| Table 3.5. | Statistical Power Analysis for the EDS Follow-up |
| Effect Size | PPVT (Child) (N=97/44) | WJ-R (Child) (N=96/45) | Story & PC (Child) (N=123/55) | WJ-R (Adult) (N=149/65) | Parent Report (Child) (N=207/91) |
|---|---|---|---|---|---|
| .80sd (large) | .99 | .99 | .99 | .99 | .99 |
| .50sd (medium) | .86 | .87 | .92 | .96 | .99 |
| .40sd (medium) | .71 | .71 | .79 | .85 | .94 |
| .30sd (small) | .50 | .50 | .58 | .64 | .77 |
| .20sd (small) | .29 | .29 | .34 | .38 | .48 |
|
Note: Assumes one-tail test (Even Start does better than control). The sample sizes reported here have longitudinal data (pretest, posttest and follow-up) for each measure. Table reads: If Even Start children gain .80 standard deviations more than control children on the PPVT between pretest and follow-up, then there is a 99% chance that the EDS sample will allow us to conclude that difference is statistically significant. |
|||||
