Facebook Data Team: Relationships and Happiness

http://www.facebook.com/note.php?note_id=304457453858

by alexandros • 16 years ago

45 32 16 years ago

So, to optimize for positivity, be in a relationship but don't get married.

mrshoe2• 16 years ago

Doesn't the article contradict this strategy? FTA: People who are in a relationship seem less happy compared to married folks

iBercovich• 16 years ago

With the massive amount of relational data being created in social networks, there is a huge opportunity in the field of analytics. This particular min-study shows trends in happiness vs relationship status, which could be used for a dating site/app spin-off. In a similar manner, analytics can be used to improve media consumption (video, sound, image, text)-- content aggregation is no longer feasible because too much information is being created everyday, so we must find a way to reduce SNR and provide the user with what he/she needs to see.

I am really excited to see many analytics-oriented startups in 2010.

-Ivan

contagionhealth• 16 years ago

Agreed. I would recommend friends pursuing an MBA go for a masters degree or PhD in behavioral economics instead. In addition, I'd like to see more analytics oriented around the individual's own delta - how many times did they switch from "complicated" to a relationship in a 5 year period? How many friends did they add? How many people did they block? A more realistic measure of personal happiness might be obtained by more robust analysis of how a person's self reporting changes over time, rather than a static snapshot of how they report at a simple point compared with how many other people report that same thing in stasis.

tpyo• 16 years ago

People in an open relationship are less happy. Why?

endtime• 16 years ago

Probably jealousy + lack of emotional gratification.

Towle_2• 16 years ago

Because most people who declare themselves to be in an open relationship on Fb aren't in an open relationship. It's most often done just for the sake of being silly or telling everyone who your "BFF" is. The latter is more likely the closer you are to being a 16-year-old girl.

badave2• 16 years ago

But why would that correlate with "negativity" across a subsection of people?

kscaldef2• 16 years ago

No, people in open relationships post fewer status updates with "happy" words. They also post fewer status updates with "unhappy" words. So, we might be lead to conclude that people in open relationships choose, for whatever reason, to expose their emotions in status updates less often than others. That doesn't particularly mean they have fewer emotions, positive or negative than anyone else.

You also see this strongly in the "widowed" category, who have dramatically less happy or unhappy postings.

adamsmith• 16 years ago

I'm surprised they didn't remind the reader of 'correlation is not causation.'

I would have liked to see time domain analysis -- so when a person goes from 'single' to 'in a relationship' does one see a pop in happiness? Is it sustained?

mrshoe2• 16 years ago

They should use the same data to make a followup post which states that happy people are more likely to find a partner or that sad people have little chance of ever getting married.

Towle_2• 16 years ago

"We already have methodology for measuring the happiness of Facebook users: by considering how many positive words people use in their status updates (see the USA Gross National Happiness Index). This method allows us to see whether a person's Facebook relationship status affects how positive and negative they are." (from post, emphasis mine)

No it doesn't. Correlation vs. Causation.

Counter-example: assume these two men (simplifying to single and non-single categories) are equally happy in real life:

Singles' non-relationship status updates: "Jeff is psyched for his Super Bowl party!" vs. Non-singles' non-relationship status updates: "Tony is psyched for his Super Bowl party!" Comparison by Fb methodology: equal happiness level

Single's relationship status updates: [non-existent] vs. Non-singles' relationship status updates: "Tony loves his super awesome girlfriend Melissa!!! Sweetheart, you're the bestest gf evarrrr!!!1!" Comparison by Fb methodology: Tony seems happier.

Conclusion: obvious nonsense. Somebody tell these guys to take a stat class.

badave2• 16 years ago

She probably is, she's just an intern.

tokenadult2• 16 years ago

Did Facebook ask me if I would be happy if they did this kind of data-mining on my status updates?

physcab• 16 years ago

That type of question is probably covered in their Terms of Service.

andylei• 16 years ago

data mining seems like one of the least offensive things they could do with my data. as is clear in the report, since it's all aggregated information, there's no information tied to any person. i might be concerned that someone at facebook could read my status updates, but lets be honest, many employees at facebook probably have the ability to read my status updates anyways.

houseabsolute5• 16 years ago

They don't have to ask you since you already gave them permission.

tokenadult2• 16 years ago

References for that, please?

bootload3• 16 years ago

"... Did Facebook ask me if I would be happy if they did this kind of data-mining on my status updates? ..."

Good question. You may not have realised you gave Fb the rights to "infer" from their server logs & interactions when you created your account. However there is a solution. Get back your right to say who can peek into your server logs by owning your own servers again. Listen to Eben Moglen and his "Freedom in the Cloud" talk to understand how ~ http://www.isoc-ny.org/?p=1338

Zilioum2• 16 years ago

Yeah, great example of how facebook does not understand privacy. Why should I care if its anonymous or not, I dont want them to be even able to find out that kind of stuff about me!? It's like they dont even understand how far reaching this is; their finding out about the happiness of millions of people and they disguise it at as some kind of "V-Day Trivia". This shit creeps me out.

houseabsolute5• 16 years ago

It's fairly easy to deactivate or delete your account. Account > Account Settings > Deactivate Account. To preempt any objection that you want to be able to use Facebook, you just _also_ don't want them to analyze your status updates: that's nice, but they are not willing to offer you service on those terms. I guess it's fine to complain about it in hopes that something will change, but the winds are not blowing in that direction, Mr. Quixote.

bootload3• 16 years ago

"... It's fairly easy to deactivate or delete your account. Account > Account Settings > Deactivate Account. ..."

Is all the data deleted? cf: Eben Moglen ‘Freedom in the Cloud’~ http://www.isoc-ny.org/?p=1338

Zilioum2• 16 years ago

I'm not against analytics or facebook in general. It's just that this company and this article in particular gives me the impression, that they dont give two shits about privacy and just do it because the last few times they tried to cross the line they got caught. I'm not expecting them to be nice, no company is, but they should at least try to show some respect for our privacy - and checking whether I'm happy or not is not doing that.

notatoad• 16 years ago

that's a rather odd way of reporting numbers. "negative 5% more positivity than average." wtf?

houseabsolute5• 16 years ago

There's definitely a problem with the readability of the graphs in particular, and with some of the sentences in the summary.

kscaldef2• 16 years ago

It's worth pointing out that the 4 graphs on the page have different normalizations, which makes naive comparisons between them misleading.

greenlblue• 16 years ago

Facebook has no shame. This is a complete ripoff of okcupid analysis.

_delirium• 16 years ago

One thing I'd like to see more on these kinds of data-mining analyses are some sort of error bars, variances, etc. They tend to just report group averages, like A=5, B=2, but there's a big difference between (using, say, 1 stddev as the bracket), A=5 +/- 1, B = 2 +/- 1, in which case they really do seem to differ, and A=5 +/- 15, B = 2 +/- 25, in which case you've got mostly noise.

ytinas• 16 years ago

I recall reading a study about happiness somewhere years ago [1] that showed people's happiness generally doesn't change that much. It reacts temporarily to positive/negative effects but then tends to go back where it was.

[1] I think it was probably a critique of this article: http://www.nber.org/papers/w10499

Facebook Data Team: Relationships and Happiness

Related Stories