We are 3 undergraduate students of Psychology at the UGA (Grenoble Alpes University). By working on our Study and Research Assignment and under the supervision of Amélie Gourdon-Kanhukamwe we participated in the CREP (Collaborative Replications and Education Project). We choose to replicate the first experiment of Sparkman and Walton described in their 2017 paper “Dynamic Norms Promote Sustainable Behavior, Even if It Is Counternormative''. STUDENT RESEARCHERS Elsa BETTEGA Lou DE RANGO Nadia GRZYBOWSKA PROJECT SUPERVISION Amélie GOURDON-KANHUKAMWE Summary This article studies the influence of dynamic norm compared to static norm on individual behavior. A dynamic norm focuses on the change of a norm over time, whereas the static norm focuses on the current state of a norm. The hypothesis of Sparkman and Walton is that by presenting a message carrying a dynamic norm, compared to one carrying a static norm, participants will be more interested in changing their normative behavior. In the first experiment this unsuitable normative behavior is the one of meat consumption. During the experiment, participants answered an online survey. They were separated in two groups, each were shown a fact about the american meat consumption. In one group this fact was presented with the form of a static-norm and the other in the form of a dynamic-norm. Next every participant answered a question about their interest in reducing their meat consumption on a scale of 1 to 7 (1=not at all, 4=somewhat, 7= extremely). They were then asked the percentage of people making an effort to limit their meat consumption as well as the number of meatless meals eaten each week by the one who limited their meat consumption. They finished by reporting : their gender, political ideology (on a scale of 1 to 7, 1=very liberal, 7= very conservative) and whether they were vegetarian or vegan. The only IV (Independent Variable) in the experiment is the kind of norm induced in the message presented. This IV is provoked, inter-participant and of two modalities : dynamic-norm and static-norm. The DV (Dependant Variable) of the experiment is the answer of the first question about meat consumption, all the others are CV (Controlled Variable). The results were on the 122 participants, 4 were put aside (by being vegan or vegetarian they could hardly reduce their meat consumption). The results follow the hypothesis of Sparkman and Walton, participants showed more interest in following a counternormative behavior, reducing their meat consumption habit, when in the dynamic-norm condition. An effect of gender and political ideologies were also present and with these effects controlled, there was still a significant effect of the dynamic norm. Our Replication We intend on doing a replication experiment, we will then explore the same hypothesis as Sparkman & Walton about the impact of the dynamic-norm on the same counternormative yet sustainable behaviour. In addition, we will verify if the same effect of gender and political orientation is to be found in France. We planned on doing an online survey using Qualtrics. Unlike the original experiment the participants will not be remunerated. On the last online survey we led we ended up having approximately 200 participants. We then hope to have as much on this survey. By using G*Power we know that, if we manage to have between 150 or 260 participants the statistic power of the study would be between 0.80 and 0.95. ORIGINAL OSF PAGE : This page contains all the collected study materials and instructions from authors that we have compiled. We also include any comments from other contributors or follow up instructions that we have learned since the beginning of the project. Click "read more" below or choose the "Wiki" option above for further information. For technical problems, please contact OSF help desk (support@osf.io). For questions or information about the studies contact either either Kiersten Baughman (kierstenbaughman7@gmail.com) or Becka Plitt (becka.plitt@gmail.com) so that the materials can be made available on this website. Abstract It is well known that people conform to normative information about other people’s current attitudes and behaviors. Do they also conform to dynamic norms—information about how other people’s behavior is changing over time? We investigated this question in three online and two field experiments. Experiments 1 through 4 examined high levels of meat consumption, a normative and salient behavior that is decreasing in the United States. Dynamic norms motivated change despite prevailing static norms, increasing interest in eating less meat (Experiments 1–3) and doubling meatless orders at a café (Experiment 4). Mediators included the anticipation of less meat eating in the future (preconformity) and the inference that reducing meat consumption mattered to other people (Experiments 2 and 3). In Experiment 5, we took advantage of a natural comparison to provide evidence that dynamic norms can also strengthen social-norm interventions when the static norm is positive; a positive dynamic norm resulted in reduced laundry loads and water use over 3 weeks during a drought. Materials The original paper is here. This project will only be replicating the first study as described in the paper. The original materials for Experiment 1 are here. Sample To obtain a CREP completion certificate the minimum N for this project is 100. Notes from Original Author Perhaps most importantly, a successful replication of Experiment 1 would depend upon a successful manipulation of the salience/knowledge of static and dynamic norm information (the primary IV). This may or may not be as easy as it appeared in our research, where we had people simply read a statement providing static or dynamic norm information. It is easy to manipulate what people read, but it is not necessarily easy to manipulate what is salient in the reader's mind as a consequence of reading it. Specifically, people may read a static norm statement and then it may remind them of the