59:19.481Z · LW(p) · GW(p)

I just don’t feel comfortable if I act.

I think that's a great trait to have and I'd strongly recommend keeping it. If you can find enough things you like about yourself (and maybe have also worked on yourself to that end), you can also acquire genuine confidence in this way that feels way more robust than acting.

Maybe you've thought about this already, but I'd flag that some people (and more women than men) don't themselves compartmentalize so much between "just sex" and "romance". Humans have some degree of sexual dimorphism around attraction (e.g., "demisexuality" is rare among men but not that uncommon among women). So, the habit you mention and the way you phrase it might substantially decrease the pool of otherwise compatible partners.

With the phrasing, I'd be worried that what many people might take away from your paragraph is not so much "This person cares about avoiding situations where they'd be incentivized to act inauthentically, therefore they prefer prostitutes over dating people with whom conversations don't feel meaningful", but rather "Something about intelligence, therefore hookers".

The mismatch in psychologies is harder to address than the phrasing, and maybe that just means you don't think you're a good match to others who view the topic differently – it really depends on what feels right all things considered.

Just to be clear, I don't necessarily mean "view it differently" on moral grounds. For instance, I don't think extraverted people are immoral, but I'd feel weird and maybe too insecure with a partner who was too extroverted. Similarly, some women will feel weird and insecure if their partner has too much of a "men are bad/threatening" psychology, whether or not they think it's immoral. So finding other ways to meet the same needs could make sense if one worries about the pool of potential soulmates already being small enough, and if one places value on some of the normative intuitions, like importance of emotional connection during intimacy with a partner and not wanting to risk it being adversely affected. (The extraversion analogy isn't great because it sounds wrong to repress a core aspect of personality – the question with compartmentalization of romance vs. sex is if it's that or more/also influenced via habit formation and so on. I don't know much about the empirical issues.)

Maybe you think what I write in the paragraphs above goes way too far in the direction of:

Also implicitly you end up showing more regard for a stranger you don’t know than for yourself, because you basically end up fighting for someones affection instead of giving someone the choice to like you or not like you.

I'd say it depends. "Accommodations" come in degrees. Also, if you make them for any stranger, you're indeed not showing respect for yourself (as well as treating other people's personalities as interchangeable). However, if you find yourself particularly motivated to be good for partners with a certain type of character, that means that you already want to be the sort of person who appeals to them.

comment by haasn · 2020-11-05T06:38:13.144Z · LW(p) · GW(p)

My anecdotal evidence is that, as you point out, a relationship formed on desperation covered up by acting tends to end the moment the acting dissolves. This implies that sustaining such relationships means acting for the rest of your life, an outcome you probably want to avoid.

Given that your criteria for a "success" are, essentially, "going on dates" and "having sex", I'm not sure whether you're content merely maximizing those two factors or actually care about the resulting relationships being fulfilling, but I'll give you the benefit of the doubt and assume you're actually interested in trying to maximize the latter.

As such, the actual advice I have on this topic is that rather than trying to act away your desperation, it's both more effective and ultimately more rewarding to just "simply" decrease your level of desperation. A discussion of how to accomplish this would be even more out-of-scope, but in my case it essentially amounted to about a year of psychiatric treatment. The benefits of this are not just that it increases your life satisfaction in general (even while single), but also, importantly for this topic, allows you to shift your dating strategy away from "soliciting a large number of short-term dates in rapid-fire manner" and towards "randomly meeting people dynamically in environments sustained over longer periods of time". I think that this is more successful and rewarding strategy in the long-term, and not one that depends on PHTG (nor where it would be beneficial).

In summary: The best way to play hard to get is to just... actually be hard to get, which implies being happy enough with yourself that you don't really need a partner in your life; especially if they're not helping you achieve your real goals. (Besides "having sex", I mean. But then, why not simply pay somebody?)

comment by amplemaple · 2020-11-05T09:14:43.938Z · LW(p) · GW(p)

Do you have any evidence that happy people fall in love slower than happy people. So far I have only noticed a slight inverse relationship where crisis retard my romance response. All attempts to slow the response by having more friends failed (although being happier is nice for other reasons). Past attempts at PHTG have often succeeded.

If you believe that I fall in love faster than other people because I am “desperate” then all my past relationships should have collapsed when I told the person I liked them. This has not occurred. As long as I wait until the fifth of sixth interaction to say “I like you” the relationships are perfectly stable. Maybe that model just applies to certain people

After I’ve had sex with a woman a few times I find being affectionate is punished much less. It’s a temporary strategy to get through the early stages of courtship. I’m sorry but “Acting for the rest of your life” is a bit melodramatic. On their first dates people pretend to be kinder, funnier, smarter, taller and hotter than they really are. They don’t keep pretending forever.

comment by voxelate · 2020-11-05T05:38:25.097Z · LW(p) · GW(p)

This is interesting, but in my dating experiences I've not come up with any solid conclusions whenever I tried experimentation.

