# What is "Data Analysis" specifically?September 2, 2008 8:10 AM   Subscribe

What exactly is "Data Analysis"? I wish to take an introduction to statistics course, but I don't know which flavour to take (there are several).

All of the courses have the same lectures, but will break up partway through and use different examples (from life sciences, political science, medicine, social sciences, etc).

I'm currently studying history, but there is no Intro to Statistics for historians. I had thought to take the course designed for social scientists (the closest to my area of study), but there is also a version which is called "Data Analysis", and the description is "An introduction to probability and statistics with emphasis on data analysis". And I realised I don't know exactly what they mean by "data analysis" (is it a specific discipline?).

What exactly is "Data Analysis" in this context?
posted by jb to Education (11 answers total) 3 users marked this as a favorite

Basically, use of statistical techniques to cope with data, usually with the aid of computer software. To verify, I googled "data analysis statistics difference"; this resonates with my understanding.

Take that version -- which I am assuming has no prereq -- if you want to be shortchanged a little on basic techniques, but pick up skills as to how to work with a bunch of data you might collect and want to manipulate.
posted by Clyde Mnestra at 8:29 AM on September 2, 2008

Is it offered or cross-listed in departments other than math? That would be a clue. My best (but uneducated) guess is that "data" refers to experimental data in the sciences.
posted by grobstein at 8:30 AM on September 2, 2008

Clyde, the reason this question is tricky and I don't think your answer is a slam-dunk is that all statistics is about "data analysis" and "cop[ing] with data," at least for a reasonably broad understanding of "data." So the poster wants to know what distinguishes the data analysis done in this course.
posted by grobstein at 8:33 AM on September 2, 2008

Grobstein is right.

When I say that I am analyzing data, I mean that the data that I am using is collected, in a stats program and now I am running different statistics on it. If the class is for social scientists, you'll probably learn correlation, regression, t-tests, and ANOVA. You won't look at probability (which seems to be a popular topic in "straight" stats classes).

It probably won't hurt you to take this class, although I'm not sure why you'd need to learn this stuff as a historian.
posted by k8t at 8:38 AM on September 2, 2008

Data analysis is incredible wide. It really encompasses start-to-finish everything from dealing with raw numbers from whatever source to drawing valid conclusions. The only person who can say what's going to be in there (and which class you should take) is the faculty in that department, specifically the person teaching it. Email her. The people in your department might have good ideas about the kinds of problems that a historian runs into.
posted by a robot made out of meat at 8:40 AM on September 2, 2008

Clyde, the reason this question is tricky and I don't think your answer is a slam-dunk is that all statistics is about "data analysis" and "cop[ing] with data," at least for a reasonably broad understanding of "data." So the poster wants to know what distinguishes the data analysis done in this course.

I make no pretensions to a slam dunk, and I think there are matters of degree. But pure statistics, to my distant recollection, is more about the meaning and calculation of certain measures, and data analysis is more about what to do with a mound of data about a certain situation -- nowadays, with a computer package.

Similar, sure, but I think of it like the difference between biology and agriculture. Others will know more.
posted by Clyde Mnestra at 8:41 AM on September 2, 2008

Response by poster: I need to know this stuff as a historian because history does involve quantitative data; I may also not always be in history, so I would like to take a course which gives me a good broad introduction, but one more useful for studying social trends than, let's say, elections or cell multiplication or something.

Basically, all of the courses share one Introduction to statistics lecture, and the sections are broken up into different things. None of the flavours of the course have appear to have pre-requisites, and I will likely be auditting.

My current use of quantitative data isn't even really statistics - I look at population trends (are they going up, are they going down?). I also look at prevalences of things - how much more prevalent is something in one region than another (do they have more cows, or fewer cows?). I would like to ask questions like: are people in region a more likely to have more cows because they are in that region (ie controlling for general wealth)? I was thinking of just dividing the number of cows by the amount of reported wealth, which seemed to be a good way of getting at the question - if it's easier for poorer people to have cows and have more cows in region A than region B (as has been claimed), then they should have more cows per £ of wealth. I realise this isn't really an orthodox way of looking at things, which is why I'm looking at this course - also because I don't really understand what "correlation, regression, t-tests, and ANOVA" are.

As to the future, I don't know where I will be going, but I will either be in history (asking similar questions) or I would like to work in public policy or social research - and I'm thinking about this course with an eye to being more employable.

