Applied Bayes' Theorem: Calculating the probability that she's over me. Could somebody check my work? TT_TT

post by abcd_z · 2013-07-19T02:59:21.923Z · LW · GW · Legacy · 23 comments

Contents

23 comments

Solve for:  The probability that she's over me given that she didn't answer my call.

Estimated probabilities:
The probability that she'd miss my call given that she was over me: 90%
The probability that she's over me: 30%
The probability that she'd miss my call given she was not over me. 10%

(P(over me|missed call)= P(missed call|over me)*P(over me))  /  (P(missed call|over me)*P(over me)+P(Missed call|Not over me)*P(Not over me))

P(O|M)=(P(M|O)*P(O))/(P(M|O)*P(O)+P(M|N)*P(N))

P(O|M)=(.9*.3)/((.9*.3)+(.1*.7))

P(O|M)=(.27)/(.27)+(.07)) = .27/.34 = .794

Probability that she's over me given that she didn't answer the phone: 79.4%

TT_TT

 

EDIT: Something I've noticed here is that people are pointing to the chosen priors and saying that they seem unrealistic.

In our 3-year relationship, she almost never missed my calls, or if she did she would contact me back as soon as she realized that she missed my call.  In the current situation, she did no such thing.

EDIT2: Yes, we broke up.  Sorry I didn't make that clear.

23 comments

Comments sorted by top scores.

comment by Suryc11 · 2013-07-19T13:12:36.051Z · LW(p) · GW(p)

I just recently went through a break-up (SO broke up with me, it was a long-term relationship).

To be frank, this is not at all what you should be doing (i.e., doing a Bayesian calculation re the probability that she's over you, or calling her and analyzing why she hasn't called back), regardless of whether your goal is to get back together with her or to move on as quickly as possible.

The best possible piece of advice I could give you is to start a reflection document. Document your feelings, your emotions, everything. It will help, I promise. What also helps is to write down every single negative you can think of about your relationship/ex. (You will be tempted to idealize the relationship/ex, and writing down negatives that you may have looked past while in the relationship will help you.)

Standard break-up advice: do not contact her under any circumstances (No Contact), remove anything that reminds you of her (includes things like unfriending ex on FB), go to the gym (or engage in other activities that'll keep you busy), and talk to friends and family.

Think of a break-up as going through withdrawal: every time you break No Contact, you're relapsing and making it much harder for your brain to get used to being without her. Another way to think about it is that your break-up is a wound, and each time you break No Contact, you're ripping up the forming scab and peeking under it. No Contact is also ideal for getting back together with her (distance makes the heart grow fonder, etc.), if you truly still want that .

Take the Outside View. Lots of people have gone through break-ups thinking that they'll feel crappy forever and then are fine just a few months later. Time really does help.

If you're anything like me, going through a break-up will really make explicit the disconnect between your reptilian, System 1 self and your more deliberative/rational System 2 self.

One final piece of advice: closure does not help in the vast majority of cases. Often, a desire for closure is just your subconscious justifying a desire to talk to her again.

EDIT: If it helps at all, given that it was a 3-year relationship and that the break-up seems pretty recent, she's probably not over you. She may be trying to maintain distance to avoid feeling guilty and confused, and so that she won't feel tempted to second-guess her decision. Also, usually the dumper has been considering this for a while, so even if you feel like she doesn't care or isn't hurting, it may just be that she started the grieving/moving on process earlier than you did (for some insight into how many dumpers think: http://np.reddit.com/r/BreakUps/comments/1htfcn/is_it_true_that_dumper_check_out_of_the/).

comment by Dorikka · 2013-07-19T04:49:57.051Z · LW(p) · GW(p)

I'm inclined to say that this belongs in an Open Thread, especially as they're posted more often now.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T05:16:58.795Z · LW(p) · GW(p)

What's an Open Thread and why would this belong there?

Replies from: ygert, roystgnr
comment by ygert · 2013-07-19T13:10:15.400Z · LW(p) · GW(p)

Hey! Why would someone downvote this? It's a legitimate question. Remember, each of us was once new here too. To answer your question, abcd_z, the open thread is a thread, posted weekly now, that is intended for posting all these minor or slightly off topic things. Basically, if it's worth saying, but not worth its own post (even in Discussion), then it goes there. Your post is the kind of thing that definitely should be in the Open Thread rather than in Discussion.

comment by roystgnr · 2013-07-19T20:21:33.768Z · LW(p) · GW(p)

I'm not sure why you were downvoted, but here's my best guess:

Google search, site:lesswrong.com "open thread" On my results page, any of four of the top five hits would have answered your first question and partially answered your second.

I'm not sure how to phrase the above paragraph in a way which doesn't make me sound like a condescending jackass, but I swear I'm trying to be helpful. If your first instinct with questions like these is "hope someone will be kind enough to answer me personally" rather than "use search engines and find most answers instantly", you can probably improve your life a great deal by correcting that.

comment by metatroll · 2013-07-19T05:29:00.990Z · LW(p) · GW(p)

You could try autoaruspicy...

