A Working Model: From Message to Behavior
Social-norm messaging becomes easier to judge once two kinds of evidence are connected: the causal pathway that a message is supposed to activate, and the evidence showing where that pathway actually ends. The working conclusion is neither “norms always work” nor “norms never work.” A norm appeal is a conditional intervention whose reliability depends on movement through several links—perception, intention, immediate choice, and maintenance—and on whether the study has measured the link it claims to have changed.
Reliability belongs to the chain, not the message
A printed statement, text message, or personalized comparison is an input, not an outcome. It supplies social information about a reference group, but the recipient still has to notice it, interpret it, and treat it as relevant. A descriptive claim about prevalence can make a behavior seem less unusual or more feasible. An injunctive claim can make it seem more socially approved. Neither claim guarantees that the recipient will form an intention, and an intention does not guarantee that an available action will be taken.
The chain can be separated into five questions. Was the message encountered and understood? Did the recipient’s perceived descriptive or injunctive norm change? Did an intention or immediate choice change? Was the target behavior observed rather than merely reported? Did the behavior persist when the social cue, observer, or momentary salience was gone? These are not five ways of saying the same thing. They are increasingly demanding claims about what the intervention accomplished.
A study that stops at the second question may still be valuable. If a carefully controlled message changes perceived prevalence or approval, it has reached the psychological target it was designed to reach. But it has not shown that a person acted. A study that finds a change in intention has moved one step farther, but intentions can be defeated by access, time, competing priorities, or routine. A field experiment that observes a choice establishes behavior in a particular setting, yet an immediate choice may be a situational response rather than a durable change. Maintenance requires a different time horizon and often a different design.
The broad review makes this separation especially important. It pooled studies that used both objective behaviors and intention outcomes, with follow-up ranging from immediate measurement to two years. A standardized average can combine these results for a broad estimate, but it cannot make their endpoints equivalent. If an intervention changes a stated willingness to exercise, that result should not be read as though it changed actual exercise. If it changes a single prescription or screening decision, that is important in its own right, but it does not establish that the underlying social norm has been internalized.
This is why the review’s limited use of perception measures matters more than a routine caveat. Only 10 of the 89 studies elicited perceived norms before intervention, and only 17 used manipulation checks for awareness or understanding. A null behavioral result can therefore arise at different points.