Assessing levels of intoxication through behavioral observation

The primary goals of this research were: 1) to develop an alcohol intoxication detection aid (IDA) that can be used to estimate reliably and accurately another person's level of alcohol impairment; 2) to evaluate the effect of alcohol on a computer-controlled critical tracking task (CTT), which assesses skills important to driving a vehicle; and 3) to examine the relationships between these two instruments and measured blood alcohol concentration (BAC).

Teplin and Lutz (1985) developed an alcohol intoxication checklist for use in a hospital emergency room. Their results indicated a high correlation (.85) between actual BAC and intoxication estimates obtained by applying their checklist. This thesis refined and extended this prior checklist procedure for use with college populations, and included procedures to measure its validity and reliability.

The CTT is a computer task which requires a subject to make a response to a brief digital display. Dependent measures include subjects' response time and error rate. Subjects' performance was consequently correlated with their BAC. The subjects were university students (n = 232) consuming alcoholic beverages at three regularly scheduled weekend fraternity parties. Subjects were recruited for testing from among those in attendance at these parties. Subjects were of legal drinking age.

Results for both instruments were instructive and encouraging. Although reliability and validity coefficients for use of the IDA during the first party were unacceptably low. the changes made in the experimental methodology after this testing proved beneficial. Results of IDA use at the third party showed substantial increases in both interrater reliability and correlation with actual BAC. These results, however, were found to vary across individual judges. With CTT performance, mean response time and mean error rate were both reliable indicators of intoxication, and results stood up fairly well under a signal detection approach to the data analysis. Implications for use of both instruments in real-world settings are discussed.