Friend of mine doing her dissertation and I said I'd help with her stats because I do maths...

Friend of mine doing her dissertation and I said I'd help with her stats because I do maths. My question is: Suppose we have two variables for example: a economic index and a secularity index and we want to find which is the best predictor for happiness. How do we find this? My guess is that you assume that the two indexes are independent then use a regression model? How do I know that they are independent though? They are uncorrelated but that doesn't always imply independece

Other urls found in this thread:

youtube.com/watch?v=u9bk2MrMGaA
twitter.com/AnonBabble

Idk, but you shouldn't have said you'd help her if you can't do it.

No pussy for you

I already have a gf so I'm gucci and we been friends for ages.

You can't, she's as thick as a brick.
>Theme music for your friend and this thread:
youtube.com/watch?v=u9bk2MrMGaA

How do all these sociology papers use so many independent variables then?

>sociologists
>not bastardising stats

Because they try and make connections between variables that have none?

it's sociology so it doesn't need to be correct. Just assume they're are independent in calculations and, if you have to, argue that the dependence is neglegible. Put some buzzwords for bonus points and just get those citation gains

OP here.

Suppose my independent variable is X and my dependent variable is Y. If X' is dependent on X can X' still be a control variable?

...

>dissertation
>needing help with stats

Pathetic. The social sciences could amount to something if rigorous courses on stats and probability theory were requirements to get into graduate school.

Soft sciences are only science in name. trying to apply mathematics and statistics to sociology without making a fuckton of assumptions and draeing conclusions where there arent any is a sisyphusian task and wont make sense to anyone that has any analytical skill. tell her to fuck off and either fuck your girlfriend(male) or jerk off into your sock because you arent gonna get any from doing her glorified homework

then don't be a beta faggot and don't say you will help a woman for literaly no pussy.

What the fuck is wrong with you autists. You realise you can be friends with women? When you actually get laid for the first time you'll see that not everything revolves around pussy.

projecting much?

You're a beta faggot. If she's not doing something for you and you're doing something for her, you're basically telling everyone around you that your time is worth less than hers.

Fuck off faggot.

>If she's not doing something for you and you're doing something for her, you're basically telling everyone around you that your time is worth less than hers.

Jesus fucking Christ. Not OP, but.. Do you not have friends? Are all your relationships based on transactions that you have with the people that are unfortunate enough to have to interact with you? Do you keep a record of all the favours that people owe you and send them reminders from time to time? Lmao.

Fucking hell. I mean, I knew this was Veeky Forums, but I'm still appalled as fuck. You guys are really top-tier autists in here; keep fighting the good fight, fuckers.

Women are meant to be fucked or to be mothers. This is their biological role. You denying it only makes you a fool.

> Are all your relationships based on transactions that you have with the people that are unfortunate enough to have to interact with you? Do you keep a record of all the favours that people owe you and send them reminders from time to time? Lmao.

you say that because you would need to. Like a good beta. If people actually respected you or admired you they would do anything to spend time with you. Those are my friends. Apparently people only tolerate you around them because you do them favors. Not because of the way you think, how funny you are, how you make them feel or anything.

Pathetic, really.

not him, but why are you strawmanning this hard?
also I agree it would be retarded to try and 1:1 but if people clearly ask for much more than they will and/or offer you it is understandable

Nor should you be a literal beta faggot and help male friends for no ass either. Obviously. Fucktard.

Oh. Right. Kys alfalfa tard

>Women are meant to be fucked or to be mothers. This is their biological role. You denying it only makes you a fool.

>assume that the two indexes are independent
No.
Assuming hat something is independent without ANY evidence is NOT a good Idea.

And if you are honestly saying that secularism and economics are independent you are either wrong or are making assumptions so broad that everything you do becomes worthless.


But that said, all people who read the dissertation likely will have no Idea what "independent" even means.

This is what I thought. Suppose I wanted to control for some economic index using statistical analysis, the control variable doesn't need to be independent of secularism though right?

nice, you completely misread the tone of my post. if you're gonna be such a fucking asshole, at least learn to read between the lines.

those series of questions were all rhetorical, you dumb motherfucker. I don't actually believe that those scenarios I included in my questions make for a good friendship at all, you misogynistic dick.

at least seemed to have picked up on it (...or so I hope)

You're that faggot op from the 'love is for brainlets' thread arent you? Cut this shit out.

...

>independence
0 correlation means independence for elliptical distros, if your data can be fit as a multivariate normal then means independence, remember to use the proper heterocorrelation measure according to your data. For the indexes you can test linear association simply through a Wald wolowitz run test. Otherwise use a copula model like a weibull/pareto or something, economists love to fit shit to those.

>indexes
Just do multivariate analysis from the index data and see if you can get to replicate said indexes with principal components, this is a crucial step since by definition said indexes would be independent.

>best predictor for happiness
It depends on how you measure happiness: as a 0,1 state then use a logistic regression. Also from the copula model you will have a exact model for the happiness marginal distribution given both indexes, and you can create your own ANOVA test from scratch or whatever, copula models give you nuclear power. Since your friend may not know how awesome are copulas then go for regression, but please, always check if your assumptions are valid.

Just do meme shit like chi squared. Ezpz publishing

Thanks for the only good response in the whole thread. At the moment all I'm doing is chucking data into SPSS, doing a partial correlation and setting control variables as gini index and GDP PPP. Then sending the correlations, p-values and df to friend. She's then taking these correlations, ignoring those with bad p-values and talking about those with interesting correlation factors. Is this sound? Do the control variables require any analysis or can I just shove them in there (the economic numerics).

married guy here. he's right. when my wife was working she was a miserable bitch. after we popped out our first kid she did a complete 180. now she stays at home and just cooks and cleans and she has never been happier.

the modern woman has been rused.