Why do social scientists use least squares linear regression for everything, even when it makes pretty much no sense?

Why do social scientists use least squares linear regression for everything, even when it makes pretty much no sense?

Other urls found in this thread:

heterodoxacademy.org/2016/01/07/new-study-finds-conservative-social-psychologists/
scientificamerican.com/article/is-social-science-politically-biased/
heterodoxacademy.org/2015/09/14/bbs-paper-on-lack-of-political-diversity/
youtube.com/watch?v=1YNjMxxXO-E
twitter.com/SFWRedditGifs

Because it's propaganda

They also cherrypick countries (in your picture)

>When all you have is a hammer...

But seriously, I think it's because they aren't interested in some complex model, just whether [math] a \propto b [/math].

>just whether a∝b
you wish

they set to prove that a∝b and cherrypick the countries to do so

>they set to prove that a∝b
I believe that's called a hypothesis; suppose [math] a \propto b [/math], then investigate it.

>cherrypick the countries to do so
P-value fishing is a problem in more than just the social sciences.

they set to prove, in the literal sense

they will do anything to prove their ideology right

Can you prove that, or have you already assumed your hypothesis is true?

I have read many papers, at least the kind that OP posted,

So you have "right wing" social scientists, who will publish papers that prove their ideas right.

And then you have "left wing" social scientists, who will publish papers that prove their ideas right.

Actual studies for the sake of truth are rare.

>I have read many papers, at least the kind that OP posted

Sounds like selection bias to me.

Some of them are more sophisticated.

>foreword by deepak chopra

>I have read many papers
So that would be anecdotal evidence. Which we all know is a very weak form of evidence. Do you have any quantitative data, and empirical evidence?

And I hope that by this point you do see the irony in your argument.

There's no irony, because I am not claiming this is scientific. I am just sharing my experience.

Anecdotal evidence is a form of empirical evidence

It's a different kind of quantumbabble. It contains chapters like "Non-classical decision making: violation of the LTP (law of total
probability) and the quantum Bayesian approach" or "The wave function and non-observed state prices".

>I am just sharing my experience.
If that were the case than you would have started this whole thing off with it. In stead you presented a statement as fact.

I did when I said that my opinion came from observation.

But it's non-quantitative.
And so here we are, discussing your opinion. I feel differently than you do.

Do you see now that we're getting nowhere with this line of argumentation?

how is violating a mathematical axiom quantum mechanics?
fuck off autismo

>fuck off autismo
stay classy Veeky Forums

What's wrong with qualitative data?

>violating axiom
more like not picking that set of axioms

>What's wrong with qualitative data?
It doesn't allow for quantitative statistics.

Didn't you even look at the picture?

I did, but after I replied. You make a fair point. I guess I had my definitions mixed up.

>I feel differently than you do.
Tell me about your experience then

Do you think that most social science studies are for the sake of truth, and that they don't discard contradictory results, that they don't cherrypick, that they don't twist facts?

Not him but, you made me questioned my assumption of a significant political bias in social sciences because you're right that I based this opinion on anecdotal evidence. So I looked into the research and the evidence strongly suggest a left wing bias. I'm wondering how you'll respond

heterodoxacademy.org/2016/01/07/new-study-finds-conservative-social-psychologists/

>305 of the 322 people (94.7%) who responded to this question voted for Obama, 4 (1.2%) voted for Romney, and 13 (4.0%) said they voted for another candidate. This gives us a Democrat to Republican ratio of 76 to one.

scientificamerican.com/article/is-social-science-politically-biased/

>58 to 66 percent of social scientists are liberal and only 5 to 8 percent conservative and that there are eight Democrats for every Republican

heterodoxacademy.org/2015/09/14/bbs-paper-on-lack-of-political-diversity/

He's wrong. Your intuition was correct

Of course, one could argue that due to their studies, they identify their positions as the correct ones. Anyone know if people who decide to study social sciences are disproportionaly liberal prior to starting it?

Nothing you've posted supports the hypothesis that their research is biased, all you've done is show that the overwhelming majority of social scientists are left wing (although the idea that America has a left wing is hilarious to anyone outside the the US, but that's a different discussion). The only way you could go from what you have to "social scientists only want to prove their own ideology correct" would be to assume that any social research is biased towards the authors political persuasion.

>image

Are you like 10?

>waaahhh why won't non-mathematicians care about my mental masturbation????

OP, I agree that it's silly. Even more fundamentally, why do social scientists (or any scientists, for that matter), assume that any possible quantity of interest can be represented by a real number? Looking at your graph, what does it mean to say a country has "Income Inequality .2"? Yes, I know people have dreamt up measures like the Gini coefficient and the like, but it's a flagrant epistemological blunder to conflate an artificial statistic with "Income Inequality". Not only is it intellectually lazy, but it also leads to nonsensical headlines like "Study shows that eating avocados makes you happier".

Because they aren't real scientists.

Because typing x=(A^t*A)\A^t*b is easy as fuck

youtube.com/watch?v=1YNjMxxXO-E

>Looking at your graph, what does it mean to say a country has "Income Inequality .2"
If you look closer, the "Income Inequality" axis doesn't even have numbers on it, it's just "low-high".