Replies from: abcd_z
comment by abcd_z · 2013-07-19T05:40:00.909Z · LW(p) · GW(p)

what's that?

Replies from: metatroll
comment by metatroll · 2013-07-19T05:51:13.729Z · LW(p) · GW(p)

Actually, I just realized that you are already engaged in an unusual Bayesian form of cyber-autoaruspicy (also known as "spilling your guts online"). It's like the character in Molière who had been speaking prose all his life, without realizing it.

comment by Mestroyer · 2013-07-19T03:10:24.560Z · LW(p) · GW(p)

The math is correct, but I don't know if those priors are realistic. Particularly the "10%" one. I miss like 50% of calls in general, having nothing to do with who's calling me. And remember that the "probability that she's over me," should be probability that she got over you since the last time you checked.

Replies from: abcd_z, abcd_z
comment by abcd_z · 2013-07-19T03:17:09.328Z · LW(p) · GW(p)

She would always get back to me within a short amount of time when she found out she missed my call. The only times that hasn't happened was when the phone glitched and she was unaware that I'd tried to contact her.

If anything, 10% is a little high.

comment by abcd_z · 2013-07-19T03:19:27.064Z · LW(p) · GW(p)

And remember that the "probability that she's over me," should be probability that she got over you since the last time you checked.

Wouldn't that be redundant?

Replies from: Mestroyer
comment by Mestroyer · 2013-07-19T03:32:07.126Z · LW(p) · GW(p)

I don't know why it would be redundant. If she has a certain probability of getting over you every day, and you called her and interacted normally Yesterday, then for it to be 30%, would require that she had a 30% chance of getting over you in a day, before you knew anything about whether she answered your call or not.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T03:38:38.773Z · LW(p) · GW(p)

I don't trust my accuracy of measurement from the time we broke up until now. The relationship between us has been...uncertain, and her ignoring my calls is the first behavioral cue I can point to and say "Okay, that means there's a good chance she's over me."

Replies from: Mestroyer
comment by Mestroyer · 2013-07-19T03:45:54.783Z · LW(p) · GW(p)

Oh, I didn't know you broke up.

You explicitly broke up, and your prior for "she's over you" is only 30%?

Also, if she was breaking up, but maybe changing her mind, a 90% chance of answering your call seems too high, but I have no experience with these matters.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T03:53:14.823Z · LW(p) · GW(p)

Ah. Sorry I didn't make that clear.
Yeah, that particular prior is low at least partly because of my wishful thinking.

Not that it seems to have helped. =/

comment by DanielLC · 2013-07-19T03:09:57.224Z · LW(p) · GW(p)

Correct, at least as far as the math goes. I prefer to use odds ratios.

Prior odds that she's over me: 3:7

Evidence ratio of her missing my call: 90%:10% = 9:1

Posterior odds that she's over me: 39:71 = 27:7 = 3.86:1

Posterior probability that she's over me: (27/7)/(27/7+1) = (27/7)/(34/7) = (27/34) = 79.4%

Replies from: wedrifid
comment by wedrifid · 2013-07-19T03:30:30.702Z · LW(p) · GW(p)

Evidence ratio of her missing my call: 90%:10% = 9:1

If I assumed this prior I'd have to conclude that I'm over, like, everyone.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T03:41:04.197Z · LW(p) · GW(p)

She almost never missed my calls.

comment by Jonathan_Graehl · 2013-07-19T04:23:48.222Z · LW(p) · GW(p)

bummer, man :) cute post.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T04:33:13.945Z · LW(p) · GW(p)

Thanks. :)

comment by moridinamael · 2013-07-20T01:31:29.535Z · LW(p) · GW(p)

I don't feel that "she's over me" is a sufficiently well-defined outcome to be meaningful. I mean, there isn't an experimental test which could confirm or disconfirm her state of over-you-ness.

comment by afterburger · 2013-07-19T04:58:51.010Z · LW(p) · GW(p)

The error bars on the result make this result meaningless. You should, at minimum, also consider how hot/successful she is relative to you (can she do better?) and whether she and you have shared goals (what are her dreams/expectations? are they compatible with yours, and are you both making progress?). It would be extraordinarily improbable for her to forget you overnight, but if she broke up with you she has likely wanted to do better for some time. The usual post-breakup advice is to focus on your own goals for awhile.

Replies from: abcd_z
comment by abcd_z · 2013-07-19T05:35:06.302Z · LW(p) · GW(p)

1) What error bars? Are you talking about the relative uncertainty of the priors? Because it's assumed that these prior are only my best estimate. You're not saying anything new here.
2) "how hot/successful she is relative to you": I haven't heard of any studies that positively correlate physical appearance and financial success with reduced time to get over a long-term relationship. I admit that it's possible that there's a connection, but as of yet I haven't seen any evidence that would persuade me to include it.
3) "if she broke up with you she has likely wanted to do better for some time." Again, you're making unjustified claims with no evidence to support them.

"The usual post-breakup advice is to focus on your own goals for awhile."
Now this I agree with.