News for hackers. Stuff that matters.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by Perceval on Saturday January 09, 2010 @02:43PM (#1041870)
I'm trying to teach myself statistics (for social sciences) right now. I got two books (Agresti & Finlay along with Knoke, Bohrnstedt & Mee) but both seem to be written in jargon instead of English.

Does anyone have recommendations for statistics books that are actually written in English?

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by kqr2 on Saturday January 09, 2010 @02:55PM (#1041898)
Zed actually recommends quite a few books in his post:

    * Statistics; by Freedman, Pisani, Purves, and Adhikari. Norton publishers.
    * Introductory Statistics with R; by Dalgaard. Springer publishers.
    * Statistical Computing: An Introduction to Data Analysis using S-Plus; by Crawley. Wiley publishers.
    * Statistical Process Control; by Grant, Leavenworth. McGraw-Hill publishers.
    * Statistical Methods for the Social Sciences; by Agresti, Finlay. Prentice-Hall publishers.
    * Methods of Social Research; by Baily. Free Press publishers.
    * Modern Applied Statistics with S-PLUS; by Venables, Ripley. Springer publishers.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by thomaspaine on Saturday January 09, 2010 @03:02PM (#1041909)
Funny you should ask, I just ordered a copy of a stats textbook I used back in undergrad because I remember thinking that it was pretty clearly written with lots of examples. It's "Probability and Statistics for Engineers and Scientists" by walpole . If you get a used 7th edition instead of the 8th, it's $10 instead of like $100. http://www.amazon.com/Probability-Statistics-Engineers-Scien...

Introductory Econometrics by Woolridge is a pretty readable econometrics text you should also check out, since it'll be especially applicable for social science: http://www.amazon.com/Introductory-Econometrics-Approach-App...

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by bokonist on Saturday January 09, 2010 @02:59PM (#1041904)
For a long time, I wanted to teach myself statistics for the social sciences too. I decided the best way to do this was to find an interesting social science issue, and then learn statistics as part of tackling the issue. But there was one problem with this. There are almost no problems in the social science where it makes sense to use advanced statistics to analyze them. Using very simple regressions can be illuminating and illustrative (though not conclusive). But when you try to get more advanced, data quality, data coding, issues with choosing the right controls, too many variables, sensitivity to variable selection, inability to separate causation from correlation, etc. render advanced statistics utterly useless. Do you have a particular social science problem that you think would be a good fit for statistics?
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by thomaspaine on Saturday January 09, 2010 @03:17PM (#1041927)
I guess it depends by what you mean by "advanced statistics". Is correcting for heteroskedasticity, a fixed-effects regression, instrumental variable regression, etc advanced? Because if so, all of that is pretty common practice in say, economics. Simple linear regression is easy and often illuminating, but good luck getting a paper published if that's all you've done.

I agree that in practice, data munging is 90% of the work and can be pretty tedious and discouraging when working with real world datasets, but if you don't know about issues like the ones I mentioned above, how do you even know where to start?

For examples of how statistics can be used to solve interesting problems in the social sciences, just browse through Freakonomics.

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by bokonist on Saturday January 09, 2010 @04:16PM (#1041998)
Yes, that's what I mean by advanced statistics. I've read many papers in economics using such techniques. They are 99.9% utter garbage. The issues I mentioned above (data quality, data coding, issues with choosing the right controls, too many variables, sensitivity to variable selection) doom the project before you can even get to using the statistics. Garbage in, garbage out. If you data munge and get a result, you can never determine if it's a real result or a data mining effect.

And yes, I understand you cannot get published in the social sciences without using advanced statistics. All this means is that the economics academia is following into the trap of Schoclastism, and insular world that uses complicated and absurd methods of research, increasingly divorced from reality.

Social science needs to get over its science envy. As I've written here before, the proper tools to use as a student of policy and sociology are the tools of product management: http://news.ycombinator.com/item?id=836196

I have not read Freakonomics the book, but I read their blog occasionally. It's interesting when they find a simple statistic that raises an interesting point.

It might be easier to explain the problems with reference to a specific example. If there is some social science paper that uses the techniques you mentioned above, and you think is particularly good, send me a link, and I can explain why I think it's likely garbage.

