Statistical testing between the 2003 ACS and CPS annual average percentage distributions for occupation, industry, and class of worker confirms the overall comparability of the ACS data for these items. Across all three variables, there are only three differences exceeding the threshold for comparability. However, several findings from our analyses are worth noting. Measurement of occupation groups does not differ for four of the six groups, but closer examination may be needed to understand why ACS underestimates the “Management, professional and related occupations” and overestimates “Sales and office occupations” relative to the CPS data. The ACS distribution of persons by industry is, in most cases, similar to the CPS, with no differences exceeding the threshold for discussion.

Comparing ACS and CPS measurement of class of worker does yield significant differences. First, the two main types of “Private wage and salary workers” differ between surveys. “Private for-profit” is measured less often in ACS than CPS, and the converse occurs for “Private not-for-profit”, with a net undercount of the overall category in ACS relative to CPS. The other main categories of class of worker also differ between the surveys. The reasons for the differences in these estimates cannot be explained with the data at hand, but these results suggest that further investigation of the origin of the differences is warranted. Perhaps slight question differences or coding protocols yield these differences. 17  Overall, the ACS and CPS estimates of industry, occupation, and class of worker are comparable. Further investigation may shed light on the determinants of the few noted differences.