We have seen that agreement between mother’s reports and medical records on asthma is moderate, with a substantial share of cases identified by one source but not the other. Each source identified some children as having asthma that the other source did not identify as asthmatic.

How does this lack of concurrence affect measures of asthma prevalence and distribution? It would be possible for prevalence estimates for the two sources to be the same even if they are calling different children asthmatic. It is more likely that the prevalence estimates will vary.

Another important question is how the lack of concurrence affects estimates of asthma distribution among children in the general population, in other words, which children have asthma? Put differently, is there systematic bias in one or both sources that cause the estimates derived from each to differ, beyond the expected random statistical variation?