Designing the Comparison: Dose, Direction, and Attainability
A comparison intervention is not defined only by the fact that it shows people what others are doing. Its effects depend on the standard, the behavioural dimension, the perceived gap, the control condition, and what else the intervention asks participants to do. This chapter examines those design choices and turns the trial evidence into conditional judgment: not a rule that upward comparison always motivates or always harms, but a way to ask when a comparison is interpretable, actionable, and safe enough to test.
The evidence offers promising signals rather than a finished recipe. Some moderator analyses associate more sessions and desired-behaviour targets with larger short-term effects. A small set of trials suggests that more attainable upward standards may support longer-term change. Direct comparisons among upward, downward, and non-restricted standards do not establish a universal winner. The differences matter, but so do the small subgroups, high heterogeneity, bundled techniques, and varying certainty of the estimates.
Direction is not destiny
“Upward” describes the standard relative to a participant on a chosen dimension. It does not describe a person’s permanent social rank, and it does not tell us whether the comparison will produce a desired behaviour. A group average can be upward for someone above the average, downward for someone below it, or neither clearly for a person near it. A standard selected from a high-performing subgroup may be upward for most participants but not for everyone. The intervention’s label and the participant’s actual comparison are therefore related but not identical.
The Nature synthesis compared social-comparison interventions with upward, downward, or non-restricted directions where enough trials were available. In short-term analyses against active controls, upward and non-restricted interventions produced significant small effects in some subsets, while downward interventions did not show a significant effect in the small available subset. But direct short-term comparisons between upward and downward, upward and non-restricted, or different types of standards did not establish significant differences. The evidence base was too thin to support a general claim that one direction is inherently more effective.
That result is important because it rejects two symmetrical shortcuts. The first says that showing people someone better will automatically inspire them. The second says that upward comparison is inherently harmful and should never be used for behaviour change. Both treat direction as if it determines meaning by itself. The relevant questions are what the standard represents, whether the dimension can change, how far away the target seems, and whether the participant can translate the discrepancy into an action.
The target’s valence also changes the interpretation. An intervention may seek to increase a desired behaviour, such as a health action or performance, or decrease an undesired behaviour, such as resource use or consumption. In the short-term active-control analyses, both classes produced small effects when examined separately. This does not show that desired targets always work better. It shows that an intervention can be framed around moving toward a desirable action or away from an undesirable one, and that the framing should be tested against the actual comparator rather than assumed from motivational language.