I see a few issues with the design of this experiment:

• You're the test designer, observer, and one of the subjects.  There's a ton of room for bias.
• The sample size is only 20 dates.
• There are 16 types of personality types if you assume Myers-Briggs is correct.
• People have a wide variety of mental states before going on dates, and various motivations.
• it is impossible to go on a date, act one way, go back in time and date the same woman, while acting differently, and compare the responses.
• The fact that you're going on dates to do this experiment may affect your actions on the date in a way that you wouldn't act if you were not doing it as an experiment.
• You may accept dates that you might not otherwise accept, for the sake of more data.
• There's multiple subjective datapoint/actions (e.g. "lots of eye contact", "teasing ..more than normal", "generally repressing".

If you want to do this for fun or as a learning exercise, that's understandable.  But I am simply skeptical that you can get actionable conclusions from this.  For actionable conclusions, I suggest reading up on existing studies (but be skeptical of them too).

comment by amplemaple · 2020-11-05T09:10:34.924Z · LW(p) · GW(p)
• Meters-Briggs has a junk, so that’s fine. My past romantic partners have consistent big five profile either.

Sure there are sources of bias. I can’t control for them. But th

1. No generalizability needed. It just has to work for me and the women I normally date. Who cares if it would work for the “average guy”

2. I have no reason to assume a small effect size. My current positive rate on dates is 20%. Lots of room for improvement.

3. The subjectivity isn’t a problem because I’m the analyst and implementer. I don’t need well defined terms like communicating an academic study. I can reproduce the study from memory whenever I want.

Finally, there is no real cost to the study. Currently I’m ambivalent between treatment and control.

comment by amplemaple · 2020-11-05T04:13:28.108Z · LW(p) · GW(p)

The main reason I settled on this strategy is the timing of my rejections. Almost all of my rejections come shortly after I make a signal of interest. It could be that women coincidentally smell my terrible BO when I mention I want a long-term patner, but I doubt it. Is that a 99/1 likelihood ratio proof?

No, but I have to start figuring this out somewhere.

comment by haasn · 2020-11-05T06:47:17.515Z · LW(p) · GW(p)

"It could be that women coincidentally smell my terrible BO when I mention I want a long-term patner, but I doubt it."

It could also be that your signal of interest is the catalyst for them deciding to act on their pre-existing feelings of doubt. Perhaps they were already harbouring doubts but didn't voice them until you, essentially, forced them to decide.

comment by voxelate · 2020-11-05T05:40:48.936Z · LW(p) · GW(p)

What kind of indicators of interest are you making, how are you making them, and when are you making them?

comment by amplemaple · 2020-11-05T09:01:39.888Z · LW(p) · GW(p)

Good question.

• on the first date I do lots of flattery, both verbal and tonal
• On the second date I may say “I like you” or mention that I am seeking a long term partner if asked
• My body language tends to be more boyfriend
• I know I want to date someone by the second date. Most women take much longer. I’m uninhibited so my intentions usually show
• I overshare about myself (1st and 2nd dates)
• I prefer to send longer/higher effort texts than most guys
• sometimes I intentionally or unintentionally hide my attraction signals (verbal and body language). My last three relationships all started with that behavior.
• I mention qualities I prefer in a long term partner on the first or second date (so stupid in retrospect)
• I ask women what they want in a long term partner on the first or second date
comment by hg00 · 2020-10-28T06:03:00.499Z · LW(p) · GW(p)

Looking forward to the results.

comment by philip_b (crabman) · 2020-10-26T20:08:16.246Z · LW(p) · GW(p)

I found this post useful, because it contains practical ideas about how to perform personal experiments in dating.

I think you are confused about how statistics work. Student's T-test is a frequentist thing, while in some places you say that you have aprior which suggests you want to do bayesian analysis.

The sentence

My prior is 70% that the H1 is true (conditional on null being false).

makes no sense because of the part in parentheses.

comment by amplemaple · 2020-10-26T20:44:34.363Z · LW(p) · GW(p)

Here's my plan for analyzing the data, let me know what changes your would make.

