Statistical Jousting

Fellow “community blogger” Geoff Caldwell recently posted under the title, “D.B. Graham or Gallup.com, who do you trust?” an interesting chart of Gallup polls on political ideology trends over the past 18 years.  The issue of the post, as best I can determine, was to contest Graham’s (another community blogger a.k.a., the “Erstwhile Conservative“) assertion that the polls on such a subject are an unreliable way to assess that sort of thing because, in Graham’s words, “. . . most people aren’t political junkies . . .”

BTW, I tried to post a comment on the matter, but Geoff’s wordPress software for some reason rejected and erased my 20 minutes of work in a heartbeat, so I resolved to repeat the effort here.  (Geoff, if you are getting fewer comments these days you might want to look into the problem.)

Image via Wikipedia

While Geoff scoffs at the idea that people don’t know what political label they should wear, I find myself in agreement with the E.C. on this, and would add that the data might well be skewed because “liberal” is sometimes used as a pejorative by some in American society, perhaps because of the traditional animus between management (conservative or regressive) and labor (liberal or progressive).  The division is as old as the concept of organized labor and is unlikely to be resolved in our lifetimes.  I think of it

as the necessary dynamic that keeps capitalism from extremes of either direction.

Nevertheless, let’s assume that Geoff is right and that people do define themselves well and  consistently over the years.  If that is the case one could analyze the 19 years of data from the Gallup chart to look for trends.  I note that the conservative and moderate percentages diverge sharply and the liberal percentage falls a little at the 2008 point.  It seems reasonable to attribute this to the Great Recession.  This is consistent with the rise of the Tea Party, a conservative movement which blames the current liberal government for the fiscal problems.

But, there is more.  I processed the data for the first 17 years, leaving out the last two (post-2008) points as non-normal due to the G.R., and got these results:

Function Moderates Conservative Liberal

Mean              39.65               37.53                  19.00

S. D.                   1.45                0.78                    1.75

n                        17                    17                       17

2008 data        37                    37                     22

Delta 2008       -2.65               -0.53                +3.00

The function n is the number of data points, or years.  S.D. is the standard deviation, a measure of data variability.  Delta 2008 is the difference between the average of the 17 years that do not include the G.R. and the values for 2008.

What the analysis tells me is this:

1.  Conservatives change more slowly over time, having dropped only about half of 1% in 17 years.  The standard deviation of the data supports this lesser variability year-to-year compared to the other two labels.

2.  There was a shift going on before the G.R. and during the 17 years of about 2.65% from moderates to liberals, and about 3.00% from conservatives to liberals.

It doesn’t seem too great a leap to me then to assume that the above trends would continue once the G.R. has run its course.  If true, this would bode a more-liberal body politic for the future.  It would also indicate that the E.C.’s conclusions appear to agree with the data after all, Geoff’s emotive rhetoric notwithstanding.

Of course the above would also mean that in about 75 more years Geoff has a 50/50 chance of becoming a Liberal [17 X(37.5-22)/ (3 +0.5)] = 75 .  But don’t worry Geoff.  I think you’re wrong about people’s ability to define themselves politically.