Viewpoint: A Great Research Paper Explains A Lot, Not Everything

ImageRecently Kathleen Vohs of the University of Minnesota and colleagues published a paper on rituals and we think it is a fantastic model for how to write a paper. We think this paper solves a lot of issues that have been discussed by the professors and writers here at InDecision about how to write and present research. Particularly it avoids certain research “sins”.

In case you are in a rush, or too busy with important ideas or data analyses to read this entire blog post (and even more so the article), or whether you’d just rather go look at cat memes on Tumblr, below is a short version of this post, followed by a longer form post.

Short Version

The Problem: Michel Pham and others have declared that one of the biggest research “sins” is to propose that a single phenomenon has a single process. This “full mediation quest,” as it is sometimes known, can often lead to research that lacks practical relevance, any external validity, and at worst misrepresents reality.

The Solution: Vohs and colleagues’ paper on rituals studies an important issue, with strong external validity, and identifies the existence of a psychological phenomenon. They then provide strong evidence for a psychological process, but do not claim to have found an exclusive micro process for the phenomenon. In the general discussion (pages 14-17 of this link), the authors discuss how the phenomenon is most likely multiply determined – something Michel Pham in his excellent 7 sins InDecision Blog post argues we need to embrace more. In the end, Vohs and colleagues are able to provide that prized “theoretical contribution” of a psychological process without committing the specific research “sin” of over claiming that the phenomenon is just one process.

Of course, you can assert that sometimes we need more nitty gritty larger research than this. We’ve got no argument with you there and we doubt Vohs would either – in fact, you can see other Vohs papers for example of “bigger” theories. It should go without saying that all research can’t be short Psychological Science reports like this one but we think there is a great lesson to be learned from the approach of this one paper (in addition to the quality scientific contribution) and that lesson is that, whenever possible, researchers should be open about multiply determined phenomenon and not seduced by the “full mediation” quest.

Long Version

Alright, so what exactly is going on in this ritual paper?

In the paper, “four experiments tested the novel hypothesis that ritualistic behavior potentiates and enhances the enjoyment of ensuing consumption – an effect found for chocolates, lemonade, and even carrots.” It identifies two processes that explain why rituals that provide no objective change to the consumption item itself, such as making the same pattern of hand gestures before consuming the item, affect consumption enjoyment. Particularly they find “a delay between a ritual and the opportunity to consume heightens enjoyment, which attests to the idea that ritual behavior stimulates goal-directed action to consume.” Further, they find rituals increase involvement in the experience.

As the paper continues it takes a real phenomenon (rituals), establishes that something psychological is going on, and then provides some enlightenment about what that is. In addition, it provides managers with an intervention idea (e.g. put time between the ritual and the consumption). However, what truly won us over was the discussion section where the authors note that, “ritualized behavior likely encompasses several mechanisms” such as preparatory mindsets, symbolic meanings, social implications, palliative functions, and lay beliefs. Some researchers might see this “multiple mechanisms” account as weak and so these researchers might meagerly attempt to hand wave away alternative mechanisms, in order to protect their precious “full mediation theoretical contribution.” Rather than reject the complications, Vohs and colleagues embrace them.

If you have ever heard a researcher talk about the endowment effect and try to claim that the entire endowment effect is simply one process, you know the problem we are alluding to. When researchers seek to claim phenomena are driven by just one thing for the most part they are wrong: science should be about nuance and openness, and this paper embraces that mindset.

Final Thoughts

With papers like this one (and there are tons of others out there, feel free to link some in the comments section), we can see that one does not need to boldly force a massive theoretical contribution and defend a single process against all others with the insecurity of a teenager and the ferocity of a honey badger. Instead one can approach things powerfully and humbly as Vohs and colleagues did, embracing the modern research movement where a single paper does not need to explain everything as long as it provides a quality contribution.

Troy Campbell is a Ph.D Student at Duke University. To read more InDecision posts from Troy click here. You can also visit his occasionally-updated personal blog People Science where Troy writes about everything from scientific methodologies to Batman.

And if that post was not enough for you, you can check out a related discussion that is even longer in this 2010 academic paper by Zhao, Lynch, and Chen. The paper examines the strengths of mediation analysis as well as the problem of full mediation and the problem of the mediation-as-theory assumption. 

Advertisements

One thought on “Viewpoint: A Great Research Paper Explains A Lot, Not Everything

  1. Pingback: Viewpoint: Why I’m Leaving Academia | :InDecision:

Want to say something?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s