Visualizing data
There’s a graph from the office of House Speaker Nancy Pelosi making the rounds. It compares job losses from the current recession to the two previous ones, and it looks pretty alarming.
However, I find the chart a little misleading for a couple of reasons. (I’m not saying we’re not having economic problems, I’m just saying this chart may be exaggerating the contrast between the current problems and previous ones.)
There are two issues with the chart. First, the scale is in absolute numbers of jobs, rather than a percentage change. Since the total number of jobs at the peak was different in the different recessions, rather than saying “We lost 2.5 million jobs in 2001 and we’ve lost 3.5 million so far”, we need to express that as a percentage of the total jobs there were to start with. A blogger at time.com did a better job, with % changes and more historical data.
The second issue is a bit more subtle, however. The data are expressed as the “job losses relative to the peak month”. OK. But aren’t the peaks just as anomalous as the valleys, in some sense? If we had really high employment during a good period, that’s great. But I’d really rather compare the % losses to the annual average of the previous 12 months, or something to that effect, rather than simply to the peak month, which may have seasonal effects and other factors at work. (I do confess that I don’t know a lot about the underlying data for these charts and how it’s collected, which may have a large influence on whether we really care or trust a number like that, however.)
# 02.10.09 @ 4:03 pm | category: world | no comments »
