difference in checking over this time period stratified by the characteristics of firms that are
identically measured in these two surveys. We also show the proportions of all firms with each
characteristic, and the weighted contribution of each firm characteristic to the mean difference in
checking over this time period -i.e., the product of the proportions and the percentage-point
differences, using the 2001 proportions to generate the weights.16
The results show that checking increased over this time period in most of these
establishments. However, we find variation in the differences in checking over this time period.
Checking increased rather dramatically in retail trade, manufacturing, in large firms (100+
employees), and in the suburbs. On the other hand, checking decreased in construction and
wholesale trade, and increased nominally in small firms, who are also among the least likely to
check in 2001. However, once we account for the distribution of firms across these
characteristics, our results show that the rise in checking in suburban, large and medium sized
firms, and firms in the service, retail trade, and manufacturing industries drove much of the
overall increase in checking we observe in Figure 5.
In addition to the factors mentioned above, firms may also do criminal background
checks because they are legally required to do so. As noted earlier, many states require
employers to check the criminal histories of applicants for certain occupations and jobs. Figure 6
provides some information on the extent to which employers check because they perceive that
they are legally required to do so.17 As pretext to these responses, our survey asks whether the
employer always, sometimes or never checks generally. It then asks whether the employer
16 We also estimated these weighted contributions to the overall increase in checking over this time period using the 1992 to 94 proportion of firms with these characteristics, but found similar results to those we show here. This is because the differences in these proportions between 1992 to 94 and 2001 are very small.
17 Of course, we are unable to verify employer responses regarding whether they are actually required to check by law. Hence, these responses can be interpreted as employer's perception of their legal responsibility to check.