trend in people eating less meat overtime (i.e. the dynamic norm). If this were the case, one should not predict the results we saw in our research, but instead a null result (see the general discussion of the paper). Further, there is good reason to believe this may be the case in this context. We ran the meat consumption studies starting in 2012. Since then, discussion of the decline in meat consumption in the US has been thoroughly highlighted by almost every major media outlet (see the New York Times, Guardian, the Atlantic, Washington Post, MSNBC, NY Post, Mashable, EVEN FOX NEWS, as well as online channels). Hence, it may be highly salient to participants, even if they're in the static norm condition. This may make it difficult to manipulate the IV successfully, as those in the static norm condition may already essentially be in the treatment condition. Perhaps worse, in the last few years there was actually a slight increase in meat consumption, and, even though there was a prior decade of decline, many media outlets are quick to jump to the conclusion that the trend has reversed, and now more and more people are eating meat (see USA today, VOX, among other outlets like blogs and such). This may easily lead participants to not trust the dynamic norm information provided in the original study. That said, I have a few thoughts on how to mitigate the above: After study 1, we began to include a free response with the IV that asked P's to answer why they thought the norm info we gave them was the case (see materials for studies 2, 3, and 4, e.g. "Why do you think this is?"). This is helpful for two reason: first, it ensures that participants on a platform like mturk actually read the information they're provided and respond to it; and second, it gives you a window into what they're thinking about. Thus, you have at least a rough way to assess if you've manipulated the salience of static or dynamic norms successfully. You can use this to see if anyone doesn't provide a coherent response (if they leave it blank, or write gibberish, they failed to complete the IV portion of the study and could be dropped). And, to the salience-related point above, it allows you to code what percent of participants in the static norm mention any attitude or behavior change overtime (i.e. if they say "People are becoming more health conscious." or "I think eating meat has declined because... " this indicates they are thinking about the dynamic norm, even though they're in the static norm condition, which would be a failure of to manipulate the IV). In our work, I believe we saw about 5-10% of people imply something had changed in the static norm group. We felt this was relatively low, so we weren't worried about it. But, if for instance you saw 20% or more of people mention that related attitudes, concerns, or behavior of others has shifted or changed in some way, that would suggest a greater portion of people are now thinking about the dynamic norm in the static norm condition. Indeed, these would only be the people who explicitly mentioned it, and as such is a conservative estimate of the true number of people who were thinking about the dynamic norm. That said, you could use this metric as a moderator: i.e. those who mention something changing in the static norm should not show differences between the statis norm, while the subset who mention no social change should be more likely to show differences compared to the dynamic norm condition. You might consider piloting materials to check for some of the concerns I mentioned above. You could ask participants for an assessment of believability. And you can ask participants "Did you already know this information?" after being given the dynamic norm. This would help you determine if the facts we shared are now simply common knowledge in the dynamic norm condition. This could also be a potential moderator (i.e. the information has to be new to generate an effect). For study 1, we didn't know what sample size to choose. Based on the effect sizes seen in subsequent studies ~d=.3, we should have ran a larger sample, probably closer to 350 total in order to be 80% powered. Or if you want to test for moderation, more would be necessary. In interpreting your results, be sure to clarify the difference between manipulating what someone reads, and what is salient to them. Our theory is about what is salient... which may be different from what they read, hence successfully manipulating the IV is an important component, and should be treated thoughtfully. In addition to the salience / knowledge point above, there are two more concerns: Experiment 1 was conducted using MTurk, but was before the explosion of bots / click farmers from outside the US that many have detected (see here and here). In our research since the rise of non-US click farmers, we have adapted our methods to screen them out. See the methods used in Experiment 5 of Sparkman & Walton, 2019 for an effective method to screen out such participants effectively. We are currently living in a time of great social change that was brought about by CV19. It feels painfully obvious that some participants will attribute a decline in meat consumption to CV 19 and meat shortages. This attribution is problematic because it means that people aren't assuming that people are voluntarily decreasing how much meat they eat (as in the original experiment), but have simply been unable to buy meat due to shortages. This suggests yet another moderator: social change due simply to the limited ability to do something may not be as motivational as social change chosen voluntarily by many people. I am less certain how you'd get around this. But I would definitely measure it / screen for this though. And it's worth thinking carefully about. Finally, I hope this doesn't sound like I'm discouraging you from replicating our work, I am actually excited about the prospect. But there are some very sizable (but interesting!) challenges that problematize the simple notion of a direct replication here.