Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

New jobs report – Journalists and statistics

July 6th, 2012, 2:10pm by Sam Wang

It’s unfortunate that many journalists lack good intuitions about how to look at data. Here’s the latest example: via Andrew Sullivan, a claim that jobs are on a downward trend based on this chart. Can you see the problem?Jobs chart - do you see a trend?
Here’s the answer:
the data fluctuate up and down, and it could be random. This jobs report is likely to have some imprecision, which is not reported. That’s too bad because this “trend,” as claimed by Matt Yglesias, might well be followed by a jump up. Take a look at other times, in the same chart, where a high report is followed by three lows in a row. I have indicated them with red arrows.
Other times with three low reports in a row are indicated with red arrows. Look what happens next...up or down?
But look what happens in the following month! A high report, a low report, and a high report. In other words, there is no predicting what will happen next based on data of this quality.

The point I am making could be made in an up-front manner by including a measure of the uncertainty. In this case, basic training on how to look at data could also help.

Update: a far better chart is here at Kevin Drum’s always-good blog.

Tags: 2012 Election · Politics

2 Comments so far ↓

  • Robert Vanderbei

    Even more important: this is a plot of the derivative. If the change is positive, then the trend is upward. Any positive value is good. The chart only shows that the second derivative has decreased. That’s not the same thing as a downward trend.

  • Sam Wang

    Good point. Which means that jobs have been added every month since late 2010.

    Do we also need to correct for the rate at which the working-age population grows? That would be about 200,000 persons per month. In which case I would score this overall trend as sputtering, but no more so than a year ago.

Leave a Comment