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by Perceval on Saturday January 09, 2010 @07:07PM (#1042246)
I agree with your assessment of the 'scientism' of social sciences, and the rather absurd uses of statistical methods (that should be inapplicable because of data collective and basic assumption issues).

What I am to do is mostly descriptive statistics, but with a few simple tests (e.g. a test of variance with covariance). I've got three populations of civil war data and would like to compare features of those populations against each other (e.g. instance of civil war in period one vs. period two vs. period three). Controlling for # of states, or # of new states (say, within five years of their founding).

Most of the complicated stuff really can't be applied to stuff like data on civil wars, since the data doesn't meet basic assumptions of the models, but I would like to be able to say something more than just 'the data from period one looks different than the data from the other two periods.'

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by thomaspaine on Sunday January 10, 2010 @12:08AM (#1042689)
> Most of the complicated stuff really can't be applied to stuff like data on civil wars, since the data doesn't meet basic assumptions of the models

I think this is precisely why we have more advanced statistical techniques. There are ways to correct for or at least detect serial correlation, homoskedasticity, etc, all of which are probably in your data. All real world data is fucked up in some way, having a large toolbox of statistical techniques helps you cope with this fact. Maybe not perfectly, but at least you'll know where/why your model is wrong.

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by Perceval on Monday January 11, 2010 @10:27AM (#1045213)
Can you recommend a book on advanced statistical techniques or non-parametric statistics that's written mostly in English with a minimum of jargon?
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by foldr on Sunday January 10, 2010 @12:57PM (#1043366)
> All this means is that the economics academia is following into the trap of Schoclastism, and insular world that uses complicated and absurd methods of research, increasingly divorced from reality.

That's being a little unfair on the scholastics. They did a lot of important work (particularly on logic), and the renaissance didn't just spring into existence out of nothing.

Re:Programmers Need To Learn Statistics Or I Will Kill Them All by hadley on Sunday January 10, 2010 @09:55AM (#1043116)
Sorry, that's rubbish. It's the "advanced" statistical techniques that help you discover and control for these problems. Sure, a lot of people apply them incorrectly and without really thinking about what they're doing and what their results mean (Sod's law: 99% of everything is crap), but that doesn't invalidate their use.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by bokonist on Monday January 11, 2010 @11:30AM (#1045378)
As I've asked for twice already, give me an example.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by jbl on Saturday January 09, 2010 @02:53PM (#1041891)
The Cartoon Guide to Statistics by Gonick and Smith is actually quite good for dipping your toes in. It should give you an okay conceptual foundation from which you can tackle the textbooks you've already got.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by silentbicycle on Saturday January 09, 2010 @07:01PM (#1042233)
I was just going to recommend this, too. It's hokey as hell, but it's actually a great intro -- After reading it, you'll know the vocabulary and fundamentals well enough to ask meaningful questions. It's also cheap, and a quick read. I just loaned my copy to my little brother for his stats class, and I'm betting he'll find it more useful than the dreadful (and expensive) intro textbook his professor chose.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by hadley on Sunday January 10, 2010 @09:52AM (#1043114)
I'd second the recommendation. But avoid the manga guide to statistics - it's nowhere near as good.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by pibefision on Saturday January 09, 2010 @02:52PM (#1041888)
Mario Triola's book is excellent. Covers all the topics with very clear explanations. Elementary Statics 10th Edition.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by tel on Sunday January 10, 2010 @01:01PM (#1043370)
I can't advise books, but I want to add that beyond the vanilla statistics often taught you'll also probably want to read a book on probability theory and/or combinatorics. These along with some basic calculus are the important underpinning maths that let you build statistics.
Re:Programmers Need To Learn Statistics Or I Will Kill Them All by korch on Saturday January 09, 2010 @07:06PM (#1042243)
If you're of the mathematical school that thinks the best way to learn any math field is to walk through the history of how we got here, then I'd recommend the recent book "The Lady Tasting Tea". It traces through the history of the great mathematical minds who created the field of statistics, explains the problems they were solving, how and why they were able to solve them, and how future problems and solutions built up over time into the field of statistics.

It is an academic book, and is more of a history book that explains the math, rather than a math text book; which is why I found it note worthy. It's also worth reading solely for the detailed citations, from which you could completely learn the core of the field by "studying the master's" original papers.

http://www.amazon.com/Lady-Tasting-Tea-Statistics-Revolution...