Do I need to prime both groups of my experiment?

Question

Hi,

I would like to use a prime at the beginning of my experiment. Therefore I want to prime the treatment group with an environmental prime to make theri environmental identity salient. What I am unsure about is, if I should prime the other group as well. I've seen studies in which e.g. on the one hand an environmental prime was applied and on the other a financial.

Wouldn't it be more correct to prime just one of the groups?

Answer 1

There's no simple response tho that question that would always be correct.

If you use no other prime, then you may confund the prime's content and having a prime. That means: If you observe a difference between your groups, you cannot be sure if it was caused by what you have shown or by the fact that you have shown something.

If you use a prime for the control group, you may run into a similar problem: You cannot be sure, if the effect was cause by the one prime or by the other.

One solution that is often apllied, is to have both: One control group with no prime, and another with a "neutral" prime. This gives you more information in the analyses, but requires +50% respondents. And, of course, you can still have situations when this is insufficient.

Comments

That sounds reasonable. The 50% more data which I would need to gather is disadvantageous, but I think I will follow that path.
In the prime, I actually wanted to make the environmental identity of the participants salient.
This is surely not a new idea and has been applied in different contexts, but I did not find the perfect fit and therefore decided to come up with my own simple prime. Sth like "Imagine you're an active environmentalist who's participating in a debating contest. Please list your most powerful arguments in ~10 sentences."

Would you advice against "creating" one's own prime and rather stick to an already succesful applied one? It's quite risky to apply my own one as, if it doesn't work, all the effort would be useless.
Secondly, how could a neutral prime look like? E.g. the same scenario without mentioning the "active environmentalist" and just invite the participant to chose a topic of personal relevance (or maybe even just "any" topic)?

Ah, and thank you for the quick reply!

---

It's always a balance between risk (the prime may fail) and having the perfect prime for your research design... If you would like to invite even more respondents, you may now consider to have one group with "your" prime and another one with that which has been working in earlier studies. But that's only reasonable, if you don't trust your prime.

I would argue that your prime should be as strong as possible in an experiment. If it's a text, then display with double-sized letters and make sure that the respondents cannot click to the next page before a specified amount of time.

> Secondly, how could a neutral prime look like?

Find an issue that few people have a strong opinion to. This could for example be: "Imagine you're an active pop-art activist who's participating in a debating contest. ...". By the way: You should probably tell the respondents what the arguments should be for or against. When actually using your wording from above, many respondenst will be confused.

---

That is something I didn't even consider to specify, thanks for the hint.

I just read a paper where they applied an environmental prime in the form of a text (~250 words) informing participants about e.g. the bad implications of excessive water usage for the planet. The text ended with a few bullet points containing tips on how to behave in a more environmental-friendly way. So listing these bad implications could be a stronger prime as you mentioned.

What bothers me here is, and my reasoning might be absolutely wrong, that I think participants will realize what I am trying to do (especially when I present an environmental product with a purchase decision afterwards) and therefore it won't work. Besides that I feel like a self-written text forces the participants to be much more involved as e.g. reading a text on autopilot.

---

That's a decision, nobody else can make for you. Getting the respondents involved is usually a good way to actually make them think about something. Yet, one might argue that this is more than priming. Or maybe a few respondents will write down that goes in a very different direction than you you intended them to think about. There's always some risk.

I personally think that it's a good priming (if you optimize the wording a bit and make a good face-to-face pretest, using the think-alout method to ensure that people understand your question/prime right). But to be honest, it depends a lot on who will grade the work, or who will review the study, should you think about publication. Above all scientific systematics there's a substantial share of individual interpretation. Finally, the world is not as unambigious as a quantitative researcher would love it...

---

I may have one last question. In case I chose a rahter seldom identity, e.g. the popstar you mentioned above. How would I justify this approach? Is it to minimize the possibility to make an identity salient that would in some way be the opposite of an environmentalist (e.g. chosing a republican identity)?

Initially I wanted to apply one environmental prime and one financial/analytical prime. The latter one could have been something like: "Imagine you are a successful investor and because of your impressive track record you are asked to have a presentation about the main factors of successful investing which you have learned troughout your career."
But here might apply the same reasoning as if I would chose sth like a republican identity and therefore possible differences would likely to be just due to the choice of primes. Is this the reasoning which I could use to justify a seldom identity?

---

The pop art artist was really just an example. If some respondent thinks about a specific artist that's very engaged in environment protection, than this will not work. But in general, bein an artist is (I guess) uncorrelated to environment. Therefore, it is "neutral" in this respect.

If you chose a member of the republican party, there's a clear correlation to environmental issues. And therefore, this was no neutral prime, but another prime. I am unsure about the financial prime, but would expect a similar problem here: Making money is often related to exhausting the environment.

I know too little about your specific issue to give you really good advice here. Finding a good neutral prime or stimulus ain't easy at all. Finally, you must be convinced by your reasoning. There's always the risk of being wrong or missing some crucial point, but that's the risk of research. If you give a good reasoning in your text, than others researchers are invited to criticize that and make it better. Try and make it good (I think, you're on the right track). And accept the fact that you'll not be able to make it perfect.