Does participation improve performance? : establishing causality with subjective data

Three possible econometric objections to these findings are addressed and answered. The subjective nature of the data does not preclude valid, cardinal measures of participation appropriate for statistical analysis. Halo effects -- changes in the measurement of one variable because of the observed state of another variable -- do not seem to induce a strong upward bias in the measurement of participation or project performance. Reverse causation is unlikely: estimation using instrumental variables, data on project timing, and documentation of case studies support the cause-effect relation between participation and better project performance. (Author abstract)