1) require longitudinal data with repeated measurements of multiple domains or levels of functioning across time

2) need to control for across-time continuity and within-time covariance (we dont want to measure correlation that were already present at the beggingn of the study)

3) distinguish between a cascade effect from earlier domains of functioning to newly emerging domains

4) distinguish between a cascade effect and developmental manifestation of progress within a broader underlying domain of functioning

5) time span between measurement intervals needs to be large enough to allow cascade effects to show up at all

6) most convincing evidence for cascade effects will be experimental (not based on correlational data)