Bayes rule odds form:

So all I need from the experiment is the ratio of how likely I am to observe this result given H1 vs. given H2. At first I thought calculating that with t-distributions would be trivial, but I'm noticing a problem now. If I have 7 successes in treatment and 3 in control, I can use the t-dist to calculate how often I would observe that in the null. But since the null is decision irrelevant, that doesn't get me the likelihood ratio which I want. I see that now. I'll circle back to this problem when I have more time later.

Found a guide

I don't care about the null because if the null is true and I get the wrong answer, it doesn't matter.

comment by philip_b (crabman) · 2020-10-27T10:03:40.979Z · LW(p) · GW(p)

What is your null hypothesis? Nowhere does your post says that. I suspect you don't know what a null hypothesis is.

From your post and your comment, I infer that you want to find the probability of "intentionally reducing reactivity and affection for the first three dates will increase attraction in partners". That doesn't work well with bayesian analysis. Instead you should try to get a posterior distribution over the value of how much it increases attraction.

I think if you want to do the bayesian data analysis, then one of the simplest ways you could model your situation is as follows.

If you PHTG, you achieve sex (or whatever it is you're after, but I'll just say sex for simplicity) with probability p∈(0, 1). If you don't PHTG, you achieve sex with probability q∈(0, 1). Currently, you don't know the values of q and r but you have a prior distribution p(q, r) over them. In this prior p(q, r), q and r are not necessarily independent. On the opposite, I would expect that they correlate (with respect to the prior p(q, r)) very strongly, because if you often achieve sex with one strategy, probably you'll also be able to do that with the other strategy, and if you can't achieve sex with one strategy, probably you can't with the other. Next, you will go and do the experiments (go on dates and randomly choose whether to PHTG). An experiment is like tossing a biased coin. If you are PHTG, you are tossing a coin which lands on heads with probability q. If you are not PHTG, you are tossing a coin which lands on heads with probability r. After n experimental results , you update your distribution over the values q and r: and this is the result you get. I think this models represents your situation fairly well.

I don't know what prior p(q, r) you should choose in order to have it fairly close to your actualy beliefs while at the same time making the computation tractable. A simplification you can try is imagining that prior to the experimental data, q and r are totally independent from each other. Then your situation is simply two separate situations, in each you are trying to estimate the biasedness of a coin. Then you take the prior of q to be a beta distribution, and the prior of r to be a beta distribution as well. Then you open "Data analysis a bayesian tutorial - Sivia Skilling" (can be found on libgen) page 14 example 1 "is this a fair coin?" and do whatever it says. Another thing you could probably do is come up with some kind piecewise-constant prior p(q, r) manually and perform the bayesian analysis by simulating everything on the computer rather than tinkering with integrals on paper. Formally, this is called Monte Carlo integration.

Also, instead of treating the outcomes as binary (sex or no-sex), you could treat them as real numbers which represent how well it went. I think this way you'll need less experiments to get a conclusion. For this case, you can read "Bayesian Estimation Supersedes the t Test - Kruschke 2012". That paper describes how to do bayesian analysis when you have two groups (treatment and control) and you want to measure what the treatment does if it does anything.

comment by amplemaple · 2020-10-27T15:35:10.614Z · LW(p) · GW(p)

Thanks for pointing me in the right direction with these! My degree is really frequentist and slow paced, excited to get to work on this analysis.

Clarification on null hypothesis:

The null hypothesis is that there is no difference in effect on the dependent variable from the treatment and control variables. I am not assessing the truth of the null hypothesis because if it is true, then I can pick whichever one I want. If control is better, then picking treatment is negative utility. If treatment is better then picking control is negative utility. If the null is true, then I am free to do treatment or control without suffering in either case. Therefore I gain no utility from a test to see if the null is true or not.

Consider shifting your tone when people know less stats than you. Saying " I suspect you don't know what a null hypothesis is" makes people feel defensive and not willing to take your useful advice. Try saying "can you clarify what you mean by ____" or "here's a common definition of a null hypothesis".

Clarification on dependent variables:

I was going to code the outcome as 1 if sex|second date. The thinking is that only women who are attracted to me have sex with me (but may not want to date, for lots of good reasons). Meanwhile many women who are attracted to me do go on a second date. But few women are attracted to me but do neither sex nor a second date. Since attraction is the concept I want to explain, this should have the best specificity and sensitivity of available measures.

I am considering a second DV using eye contact during date (qualitative) as a robustness check. I think some people do lots of eye contact on all dates as a subconscious influence strategy, so it has more false positives than the other two.

comment by philip_b (crabman) · 2020-10-28T09:52:33.268Z · LW(p) · GW(p)

Alright, it seems you do know what a null hypothesis is. Glad I could be of help.