Healthy Influence – Persuasion Blog

Here’s an interesting field experiment that tests the impact of future facial appearance on quitting tobacco. You can take an image of a person’s face and run it through a software program that will show the effects of aging and of smoking. Consider this example.

Grogan et al. recruited a group of 70 young women who smoked, but wanted to quit. The women were randomized to a Control Group with Usual Care or a Treatment Group that received the Usual Care, but also looked at a series of images that portrayed their faces as they aged under conditions of Smoking or Quitting. The Grogan team collected TpB measures at three time periods (baseline, post intervention, post plus one month) along with self report measures of smoking, and a carbon monoxide (CO) breath test as a physiological measure of smoking.

Quick summary: All participants want to quit and are given standard clinical support, but half view those age progression images of their faces as Smokers or Quitters. What outcomes do you expect from young women watching their faces get old with and without smoking? Let’s start with the TpB models.

Compared to the Control women, participants who viewed those progression face images showed Medium to Large favorable increases in Attitude, Norm, Control, and Intention to quit smoking, ranging from 35/65 to 25/75. Attitudes changed the most. The Windowpane effect sizes are consistent with the literature and demonstrate that the progression viewing had an obvious, practical impact on TpB elements. So far, so good. Now the self report smoking.

The Grogan team collected three different kinds of self reported smoking and all three showed Small to Medium positive changes for the Treatment women compared to the Control. Averaged across all three types of self report the effect size appears to be a Medium Windowpane, about 35/65, another obvious, practical differences. So far, so very good. Now, the objective measure of smoking, that CO score.

To get this measure each person has to blow into a machine much like those DUI devices. It measures how much carbon monoxide is in the person’s breath which is a reliable indicator of recent smoking. The Grogan team found no difference between the Treatment and Control groups and worse still there was no CO difference from Baseline to Post for either group.

Quick outcome summary: great self report results that favor the Treatment Group on TpB and self reported smoking; terrible results for objective indicators of smoking.  Here’s their Table 1 that displays all results for your viewing pleasure.  Click to enlarge if needed.

If you read the smoking literature these mixed findings are a common and disheartening outcome. Many studies that include both self report and physiological measures of smoking produce these mixed findings with the self report data coming in like gangbusters and the physio data falling dead on the floor. If you’re an ex-smoker like me, you know why this happens: Smokers lie to themselves about quit success. If you actually smoke one less cigarette that day, you tell yourself it was five. And, if that’s what you’re telling yourself, imagine what you’re telling a member of the Grogan team.

You find this same problem in all the addiction research which is why everyone who gets funding collects both self report and objective physio measures. Early on, everyone was flying with self reports and discovering all manner of proven interventions that caused quitting – we’ve got the data!!! – but then these interventions never produced the same positive outcomes in the field. Addiction research has become more suspicious or scientific or both depending on your outlook as a result.

My persuasion analysis of the Grogan team study is that the progression images are fabulous persuasion Cues. They generate Medium to Large changes in TpB variables, especially Attitude, but also Norm and Control . . . which should be a red flag to anyone who’s thinking about this data. Why should information about appearance whether Argument or Cue cause any change in Norm or Control? It’s a pure Attitude play and if appearance is also generating noticeable changes in Norm and Control, my radar goes off. That doesn’t make sense, especially if people are going High WATT, which, of course, they are not. It does makes sense, however, if people are Cue-ing off the progression images. Their emotional response to the image progression causes an immediate change in everything related to quitting, the kind of global response you can get from Cues, but rarely from Arguments in this Local.

Since it is called the Theory of Planned Behavior it sounds like a Central Route play all the way, but realize that you can change Attitudes, Norms, Control, or Intentions with Peripheral Route Cues. You need to see that the age and smoking progression software doesn’t cause people to think about quitting, but about getting ugly. They may get incredibly thoughtful while viewing those grim images, but that’s not elaboration activity over quitting behaviors like not buying tobacco or hiding it or giving it to someone else to hand out or using that awful gum or whatever tactic that actually addresses the behavior. Thus, these progression images cannot be Arguments for any quitting behavior, but only address motivational elements that may precede quitting. That distinction is crucial.

The intervention never directly connects the software images to actual quit behaviors. It, at best, hopes to create some kind of magical motivation in a chain from

Ugly Progression Images -> Disgust at Getting Ugly -> I Quit

Realize that step from Getting Ugly to I Quit is not a straight line, but includes several other intervening concepts that are not addressed here. Realize the leap from Step 2 to Step 3 is Cue driven, like one of those crazy daredevil drivers leaping over a line of cars packed between the ramps. You think the energy from that image progression will cause you to clear all the obstacles, but the carbon monoxide outcomes prove that largely wrong.

This age progression software has been around for awhile and several tobacco control groups have touted it as Cool Beans (here and here). Also notice the great confidence and even some data with Really Big Effects!!! Except, as Grogan’s team demonstrates, the intervention only affects everyone’s perception, but not their bodies. The disconnect between the TpB data and the carbon monoxide outcomes underscores it.

Now, let me be quick to applaud the Grogan team for their restraint in interpreting their results. They do not try to sell this as a major breakthrough and indeed express appropriate concern over the CO results. They also suggest interesting lines of future research and in general do a great job of discussing mixed results. Kudos.

I still want to hammer on those mixed results. Zealots will seize upon the great self report data and offer cheers and huzzahs for image progression as a cure for addiction. The data taken as a whole, clearly do not provide that support. They strongly suggest you can whip people around with those images and generate a very large emotional response that floods immediate cognition. This may not be useful with addicted behaviors, but you take nicotine out of the equation and image progression may work. Combine it with the opportunity for immediate action and you’ve got a great little persuasion tool.

Think about those cosmetic areas in large department stores. Take a picture of a women in the shop, then run it through an age progression series with and without Our Makeup Treatment. Or, hey, think about mortality salience from Terror Management Theory. Don’t have people merely think about death, but show them the inevitable progression of age on their faces – think that might motivate Worldview actions? How about job counseling? Show young people an age progression of themselves and then start talking about future career development. Think about selling life insurance with this.

In just this quick sketch, it’s obvious that the age progression software could have powerful persuasion effects when the Local makes a strong link between self perception of aging and an immediate consistent behavior. I believe, however, that the Grogan et al. data demonstrate it’s unlikely to have much impact on addiction. It’s just so damn hard to generate a psychological change that lasts long enough and is strong enough to continue when that next physiological craving arises. Jeepers, I have trouble reading the tobacco control literature without wanting a Lucky Strike filterless between my fingers and I quit 30 years ago.

Grogan, Sarah; Flett, Keira; Clark-Carter, David; Conner, Mark; Davey, Rachel; Richardson, Deborah; Rajaratnam, Giri
Health Psychology, Vol 30(6), Nov 2011, 805-809.

doi: 10.1037/a0024745

P.S. A Peitho nomination to whomever invented this progression software that scientifically shows the effects of age, smoking, drinking, celibacy, or any other behavior targeted by the Lifestyle Drum and Bugle Corps.  How could you possibly do anything other than fake the math?  I mean, where’s the data that proves aging follows some Photoshop algorthim?  Yet, you’ve sold smart people into thinking the software portrays an accurate reflection of facial appearance given different conditions. People will believe anything that comes out of a computer. I’m going out on a limb here, but has anyone done a study that takes people who are currently older, feeds a young picture of them into the software, then compared the output from the computer to their actual older faces?