Posts Tagged ‘statistics’

Franken Does Truth a Service

Thursday, 8 October 2009
Franken gets testy over statistics by Eric Roper of the Star Tribune

The senator spent the bulk of his time attempting to debunk the witness, particularly a statistic in his testimony that employees have a 63 percent chance of prevailing in arbitration compared to 43 percent in litigation.


De Bernardo eventually conceded that he did not know whether $50 would be considered "prevailing" in the statistic,[…].

While Franken has at times resorted to worse intellectual dishonesty than in this case he exposes, here he is right on the mark. As Franken's line of questioning and the answer that it elicts show, the statistic in question tells us virtually nothing about whether the outcomes of arbitration would be considered equally or more favorable to employees than are the outcomes of litigation. It is, in other words, a garbage statistic.

Firms are entitled to require arbitration as a condition of doing business with them, but those who deal with these firms are likewise entitled to require that there be no such imposition as their own condition of doing business.

Other Statistical Analysis

Sunday, 14 June 2009

Meanwhile, Andrew Sullivan notes the The Results as They Came In for the Iranian Presidential election. Specifically, as the official figures were up-dated, they fell almost perfectly along a straight line. The fit has an R2 of .998, which is a virtual impossibility. So the official figures are a bald lie. That doesn't mean that Ahmadinejad wouldn't have won under a fair count, but it means that he chose to steal the election rather than to risk losing under a fair count.

Tweaking the Truth

Friday, 9 January 2009

Here's an example of a wretched journalistic practice:

US job losses hit record in 2008 from the BBC
More US workers lost jobs last year than in any year since World War II, with employers axing 2.6 million posts and 524,000 in December alone.

The US jobless rate rose to 7.2% in December, the highest in 16 years.
One should not begin with a news story with a lie, and then correct it. The head-line here doesn't refer to a post-war record. A head-line is often the only part of the story read, and almost always the first part of the story read. So people with no sense of history — because their educators in the school system and in the main-stream media don't impart one to them — are filled with anxiety and rage. The emotional effect lingers even after the correction is given, and some never get the correction.

The economic news has been bad, but it simply doesn't compare to that of the Great Depression — which itself shouldn't be seen as necessarily our worse down-turn. (Those who uncritically presume that it was should look into the Depression of 1837.)

One should, BTW, be careful to distinguish amongst different statistics:

  • the number of jobs lost,
  • the unemployment rate,
  • the employment rate,
  • changes in these rates.
Note that the article acknowledges that the unemployment rate is not at a post-war high; it has, rather, climbed faster than at any time previous in the post-war period. The unemployment rate itself was worse at the end of the kinder, gentler Administration of GHW Bush.

Second, when there is job creätion as well as job loss, people may lose jobs but spend relatively little time unemployed. (Being dismissed from a job is still a stressful experience for most people, but not necessarily equivalent to being materially impoverished.)

Finally, the unemployment rate is not simply the complement of the employment rate. The unemployment rate, which tries to measure the number people who are seeking employment and unable to find it, is a fairly junky statistic. On the one hand, it doesn't count people who would choose to work if the were offered a job, but who have just given-up hope of finding one; and it doesn't count people who are under-employed, wanting full-time jobs but only able to secure part-time employment. On the other hand, it does count people who aren't sincerely seeking employment, but are going through the motions of seeking a job so that they can continue to collect benefits from programmes that require them to seek employment. The employment rate is simply the percentage of people of working age who have jobs. It has problems — including that it counts under-employed people — but it's less junky that the unemployment rate.