Validity is important to the credibility of research findings. As Babbie (1989) explains, there are three types of validity, one of which is criterion-related validity. Criterion-related validity is a predictive validity and should not be neglected when one is doing research. Operational definitions must be such that they measure what the researcher purports to measure if any valid prediction about human behavior is to be made. Therefore it is important to help sociology students understand validity; conclusions drawn from invalid research are not useful in discussing the problem at hand. In this paper we describe a project that provides novice researchers with experience in dealing with the problems of achieving valid research findings. This demonstration involves obtaining the actual time people spend brushing their teeth, as well as obtaining their subjective estimates of time spent brushing. Perhaps because of the social desirability effect (see Albrecht, Chadwick, and Jacobson 1987), people may be expected to overestimate the time they actually spend. Another possible reason for inflated estimates is that people are bored with toothbrushing, and the time it takes to brush one's teeth seems longer than it actually is. Finally, because people simply may round their estimates up to the nearest minute, estimates of brushing time may be inflated. Instead of reporting that they brush their teeth for 1.7 minutes, people probably will report that they brush for two minutes. Whatever the case may be, each of these instances would yield invalid data, and any prediction of actual toothbrushing time would be erroneous. We asked 448 students from introductory sociology classes at a large private university to find out how much time their roommate, spouse, or another person to whom they had access actually spent brushing his or her teeth. The students were instructed to observe that person brushing his teeth without the person's being aware of the observation. Actual brushROBERT M. FERNQUIST