Labouring: A Reply
回应了Peter Wardley对作者1983年文章的两处数据错误指正，说明这些错误对原文结论影响甚微，并感谢其纠正。
We thank Peter Wardley for pointing out two errors in our 1983 Economica article. Replication of empirical results in economics is a time-consuming and relatively unappreciated endeavor (Dewald, Thursby and Anderson, 1986). We should emphasize, however, that Wardley's corrections are minor and do not affect significantly the discussion, results or conclusions reported in our original paper.' Wardley's first correction results from our use of an 1880 labour productivity figure listed incorrectly (as 302) in R. H. Walters's 1977 book, rather than the correct figure (312) listed in his 1975 article. We consulted both sources, used the book to obtain our data series, but cited the article since it was more readily available to readers. For this we apologize. The differences between Wardley's Table 1 and our Table 1 arise from the correction of this single observation. Not surprisingly, changes in coefficients and t-ratios are trivial. For example, Wardley reports coefficient (|t|) values of -0 303 (3 36), -0-287 (2.87), -0-26 (2.34), -0-265 (2 25) and -0-28 (2-28) for the wage variables. By contrast, the corresponding values reported in our 1983 paper are -0-301 (3 35), -0-287 (2 86), -0-261 (2-36), -0-258 (2.18) and -0-276 (2 24). The second correction results from an error made by us in obtaining our FIML estimates. We 'reached back' for a lagged price term that was not in our data set. TSP set the log of the non-existent 1873 price equal to zero. (More recent versions of TSP provide an error message when one reaches outside the data set.) Wardley's Table 2 provides FIML results estimated with the correct 1873 price. While our conclusions were not affected significantly, we regret the error and appreciate Wardley's correction.