Steve Sachs Duke


Sunday, June 22, 2003


What I've been working on: For the second year in a row, all students in Philosophy, Politics, and Economics at Oxford have been required to complete an "IT Project" using statistical software. It's an absurd requirement, and academically indefensible. Not for the reasons that most are complaining about, that few students will ever need to perform a statistical analysis for their classes; it's useful for any educated person to understand a statistical argument, recognize common fallacies and know how to test for statistical significance. (I say this as someone who's pretty ignorant about the field and would appreciate learning more.)

Instead, the real problem is that the requirement doesn't teach anything about statistics. It only teaches how to use SPSS, a specific statistics package allegedly popular among social scientists. What's worse, the requirement doesn't even teach students how to use SPSS very well, since most of the program's commands require a detailed understanding of the statistical functions involved. (Quick--what's a "One-Way ANOVA"?) So all that we can do--and all that we're graded on, it seems--is try to fill our papers with pretty graphs. It's a classic example of a good idea filtered through enough Oxford bureaucracy to render it meaningless.

In any case, faced with the requirement, I tried to make something useful of it. Comparing the membership of the U.N. Commission on Human Rights and historical data on political and civil liberties from Freedom House, I wrote a short report entitled "Sins of Commmission: Repressive Regimes and the U.N. Commission on Human Rights." Its basic conclusions are these:

  • In the past three years, regimes with the worst or near-worst human rights records have been substantially over-represented on the Commission, a trend that is statistically significant and in certain ways historically unprecedented.
  • This over-representation cannot be explained by the Commission's geographic distribution requirements, and holds true even within the geographic regions used for selection.
  • There is no statistically significant evidence that Commission membership causes regimes to improve their behavior over time. In fact, membership may retard progess by sheltering regimes from the Commission's scrutiny.

None of this is particularly earth-shattering, but some may find it interesting. The zipped data files are also available for others to make better use of.




Blog Archives

Front page
XML Feed


© 2011 Stephen E. Sachs


Anglia Regnum