Why Tim Ball is Wrong

(c) 2007 by Barton Paul Levenson



Dr. Timothy Ball is a retired (since 1996) University of Winnipeg professor of geography, though he lists himself as "Emeritus Professor of Climatology." He is also an often-quoted global warming skeptic. One of his most famous lines is that "[T]he global temperature has declined since 1998 while human addition and levels of CO2 continue to rise." Deniers often summarize this quote as "Global warming stopped in 1998!"

Let's examine this assertion.

First of all, 1998 was an exceptionally hot year because it was an El Nino year. The choice of dating from 1998, when we actually have more than 120 years of time series data available, is called "cherry-picking," and is considered a mistake in data analysis. You can't pick out part of the data that seems to support your hypothesis, you have to use all of the data. Dr. Ball must have taken some kind of course in statistics in his years as a scientist; he must know that basing a trend on nine years of data when 120 years are available is a beginner's mistake that would get him a flunking grade in any introductory data analysis class. But let's allow his cherry-picked start date and examine the numbers.

Here are the mean global annual temperature anomalies (in hundredths of a degree) for the years 1998 through 2006, according to NASA's Goddard Institute for Space Studies:



199871
199946
200041
200157
200268
200367
200460
200576
200665


I entered these columns into a Microsoft Excel spreadsheet and ran a linear regression of the anomalies on the years. As expected from the small sample size, the regression was statistically insignificant (p = 0.22). But -- here's the kicker -- what trend seems to exist in the data is UP, not DOWN. The regression equation is:

Anom = -3742.577778 + 1.9 Year

I.e., on average, the temperature anomaly increased by 0.019 degrees K. each year in the period discussed. This brings up another point about statistical analysis -- you can't tell a trend from drawing a line from the starting point to the end point. That gives too much weight to "outliers" instead of weighing all the points equally. What you do is what I did -- run a linear regression. And if you do that, Dr. Ball's idea turns out to be wrong. The trend is up. Not in a statistically significant way, though it becomes very significant indeed if you run the regression from, say, 1880 to 2006. But up, not down. So global warming did not stop in 1998, and Dr. Ball, and anyone else who uses the line "Global warming stopped in 1998!" is WRONG.



Page created:09/26/2007
Last modified:  02/01/2011
Author:BPL