Everything

Businessweek B-School rankings are BS

Fuqua is a great business school. However, in no world would someone rank it above HBS, GSB, Darden or Wharton. So when Bloomberg Businessweek released their 2014 Business School rankings, I was a bit surprised. Not only was Duke #1, but Harvard was #8 and UVA was #20. Holycrap.

Thankfully, they provided their top 85 list in an easy-to-copy-to-Excel format on their website. They list three key variables that factor in to the score (Student Survey, Employer Survey, and Intellectual Capital) that they *implied* factors in to their scoring. It turns out that previous overall ranking also factors heavily in their judgement…more on that later. They also included a helpful “Ranking Index Score” with Duke holding 100 and every school holding a value relative to that.

So I had a simple question. How Much weight does Bloomberg Businessweek give to each of these variables? Unfortunately, coming up with the raw numeric values for the three variables (as opposed to their ranking) seems to be impossible. However, they did give us more than just a final ranking to work with and the Index score will help. I am still forced to do a basic linear regression where the independent variables are rank but it should give us a clear directional indicator on what Businessweek weights.

Spoiler: It’s all about what employers have to say. Education isn’t worth much at all. 

The naive approach:  Simple linear regression with three variables.

Ok, the independent variables are the three mentioned above and the dependent is the index score.

What do we find? Oh wow. With an adjusted R Square value of .97, the biggest variables are the student and employer survey.  More on this later.

A note on the output. The function is setup so you start at 100 (the highest sore) and subtract from it. So, a school ranked #1 in every category would be 100.9182+(1*-.39856)+(1*-.36886)+(1*-.11752). As you work your way down the rank (from 1 to 85) you subtract more and more to come up with the index number. A lower index number means a lower rank.  I have included the whole Excel output for those interested.

Simple BSchool Regression

Approach Two: Including prior survey rank

I assume that these scores aren’t calculated in a vacuum. Did Businessweek take in to account prior rankings? Not every school was ranked in the last survey. I can’t run a regression with holes in the data and I don’t want to put in 0’s for the prior years. So, when there is no prior year information, I am filling in “85” – the lowest rank in this year’s data. It isn’t a perfect approach but I’ll address that in the next analysis.

The results? I’ll save the stats lecture but will say that the previous survey rank isn’t significant using this method.

Simple BSchool Regression with previous

Approach Three: Including prior survey rank, discarding new schools

So I have to re-run the regression and remove schools who haven’t appeared in prior years. I think throwing in 85’s is throwing off the analysis. For the results below, I am only analyzing schools that have a score this year AND in the previous survey. This is kind of cheating but I’m at a loss for a better way to do this given the data I have.

The results? Well, this is interesting. The P-Value for Overall Rank is now in an acceptable range and in fact appears to be significant. The table is below:

Simple BSchool Regression with previous No 0s

So where did we end up?

Well, the final regression formula that I came up with was: 101.7979 + (Previous Rank * -0.19207) + (student survey rank * -0.28878) + (Employer Survey Rank * -0.36018) + (Intellectual Capital Rank * -0.09254)

In a more human readable format – the factors that Bloomberg Businessweek use, in order are:

  1. Employer Survey
  2. Student Survey
  3. Previous Survey Rank
  4. Intellectual Capital

So let’s stop dancing around the obvious. This ranking system is bullshit. Employer surveys are worth >3x the quality of the professors. It also doesn’t account for things like which schools actually did the most to improve their student’s lives. For example, someone who goes to a top 5 school would probably have done alright in life. Someone who went to school 50 probably had a much larger boost in quality of life. And never mind the practical question of “how does one rank intellectual capital?” What this tells me is that there is an outsized emphasis on students finding jobs and a minimal value placed on the quality of the education.

You can have a great philosophical debate on whether schools exist to teach people or to find them jobs and I’m not about to tackle that one without a few more drinks. I think it is fair to say, however, that this goes a long way to explaining why business schools focus on their recruiting and may explain some of the non-obvious admission, academic, and employment things that any top-50 business school student had seen. Skipping classes regularly to go hunt for jobs while paying >$100k for a 2-year education? Check.

 

More coverage:

http://poetsandquants.com/2014/11/11/duke-tops-2014-businessweek-ranking/