Table 4

 Results for reduced multivariate regression models for traced handguns linked to violent and firearm related crimes, and for all traced handguns*

VariableHandguns linked to violent and firearm related crimesAll traced handguns
RR95% CIp ValueRR95% CIp Value
*Gun years of exposure/1000 is employed as an offset variable; results are on a per 1000 gun year basis. Time to crime was not incorporated in these models.
†Measured at the county level.
RR, rate ratio; CI, confidence interval.
Variables measured at the retailer level
    Gun years from police sales, %0.990.99––1.000.007
    Gun years from sales of multiple guns, %1.011.00–1.020.07
    Denials, % (of sales + denials)1.431.32–1.56<0.00011.371.28–1.47<0.0001
    Median age of purchasers, y0.950.94–0.97<0.00010.950.94–0.96<0.0001
    Retailer type (pawnbroker  = 1, gun dealer  = 0)1.261.02–1.550.031.301.10–1.530.002
Variables reflecting local gun commerce and tracing policy
    Federal Firearm Licensees per 100 000 people†0.970.96–0.98<0.00010.970.96–0.98<0.0001
    Retailer’s city traces all recovered guns (yes  = 1)1.431.15–1.780.0011.521.28–1.79<0.0001
Crime rates per 100 000 people†
    Latino population, %1.011.00–1.020.07