I just don't understand what the difference between the two versions of the course (social sciences vs data analysis) might be. I'm going to the introduction lecture today - perhaps that will clear things up a bit.
posted by jb at 9:13 AM on September 2, 2008

This answer may be a bit wider in scope than the question calls for, but here goes.

Statistics, as a tool, is going to be difficult to integrate into your knowledge without a series of supporting bits of knowledge (packaged, usually, as "Methods" courses). Now, if you are more quick-witted than me (not hard) you may be able to overlook those gaps in knowledge, but be assured that there will be gaps.

Without seeing the "Data Analysis" syllabus, I can't say what that course offers. What I would suggest is that the Stats courses offered by the Social Studies College/Department/Division would probably offer the benefit of using a lexicon you would probably understand.

Keep in mind that many statistics courses will require you to actually input data into a computer program and interpret the output, but that the practice in doing this is actually less valuable than your ability to understand these interpretations in other people's work. Doing this in a Social Studies environment may help you tease out what Social Scientists care about (and, thusly, present) in research.
posted by Hypnotic Chick at 9:32 AM on September 2, 2008

From my experience (I have a degree communication & media and worked in media research for a while) the first course will be a basic introduction to statistics with a focus of math and the second will focus on collecting and managing data samples and analyzing this data with certain software (most likely SPSS or R). While you can do the 2nd one with the basic highschool knowledge about math (as long as you know at what to look in the results) you'll never be particularily good in data analysis as you lack the theory background and can't do anything more than follow certain patterns.

If you really want to become good at it - take a basic and advanced statistics class, we had them offered from the sociology and also the economics department, they usually are more beginner friendly than the "real" statistics classes. Then take 1-2 classes in a social science methods course where you learn how to do a content analysis or statistical survey, this will also teach you how to collect data and put it into an analysis friendly form. Last step - take 1-3 courses on real data analysis, where you learn how to use the common software, how to run the different tests and how to interpret the results.

An awful lot of people in my major only took the basic statistic class (which they hardly passed) and then the basic data analysis class (which usually doesn't show you anything you couldn't also do with Excel). You can imagine, that the skills you get from that are hardly sufficient if you want to really work with statistical data later.
posted by starzero at 9:38 AM on September 2, 2008 [1 favorite]

My current use of quantitative data isn't even really statistics ... I realise this isn't really an orthodox way of looking at things, which is why I'm looking at this course - also because I don't really understand what "correlation, regression, t-tests, and ANOVA" are.

You are already doing "stats" with your data, it's just that you aren't aware of all of the assumptions and uncertainties in your data. It's to your credit that you appreciate that. A good stats background is the best tools you'll have for understanding populations and changes in populations.

Data analysis is a very generic term, rather like "language arts" or "composition". I would suggest that you talk to each of the prospective profs to get a sense of the course content.

The courses are probably very similar in coverage, but intended for different audiences. At a guess, the first sounds like a service course for the social sciences, the second sounds more mathematical, and more likely for math/stats students. At a guess, you sound more comfortable with the stats for social sciences course, rather than the data analysis course.

Talk to the lecturers before you decide.

Btw, even if you audit, do the assignments. Stats, in particular, is a practical art, best learned by working through examples.
posted by bonehead at 11:03 AM on September 2, 2008

I teach a "Data Analysis" course in a major research university, and I will repeat exactly what everyone else told you.

You seem to know what you need to learn in the class: probabilities, distributions, comparison of groups (t, z, p, and F tests), ANOVA, correlation, and regression. I would only add that some exposure to non-parametric methods (and in particular, contingency table analysis) would also be useful given what you described above.

Do NOT choose the class based on the name. In a lot of universities, there is a lot of bureaucracy involved to change the name and description of the class as it is listed in the catalog. So, professors make somewhat drastic changes to classes without modifying how it is listed. I currently teach my "Data Analysis for Managers" class primarily as a time-series forecasting class since that's what I believe my students will use the most. The previous professor that I took the class over from used to teach it primarily as a quality control class because that was his background and focus. Yet, the catalog description would make you think that you are going to see some of the things I talked about above.

So, ask for a syllabus or a table of contents for the modules in the courses you are evaluating. Try to see which textbook they use and look at the table of contents of that book. Email the lecturer, put in the explanation you mentioned above and ask if they think the class is appropriate for you.
posted by tuxster at 11:57 AM on September 2, 2008

« Older News Parents Can Use?   |   Why should I not run after lasik? Newer »