The measurement of happiness has received increasing attention in behavior analytic literature. Happiness in individuals with developmental disabilities has been assessed by 1) counting a specific behavior, or 2) sampling constellations of behaviors. The purpose of this study was to examine the two approaches while observing nine child and teacher dyads at an autism treatment center. Results showed that, overall, a constellation of behaviors can yield similar patterns when compared to a specific behavior count. However, the affect of one person did not predict the affect of the other and similar instructional conditions did not predict affect either. The implications of these results and future directions are discussed.