The New York Times has corrected the article I blogged  about recently. I had contacted the ombudsman who routed my concern to the reporter.
Corrections are interesting in the internet age. A dead tree correction sits by itself a day or a week after the original article. The original mistaken article survives on microfilm misleading the unaware reader. And of course the consciousness of the reader who does not see the correction survives blissfully unaware as well.
On the web, corrections at the Times at least, are noted on the original article . Under the headline, the phrase “Correction Appended” appears and at the bottom of the article is the correction for the ambitious reader:
Correction: Aug. 11, 2004, Wednesday
An article in Business Day on Monday about the number of new low-income jobs compared with high-income jobs misstated figures from an analysis of Labor Department payroll data conducted by Economy.com, an economic research firm. Since January, industries ranked in the bottom fifth in terms of median wages have generated 177,000 jobs, not 477,000. Industries in the top fifth generated 135,000 jobs, not zero.
Unfortunately, the correction does not discuss the damage this change does to the central claim of the article. The article claimed that the economic recovery was producing too few jobs that paid too little. In support of the last point, the single fact mentioned in the correction was the only evidence. Here is that fact, correctly stated: Since January, the U.S. economy has created 1.1 million jobs as measured by the payroll survey of the Bureau of Labor Statistics. Of these new jobs, 17% are from the industries whose average wages are in the bottom quintile of all industries, while 13% of the new jobs are from industries in the top quintile.
Cause for concern? Hard to say without more information on the details, whether the quintiles are stable, whether the actual jobs, rather than the industries, pay less, whether the same pattern emerges when benefits are included and whether this is standard during a recovery.
I have the comparable data for the recovery after the 1991 recession. I went four months out from the trough of the employment numbers to make it comparable to January 2004. Then I went out for the next seven months. During that seven month period after the 1991 recession, payrolls grew by 924,000. Of that increase, 18% of the jobs were in the lowest quintile and only 10% in the top quintile, yet somehow, the recovery and the near-decade of expansion survived. (These calculations are based on the data provided to me by economy.com that were used to draw conclusions from the current recession. I wouldn’t quote them without verifying their accuracy. They are supposed to correspond to private non-farm payroll employment).
To really nail this down, I could do another calculation for a comparable period relative to the end of the recession rather than the trough in employment. But you get the idea: there’s no there, there. No cause for concern in the recent data and no cause for a front page Business Day article either.
Post-script: Ironically (embarrassingly?), this post has been amended after I discovered a numerical error in the earlier version of the post. The correction does not change the conclusion. Phew!