Weekly Drinking Is Not Binge Drinking
Alcohol is the clearest test in the supplied evidence of why an if–then plan must be matched to a precisely defined behavior. The central question is not simply whether implementation intentions reduce “alcohol use,” but whether they change a weekly quantity, prevent a heavy episode, or do something different under different social and delivery conditions. Comparing those outcomes lets readers use moderator evidence as a conditional guide rather than as a ranking of universally best formats.
One label hides two different targets
The alcohol meta-analysis separated weekly alcohol consumption from heavy episodic drinking. Weekly consumption is an aggregate quantity across a period. Heavy episodic drinking, by contrast, concerns concentrated drinking episodes that can carry their own risks even when a weekly total does not look dramatically different. A person might reduce the number of drinks across several ordinary occasions while still experiencing a heavy episode. Conversely, a person might avoid one heavy episode without changing a weekly total very much. The measures overlap, but neither is a substitute for the other.
That distinction changes the planning problem. A plan aimed at weekly quantity can attach a response to recurring occasions: if I am about to order another drink on an ordinary evening, then I will switch to a non-alcoholic option. A plan aimed at a heavy episode may need to operate inside a rapidly changing social situation, perhaps when friends encourage continued drinking or when a round is already being ordered. The cue may be harder to isolate, the response may be more socially costly, and the person’s control over the situation may be more limited. These are plausible reasons for different effects, not proof that one particular mechanism explains the results.
Across sixteen included studies, the review found a small but statistically significant reduction in weekly alcohol consumption when participants formed implementation intentions. The sample-weighted average effect-size difference was d+ = −0.14, with a 95% confidence interval from −0.24 to −0.03. The negative direction means that the implementation-intention conditions reduced consumption more than their comparison conditions. The estimate was calculated after controlling for baseline differences, which matters because groups can differ before an intervention begins and because a follow-up-only comparison can misrepresent change.
The result for heavy episodic drinking was different. The pooled effect was d+ = −0.01, with a 95% confidence interval from −0.10 to 0.08. That interval is compatible with no average effect. The correct conclusion is not that implementation intentions can never affect heavy episodic drinking. It is that the included evidence did not show a reliable reduction for that outcome. Absence of a detected effect in these studies is not evidence that every possible plan, population, or delivery format would fail.
The baseline adjustment is part of why the weekly estimate is easier to interpret. If the intervention group already drank less before planning began, a follow-up difference could falsely make the plan look effective. If an effective intervention lowers consumption in one group while the comparison group stays similar, the variation in follow-up scores can also change. The review therefore calculated the effect from the change in each group rather than treating the final measurement as the whole story. That does not remove every source of bias, but it makes the reported contrast more responsive to what happened after the intervention.
The two outcomes also imply different standards for a useful plan.