The effectiveness of an intervention in single-subject research is evaluated by observing combined patterns of level, trend, and latency in data graphs. A highly effective treatment typically produces obvious, immediate changes in level, a reversal of negative trends, and short latencies upon introduction. Conversely, an ineffective treatment is characterized by minute shifts in level, ambiguous latencies, and trends that appear to be mere continuations of patterns already established during the baseline phase.