Because neither the survey data nor the BWDC data were free of measurement error, we created four alternative indicators of filing status.
9 percent of the sample had wage-loss claims in the BWDC data that matched by date and body part with their occupational disease report.
To estimate benefits, we used the BWDC algorithm to calculate after-tax weekly earnings, and then a weekly benefit variable was constructed for every observation with sufficient data.
9) The BWDC data do not include claims that were only for medical benefits with no wage-loss component.
10) In a handful of cases we can be almost certain that this is what occurred, in that after being asked when they filed the claim for an episode reported in 1996, respondents gave a date from some earlier year that corresponded to a date on a record in the BWDC data.