Similar to Brandon I’ve began playing around with new statistical packages. Like many libertarian scholars I have my skepticism about the limits of what we can learn from number crunching. I think there is a place for statistical analysis in the social sciences, but it is definitely meant to be a tool, not an ends to itself, and should be complemented with additional methods.
Recently I’ve begun trying to find a Geographical Information Systems (GIS). I had initially intended to buy a copy of ArcGIS, one of the dominant GIS packages, until I looked at their pricing plans. A single license for the basic version costs $1,500 USD. I’m sad to say this price tag is not abnormal. STATA, one of the larger statistical packages, sells an annual licence for its bare bones version at $125 USD. SAS has its pro version going for $9,000 USD.
What is abnormal is that several freeware packages exist that provide comparable services. Are you an undergraduate student taking a class on univariate regression analysis? Download Gretl. It has a menu based system that is relatively easy for even the newest of users to play around with. If you’re looking to challenge yourself opt instead for R.
Likewise, for those who like me are on a budget, there exists several freeware alternatives for GIS systems such as GRASS and QGIS. I’m still learning GIS so I can’t comment on either package, but I will be sure to provide reviews once I’m comfortable with them.
If several freeware alternatives exist, why do retail versions remain dominant in the industry?
Part of the answer is that corporations and universities value the customer help hotline if their software starts to malfunction. Poor graduate students don’t have much money, but tend to have a surplus of free time to use trying to figure out why their software isn’t working. Corporations have the opposite constraints, they have infinitely more money than graduate students but have much stricter time constraints.
Surely that can’t explain it all though, can it? If what you are purchasing with retail packages is the customer hotline, why haven’t a group of entrepreneurial (and hungry) grad students set up a business where they provide dedicated IT support for freeware? Several attempts have been made by Linux enthusiasts to provide such services for corporations looking to replace their Microsoft OS systems, so the idea has surely been thought of before.
Another possible answer is that what these retail packages are selling is their community. STATA may not be so technically superior to Gretl, but the former’s community is larger than the latter. If you have a problem with Gretl you can’t easily find another user to help out outside of a few niche forums. Meanwhile you are sure to find a STATA compatriot just by walking down a social science college’s halls. I am not really convinced by this idea though. There is a value to joining an existing community, but in the long run people do move across networks. Consider Myspace, which less than a decade ago was the social network, until it was defeated by another social network. How much longer will STATA and ArcGIS last before its user base migrate to R and GRASS?
What do you all think? What other reasons might explain why pricey retail statistical packages remain dominant over comparable freeware alternatives?