Editorial NexoScience
Nudge someone toward recycling, and chances are it sticks. Nudge them toward driving less, skipping a flight, or buying an energy-efficient fridge, and the effect tends to fade almost as soon as the campaign that produced it ends. That is the uncomfortable pattern that emerges from the largest test to date of whether “soft” climate campaigns, the pamphlets, the app reminders, the neighborhood comparison letters, actually change what people do.
Researchers Claudia F. Nisa, Jocelyn J. Bélanger, Birga M. Schumpe and Daiane G. Faller, from New York University Abu Dhabi, pulled together every randomized controlled trial they could find that measured real, tangible household behavior rather than self-reported intentions. Their meta-analysis, published in Nature Communications, combined 144 independent estimates from 83 studies conducted between 1976 and 2017, covering more than 3.09 million observations across households and individuals in Europe, North America and beyond.
The overall effect is real, but very small. Averaged across every behavior and every type of campaign, behavioral interventions nudged people toward lower-carbon choices with an effect size of just d = -0.093, which the authors translate into a probability of benefit of 6.6%. In plain terms, a randomly chosen person exposed to one of these campaigns has only a 6.6% higher chance of acting greener than someone who got no campaign at all. Restrict the analysis to the largest, most rigorous trials, the ones without self-selected volunteers, and that number drops to somewhere between 2% and 3%.
The effect disappears once the campaign stops. Fifteen of the included studies tracked what happened after the intervention ended. Once it did, there was no statistically detectable lasting change (d = -0.13, 95% CI crossing zero). Whatever behavior shift a poster, an email or a smart meter produced, it mostly evaporated the moment the nagging stopped.
Recycling is the outlier that actually works. Among the six behaviors studied, energy use, transportation, animal product consumption, food waste, water use and recycling, only recycling reached what statisticians would call a medium effect (d = -0.457, a 32% probability of benefit). People given a recycling-focused campaign were, on average, about a third more likely to recycle than people who got nothing. Everything else barely budged: private car use showed an effect size close to zero (d = -0.036), and the purchase of energy-efficient appliances, arguably one of the highest-impact actions a household can take, showed no significant effect at all.
Small studies told a much rosier story than big ones. This is where the paper turns into a lesson about hype versus evidence. Studies with fewer than 100 participants per group reported effect sizes nearly twelve times larger (d = -0.335) than studies with 500 or more participants per group (d = -0.028). A formal funnel-plot test (Egger’s test, p < 0.001) confirmed the pattern is not random: the smaller and less rigorous the study, the bigger the effect it tends to report. Two decades ago, the field's own tracked history shows medium-to-large effects; the most recent, best-designed trials converge toward almost nothing. It is the sort of self-correction that inflated expectations rarely survive once enough good data accumulates.
Not all strategies are equally weak. Plain information, statistics, energy labels, in-home displays, is the most commonly used tool and the least effective, producing barely a 3.4% probability of benefit. Choice architecture, so-called nudges such as moving recycling bins closer or defaulting air conditioning to a higher temperature, produced the strongest results (d = -0.352, 25% probability of benefit), and social comparison messages, the “your neighbors use less energy than you” letters, came in second (d = -0.077, but a comparatively consistent one). The catch: nudges have only been tested on a handful of behaviors, mostly meat consumption and food waste, and almost never on the big-ticket items, flights, cars, home energy retrofits, where the carbon savings would actually matter most.
The authors are careful not to overstate the opposite conclusion either. A small effect is not the same as no effect, and their results do not argue for abandoning behavioral campaigns in favor of financial incentives or regulation. What they do argue is that treating information campaigns as a serious climate strategy on their own, without pairing them with pricing, regulation or better-targeted nudges, is not backed by the evidence. Getting a household to recycle turns out to be a solvable problem. Getting the same household to fly less or trade in the car is, so far, a much harder one.
Source: Nisa CF, Bélanger JJ, Schumpe BM, Faller DG. “Meta-analysis of randomised controlled trials testing behavioural interventions to promote household action on climate change.” Nature Communications, 2019. DOI: 10.1038/s41467-019-12457-2.
