Need help with statistics software
Karli — Sat, 09/12/2009 - 23:07
I have been asked by my father in law to help him analyze some data for his graduate thesis. Back in undergrad I spent a year as a research assistant and took several statistics courses and did quite well in them, but all we ever used to analyze data was Excel and calculators. I also spent a year as a graduate research assistant, and again did nothing but imput data into spreadsheets and then manually calculated results.
He is analyzing existing data (census information, to be exact) having something to do with correlating where people lived over a 30 year period (sorry, I don't currently know more....let me know if you need to have more info!). The data is being provided by a public entity, and is prepared and then delivered to the end user (i.e. him) via an extractable file that is described as follows:
"The IPUMS extract system does not provide data in the format of any particular statistics package. Instead, we provide compressed data files and the command files necessary for reading the raw data into SPSS, Stata, and SAS."
I know there are open source versions of SPSS available on the web; I had a former co-worker who used that software to do agency quality assurance analysis and have a general understanding of how it works. Beings that I like finding random reasons to teach myself new software and computing skills, I am not too intimidated by the idea of obtaining this software and using it to perform the analysis he needs.
I don't, however, know enough about the various statistical analysis software that is out there currently, and types of files they support. My specific questions are these:
- If anyone has ever used SPSS and/or the freeware version of SPSS (http://en.wikipedia.org/wiki/PSPP), do you have a preference? My Father-in-Law can get a student version of SPSS for $100 or $200 or so, but free is such a nice price tag..
- All things aside, the freeware version does me no good if the raw data can't be read by it. Based on the blurb from the company, do you think this will be a problem? My guess is no, but obviously I'm not sure and some clarity would be nice.
- And since I HAVE to ask, I am a huge Excel dork....anyone know if this data could be manipulated in Excel? I'm very curious.
Thanks in advance!!!
I'm sure Scaron is probaby
Dragon — Sun, 09/13/2009 - 11:07I'm sure Scaron is probaby the best person atm judging by the two responses thus far. If you do need any other help, all the computer labs (well most of the computers in them) have both SPSS and SAS on them and I can dork around on em sometime this week if you need.
I have what I would consider
Scaron — Sun, 09/13/2009 - 10:54I have what I would consider to be extensive experience with SPSS (which is now, as of the current release being renamed to PASW). I have not used the freeware version.
A couple notes:
SPSS probably is the best, most intuitive, solution for the casual number cruncher. They're also a bunch of corporate fucktards, but that's another story. SPSS with it's menu (rather than syntax) driven GUI is quite intuitive. I don't know anything about the freeware version, but I can tell you that you'll be able to do what you need to do with SPSS.
STATA and SAS are less user-friendly (although generally more powerful) solutions. Don't quote me on this, but they also are substantially more expensive options.
If your father-in-law opts to go with SPSS, I would make one suggestion. Instead of buying the student version he should go for the grad-pack. There are a couple of key differences. First, often the student edition has limits on the number of variables (last I checked you were limited to 50) and cases (again, this probably has changed but last student version I had could only handle 1500 cases...in the case of census information, this is not likely to cut the proverbial mustard.) The grad-pack is more money (by like about 100 bucks) but in my mind is well worth it...also if you buy the PC version it comes with AMOS (path modelling software). All with a 4-year license.
Most statistics CAN be calculated in excel, however they often are not accompanied by some of the more common reporting features one would find in SPSS outputs (things like degrees of freedom, precise p-values, effect size estimates, etc) which will be required for a APA presentation of findings. All that said, you most likely will be able to, quite easily, calculate the correlations you're looking for in excel. Try it first and if you get everything you need you just saved yourself 200 bucks.
Finally, if you have questions about anything, I'm happy to help. In my RL life I teach the statistical analysis of social data to grad students, so if there are any rough spots you run across, I may be able to lend a hand.
Wow, couldn't have asked for
Karli — Sun, 09/13/2009 - 11:28Wow, couldn't have asked for a better response. I don't know how this guild manages to do it....every time I have come to the forums with a random tech-geek question I end up getting expert help, so THANK YOU!!!!
I forwarded your suggestions to my father-in-law and more or less recommended that he pony up for the grad student version. I have always been very good at math and I definitely spent more time taking stats courses and doing research assisting than the average person who got my degree, but it's been 5 years since I left school and more than that since I've really done major data analysis. I did do a lot of it at work, but not anything that had to really be checked for a whole lot of integrity, ha ha. I would administer confidential surveys where people could report how they were feeling about various aspect of their jobs and report it to upper management. I generally did this by using google docs (forms) to create surveys, and then I would import the responses (which automatically go in to a spreadsheet) in to excel. Then I would create formulas to change the responses to numeric values, and manipulate the data from there.
Anyways, the next steps are for him to get me access to the software. I suggested he purchase it so that he can get a downloadable version and then gets a hard copy mailed to him. That should save us some time. I have the data saved on my computer now in the SPSS format, so once I get the software installed I should be able to load it in there to analyze.
I'm all for easy user interfaces (good sell, by the way!), and I am about 99% sure that he's going to need to present his data to APA standards and I know in the end it will save me a lot of time and headache to just let the software work its magic.
I'm sure once I get the software up and running and begin to do the analysis, I'll have more questions, if nothing more than to make sure I'm doing everything correctly. I don't want to screw this up!
I really appreciate your help an expertise, thanks again!!!
No problem. I'm happy to
Scaron — Sun, 09/13/2009 - 11:30No problem. I'm happy to help. Hit me up with a pm or whatever if need anything.
I've never used SPSS or
Dufe — Sun, 09/13/2009 - 08:45I've never used SPSS or PSPP. I did use Stata extensively in school, and it had some options to export stuff to excel. Is this a one-time thing? Or will this need to be done repeatedly?
It's a one time thing. It's
Karli — Sun, 09/13/2009 - 11:09It's a one time thing. It's for his Master's Thesis. He is focus is on contemporary american history, and he HATES math so he's enlisted my help =).
The data analysis I did was definitely limited, but I like to periodically take on random challenges, as I find that just being willing to stare something in the face and think creatively to problem solve often garners me new skill sets. Good or bad, I'm self taught in Windows Server 2000/2003 management, and website maintenance. I've never used Dreamweaver or any sort of interface software; it's just straight up HTML coding. My methods probably take longer than they need to, and at some point if something really grabs my intereste I'll probably take a class on it, but I have had a high degree of success just tackling challenges like these head on. It will be interesting, to say the least!