cruits, disqualifying them would have a significant effect on aggregate attrition, reducing first-year attrition from 21.4 to about 18.9 percent.
The analysis suggests that an enlistment standard for light smokers, however, would not be cost-effective. The net benefits are negative, and significantly negative, under any standard that limits the proportion of light smokers in the Navy solely on the basis of their smoking behavior.
We now conduct the same type of trade-off analyses for the Army. In the case of the Army, recruits were classified as “nonsmokers,” “less than daily smokers,” and “daily smokers.” The recruit population shares and the 12-month attrition rates are shown in Table 7-4. Immediately, we observe that the attrition rate differences between smokers and nonsmokers are much less dramatic in the Army compared with the Navy. Because the differences in the attrition rate between nonsmokers and less than daily smokers are insignificant, we combine them into a single category for the purposes of analysis. This new category “nonsmokers or infrequent smokers” constitutes 68.4 percent of the recruit population and has an attrition rate of 14.3 percent.
If daily smokers in the Army were ineligible to enlist, the savings in reduced attrition costs would be about $38,400,000. However, the additional recruiting costs resulting from eliminating 32 percent of the market more than offset the savings from lower attrition. The loss that would accrue from such a policy would be about $21,000,000 per year. Policies that would restrict, but not eliminate, daily smokers would also increase costs.
Although a formal analysis was not carried out for the Air Force, the similarity with the Army attrition rates suggests that screening out smokers as identified by the Air Force study would not be cost-effective.
Flyer and Eitelberg (2005) found that the behavioral issues that may help to explain the relationship between preservice smoking and first-term attrition are similar in many ways to those underlying the relationship between dropping out of high school and early release from the military. Moreover, there is an interaction between education and level of smoking such that certain levels of smoking can magnify the effects of education on attrition.
This point is illustrated in Table 7-5, which presents 12-month attrition rates by education category. The table shows that recruits with Gen-