Wednesday, July 31, 2013

Statistics, Inflation, and Higher Education

Statistics, Inflation, Higher Education

By Professor Doom


     I’ve been asked why I keep saying administration only uses retention and growth to decide how to run institutions, and don’t care at all about improving higher education, so before talking about faculty’s hand in destroying higher education, allow me to explain this conclusion.

     My training makes me use statistics to explain this conclusion, in this case a part of statistics called hypothesis testing, which is a method to check when you’re suspicious about something strange happening. The way how hypothesis testing works is first you make an assumption, typically a hypothesis that “what you’ve been told is the truth” as opposed to an alternative hypothesis that “something suspicious is going on.” Then, you look at your data. Next, you calculate the probability that you’d get data like what you see (or anything stranger) if you’ve been told the truth. If that probability is low, you reject what you’ve been told in favor of the alternative. If the probability is not low, then you figure that nothing unusual is going on, and your suspicions are unwarranted. “Low” is a matter of personal choice, but accurately calculating that probability is the core idea of modern statistics.

     There are many ways to calculate the probability, but a crude yet simple way is just to look at it as coin tossing, assuming a 50% chance that each data point will go to either one hypothesis or the other.

     So let’s take an example of hypothesis testing using this very simple method:

     Consider officially reported inflation of 2% (or so, varies a bit every year, but this is basically the number government insists is true). Now, the government gets this number using a fairly arcane method, and they change the rules for how they calculate inflation every few years. The government claims these rules changes are to get a more accurate representation of inflation.

     Maybe the government is telling the truth (stop laughing!), but it really seems to me that prices on average are going up more than 2% every year, so I’m suspicious about these rule changes. My basic hypothesis is the rule changes are as the government says, simply to better represent inflation. Since that’s supposedly the case, half the time the rules changes would move inflation up, and half the time inflation would move down (since the goal of the changes is supposedly accuracy, and not to minimize reporting of “real” inflation).

     Suppose the last eight changes to how inflation is calculated show that all eight changes reduced the reported inflation number. It doesn’t matter what the changes are, but some of them include geometric weighting to goods that aren’t increasing in price, assuming consumers don’t eat or use gasoline, assuming consumers won’t buy goods with increasing prices, assuming consumers get pleasure from increasing prices on improved goods to the point that price theoretically decreases even if it goes up—we’re talking really obscure arguments that are difficult to follow, much less calculate, but that’s the advantage of using a simple statistical method. It doesn’t matter what the changes are, what matters is that every time, the official inflation rate goes down.

     Basically, the coin was tossed 8 times, and came up “heads” eight times in a row. The probability of this happening, assuming the government is telling the truth, is 1 out of 256, or about 0.004. This probability is low enough that I, personally, believe the government is indeed making these rule changes to under-report inflation, and I reject the claim that the changes are only about better accuracy.

     Granted, any yahoo can pick up a pocket calculator, buy some stuff, come back in a year and buy the same stuff, and see with his own eyes that something’s screwy about government inflation numbers, but the previous is loosely how it’s done in a quantifiable way with statistics.

     Ultimately, statistical decisions are a matter of personal choice. 0.004 is low for me, but if you have enough faith in the government, 0.004 might not be a low enough probability for you to say the government is lying. Such people do exist, by the way (they don’t trust statistics, which I can accept, and won’t pick up a calculator and see with their own eyes, which I find bizarre).

     As an aside, statistical tests on whether smoking causes lung problems can get probabilities around 0.00000000000001, but it’s ultimately a personal decision on whether that’s low, which is why tobacco company executives can say with a straight face that they don’t believe there’s a relationship between smoking and lung problems.

    Now, back to higher education. Let’s assume administrators are changing policies just to make higher education better, and that improving retention and growth of the student base has nothing to do with it. Let’s list the changes in the last twenty years I’ve identified so far:

Easier admission (improved retention and growth)

Reduced standards for staying in college (improved retention and growth)

Encouragement of low content courses (improved retention and growth)

Encouragement of the belief that getting a degree is the path to riches (improved retention and growth)

Infinite withdrawal policy (improved retention and growth)

Introduction of remedial coursework to 3rd grade material (improved retention and growth)

Reduction of difficult course requirements (improved retention and growth)

Extensive convenient online offerings (improved retention and growth)

Encouragement of cheating (improved retention and growth)

Encouragement of group projects (improved retention and growth)

Extension of time to complete degrees (improved retention and growth)

Addition of many dubious degrees (improved retention and growth)

Ability to re-take courses that have already been passed (improved retention and growth)

Ability to take the same course several times simultaneously in a semester (improved retention and growth)

Heavy-handed encouragement of faculty to increase retention (improved retention and growth)

Encouraging students to get loan money to the students own detriment (improved retention and growth)

Elimination of mandated advising and course registration for students (improved retention and growth)

Reduction of tenure (no relation to improved retention and growth).


    I may have missed a few, but once again, the coin seems to be coming up heads nearly every time. Heavy reduction of tenure admittedly has no direct effect on retention and growth; I do feel that giving faculty no protection or means to protect higher education has indirectly led to the other policy changes that do affect retention and growth, but perhaps I’m biased. Other factors (such as making student evaluations the primary part of job performance) I haven’t yet discussed in these essays, so I’ll leave them off as well.

    That’s 17 heads out of 18 tosses, so we’re looking at a probability of around 1 in 30,000. This probability is just too low for me to consider that the rules changes are being made in the interest in education. So, I reject that hypothesis, in favor of believing that the administrators of higher education are using retention and growth as their primary means of policymaking.

     Maybe higher education administrators really do have the best interests of education at heart…but the data says otherwise, at least in my opinion on what “low probability” is.

      We’re closing in on August now, the time when people seriously consider going back to college. If you’re looking to go into higher education, realize the odds that administration isn’t going to screw you are about 30,000 to 1 (which, outside of the Ivy League, is close to the proportion of people that come out ahead after entering higher education).

      Still going to go to college? Good luck. I hope you beat the odds.