Follow-up period. In contrast, when examining long-term effects, both commitment only and commitment plus another treatment yielded somewhat stronger results compared with other interventions (right side of Table 2), with small but reliable significance levels and average rs but very small buffers against unpublished nonsignificant findings.
Presence of moderators. Diffuse comparisons, or tests of heterogeneity, were conducted to see whether we could detect the presence of moderators in these analyses. That is, when significance levels or effect sizes are highly variable, it may be possible to break the set of studies into subgroups according to moderating factors, such as types of behaviors (e.g., recycling vs. conservation) or geographic area of the research (e.g., rural vs. urban). The Bonferroni adjusted alpha for our 16 meta-analyses (weighted/unweighted by intervention/follow-up by commitment only/commitment plus by control/other) is .03, and none of our tests of diffusion was significant (all but three p values were > .20). Naturally, these tests are limited by the small number of commitment studies; it is possible that as more studies are conducted, researchers will be able to identify moderating factors.
Summary. Meta-analysis results showed that—compared with control conditions—commitment only and commitment plus are more effective both during the intervention and after people had been released from their commitments. Results are more complex when commitment is compared with other treatments. During intervention phases, commitment-plus conditions were more effective than other conditions. However, during follow-up periods, both commitment-only and commitment-plus conditions yielded more durable change when compared with other typical interventions. These latter results are tentative as they are based on a small number of studies. However, using commitment in combination with other interventions is consistent with the idea that in applying psychology to natural settings, it may require more than one intervention to effect consistent behavioral change (Werner, 2003).
Theoretical Underpinnings: What Processes Underlie Commitment Effects?
Although the meta-analysis was able to show us that commitment making is an effective instrument for altering environmental behaviors, it could not demonstrate why it is effective. To explore the possible psychological