Myth and Measurement by Card David; Krueger Alan B.; Card David

Myth and Measurement by Card David; Krueger Alan B.; Card David

Author:Card, David; Krueger, Alan B.; Card, David
Language: eng
Format: epub
Publisher: Princeton University Press
Published: 2015-03-21T16:00:00+00:00


NOTES

1. This description is simpified somewhat. Many researchers also adjust the Kaitz index for the minimum-wage rate that was applicable to newly covered workers in 1980 and earlier years. After 1980, the minimum wage for newly covered workers applied only to a trivial fraction of covered workers.

2. For example, Murray Weidenbaum (1993), chairman of the Council of Economic Advisors during the Reagan administration, recently claimed that, “the Minimum Wage Study Commission concluded in 1981 that a 10 percent increase in the minimum wage generates a 1–3 percent increase in the unemployment among those holding minimum wage jobs, mainly teenagers.”

3. We omit Adams (1989) from this table because it is unclear whether his estimate pertains to teenagers or to all workers, and because his sample period was not reported.

4. The output-constant demand curve also is homogeneous of degree zero in factor prices.

5. Fisher (1973) and Hamermesh (1980) also raised related criticisms.

6. A classic example of this type of problem was noted by Cochran (1957). Cochran described an agricultural experiment in which oat fields were randomly selected for fumigation to reduce eelworms and thereby increase yields. There were two outcome variables in this experiment: (1) the crop yield; and (2) the number of eelworms remaining. Cochran warned against controlling for the number of eelworms in trying to estimate the effect of the fumigation treatment on crop yields.

7. De Long and Lang (1992) provide evidence of publication bias in economics articles. Concern about publication bias is by no means unique to studies in economics. The problem seems to be especially important in medical studies of cancer treatments (see Berlin, Begg, and Louis 1989).

8. See Begg and Berlin (1988). One factor that Begg and Berlin examined in their study of publication bias is the relationship between sample size and statistical significance. They interpreted the absence of an association between sample size and statistical significance in clinical trials of cancer treatments as evidence of publication bias.

9. Not all studies begin with 1954 data. Thus, the date of publication is not perfectly correlated with the sample size.

10. Notice also that, even with dependent observations, the t-ratio is expected to increase as the sample size increases.

11. Ragan (1977 and 1981) reported t-ratios only for disaggregated groups of teenagers. In this case, we used the average t-ratio. Mincer (1976) does not report a t-ratio, but did report that the minimum-wage effect for white teenagers was significant at the 0.01 level. In this case, we used 2.39, the critical t-value for a two-sided test of a null hypothesis at the 0.01 level.

12. The degrees of freedom are equal to the sample size minus the number of explanatory variables.

13. The standard errors reported in Table 6.1 assume that the errors from the regression equation are independent. This assumption is incorrect, because the underlying studies use overlapping data sets. The standard errors should not be interpreted literally.

14. A regression of the elasticity on the standard error (without a constant) yields a coefficient of 1.51 and a standard error of 0.21.

15. Welch



Download



Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.