Start with a convenience sample of UK adults who drink a bit too much. Recruit them into a study, then send all of them four weekly email messages.
The remaining 101 (66 female) participants (n = 29 in Condition 1; n = 27 in Condition 2; n = 23 in Condition 3; n = 22 in Condition 4; see Table 1 for sample characteristics) were sent four weekly e-mail messages that started with the words, “Before taking part in this study you were drinking more alcohol than is recommended by health experts . . .”
You see Uncle Norm in all messages. The messages provide a standard of what’s going on with drinking. Then you see the inclusion of Ms. Comparison as the Other Guys encounter not only information about Uncle Norm, but gets guidance from Ms. Comparison about how they stand against the norms.
To test the impact of these 4 persuasion plays, the researchers provided information links in the weekly emails. They simply counted how many pieces of information the drinking Other Guys took as the measure of persuasion effectiveness. So the key What element in the TACT is information seeking.
I want to focus on the least favorable count the researchers reported. As often happens in field experiments, some Other Guys drop out. How do you count them? Some researchers drop the drop outs then do the count which tends to inflate the effectiveness of the experiment. Other researchers will report both sets which includes the drop outs. That latter approach is sometimes called Intention To Treat and approximates the living Local better. In the mess of life if you run a persuasion play and Other Guys walk out in the middle, you’re stuck with reality. I prefer Intention To Treat counts and report that here. Look at this table (from the Supplemental Materials, not the published paper).
The first three rows show virtually no impact for Uncle Norm, with or without Ms. Comparison on information seeking. The Other Guys randomly assigned to these three conditions clearly blew off the available information. But, the Other Guys in the last row behaved quite differently.
Again, those in the rank comparison condition (Condition 4) were significantly more likely to request information than those in Condition 1, χ2(1) = 11.49, p = .003, OR = 7.50, Condition 2, χ2(1) = 18.12, p = .003, OR = 31.20, and Condition 3, χ2(1) = 12.82, p = .003, OR = 12.60 . . .
So, compare treatment 4 to those in 1, 2, or 3 and you open Huge Windowpanes. Remember a Small odds ratio is 1.5, Medium 2.5, and Large 4.25. And what is that successful message?
You drink more than X% of people like you.
You see Uncle Norm in that we are dealing with standard of behavior, in this case about drinking. And, we also see Ms. Comparison where each Other Guy gets compared in simple numbers (either percent or count) to people like them. This play is explicit about both the norm and the comparison. Instead of leaving it up to the Other Guy to do the work, the play directly tells them that they are drinking more than Guys like you.
As we consider this experiment, start with limitations. One experiment, no replication. About 100 Other Guys are in this study. And, we’re only talking about WATtapping a link in a convenient and repeated email for information seeking. That’s a pretty simple and easy TACT and has nothing to do with actual drinking behavior.
That noted, you do see how the combination of Uncle Norm and Ms. Comparison can strengthen the persuasion effectiveness of this kind of play. We know that norms are crucial forces in human thought, feeling, and action, but you need to make that norm clear and obvious to Other Guys. Merely stated official recommended guidelines does express the norm, but it requires the Other Guys to do head work about how the norm relates to them.
Inclusion of Ms. Comparison activates the importance or relevance of the norm to a specific Other Guy. Ms. Comparison specifically calls out the Other Guy and says, hey, you are doing something more (or less) than everyone else! The comparison does not criticize the Other Guy; it just makes the norm and the deviation from the norm explicit.
For me these data also illustrate the Low WATTage of this play. All of the messages contain enough information for people who are already drinking a lot to reason out that they are indeed drinking a lot. That should provoke the Long Conversation in the Head and they should of their own accord come out on the other side thinking, “I will drink less.” But, the results demonstrate even when you provide sufficient evidence, strong Arguments in Dual Process parlance, if the Other Guys aren’t High WATT, they won’t do this thinking.
You need to simplify the message so the Other Guys don’t have to do much thinking at all. The combination of Uncle Norm and Ms. Comparison accomplishes that.
You also see the Low WATT operation of these Cues with the TACT. All the Other Guys had to do was hit a link in a repeated email message. Literally just WATtap. And, in 3 of the 4 treatments, virtually none of the Other Guys could exert that little effort.
While I’d like to see replications from different teams with different TACTs, I’m pretty sure this is just a good idea for practical persuasion. You cannot go broke underestimating the WATTage of Other Guys, so you often have to employ Cues to get any change started. And even with the Cues, you must make them obvious enough just to get Other Guys to WATtap a link.
Keep it simple, sweetie. And obvious.
Taylor, M. J., Vlaev, I., Maltby, J., Brown, G. A., & Wood, A. M. (2015). Improving social norms interventions: Rank-framing increases excessive alcohol drinkers’ information-seeking. Health Psychology, 34(12), 1200-1203.