Table 3.
Accuracy-Contingent Analysis. Best fitting model (Experiment 2).
| Effect | Estimate | S.E. | z-value | p-value |
|---|---|---|---|---|
| Intercept | -1.20 | 0.25 | -4.79 | <.00001 * |
| Accuracy on Prev. Inst. (Incorr. vs. Corr.) | 1.16 | 0.30 | 3.90 | .00001 * |
Note: Results of multi-level logit model with crossed intercepts for both Participants and Items. The lmer function in R was used [syntax: Accuracy ~ 1 + PreviousAccuracy + (1 | Participant) + (1 | Item)]. Model was a better fit than a model with fixed effects (χ2(1) = 15.15, p < .001). Models adding fixed effect of Condition or Condition × PreviousAccuracy interaction did not reliably improve fit. Adding random slopes to null model did not improve fit.
statistically significant.