JAMA Internal Medicine provides a research report on the effect of decision aids on mammogram discussion and usage among women over the age of 75. It is a simple study that exposes complex persuasion. Let’s start simple.
Researchers designed an 11 page decisional aid (pdf Decisional Aid Supplement) with text, charts, graphs, and numbers as rates or percentages. They recruited 45 healthy women over 75 in a clinical setting and asked them to complete a pretest and a posttest about breast cancer and mammography. In between, all 45 women read over the decisional aid, interacted with it as designed, and provided an evaluation of the decisional aid itself (comprehension, bias, usability, and so on). The crucial outcomes measured at pre and posttest were knowledge, intention to get a mammogram, and decisional conflict (how uncertain they felt about the issue). The results.
Participants’ knowledge of the risks and benefits of mammography improved: women answered on average 1 (interquartile range, 0-2) more questions correctly on the 10-item index. After reading the DA, fewer participants intended to be screened . . . Decisional conflict declined after reading the DA but not significantly.
So, the Other Guys got smarter, rejected mammography, and tended to feel more certain about their decision. Here’s the table for the intention to get a mammogram.
You see clearly that the decisional aid really moved the dial with these 45 women from pre to post. It either decreased intention or increased uncertainty. But, it in no way motivated these women to sign right up for a mammogram.
Interestingly the women perceived the decisional aid as somewhat biased against mammograms.
Forty-two percent (18 of 43) found the information balanced, 42% (18 of 43) found it slanted toward not getting a mammogram, and 16% (7 of 43) found it slanted toward getting a mammogram.
Of course, there are serious weaknesses in this study. Only 45 Other Guys in a convenience sample with a pre-post test! This is a repeated measures t-test design without a control group. Any persuasion panther, much less scientist, who did “message testing” like this would deserve any beating she gets. One sample with 45 people ain’t much to go on. Replication. Vary the contents of the decisional aid. Provide other comparisons (recommendations from the AMA or a US health task force, for examples).
Science aside, now consider the persuasion.
What’s a decisional aid? A decisional aid is medspeak for an interactive sequence of information and questions that guides Other Guys through the known science about a particular medical issue. Here the decisional aid takes Other Guys through breast cancer risks and mammograms.
This aid begins with an orientation about the topic (breast cancer) and headline scientific knowledge in both text and numbers written in a very easy to understand style. The aid then asks a series of health questions (height and weight, recent hospitalizations, diabetes, etc.) about the respondent and she has to write her answer to the question and convert the response into a number written on the right side of the page. At the end of the 10 or so health questions, the woman then totals up her score and sees this.
The aid continues with a lot more numerical information very simply designed and reported about risks and outcomes related to breast cancer and mammography. Here’s an example.
An infographic or Pretty Picture, but more scientific than usual. It is an exact visual representation of numbers, specifically, rates. Now, the aid presents one final interactive element. This is the last page in the aid before a question about intention to get a mammogram with completes the 11 page decision aid.
Each woman takes a moment reflect over the preceding 9 pages of text and numbers then checks off any of the Pro or Con thoughts already printed on the paper. She can also then add her own thoughts in those blank lines.
[ELM/HSM Brownie Point: Yes, that is what you think it is and we will get to it. Now, spin the propeller on your head and fly back into the post.]
From this description you see the sequence of information and interaction. The aid presents data, facts, numbers, conclusions, and so on from the scientific peer review literature. The aid asks the Other Guy to provide personal health information that is then immediately scored for comparison against an easy to understand summary in graphic form. Then more information and that Pros and Cons interaction.
Now, you can call this a decisional aid.
You can also call it push surveying.
You can call it choice architecture from Nudge.
Or you can call it a Biased Central Route persuasion play.
Each label means the same thing: sequenced information and interaction that guides the Other Guy through an issue. In this particular instance, the WATTage is probably quite high for each Other Guy, the information is nothing but Argument, that summary health score manipulates a biased schema, and the Pros and Cons interaction provokes the Long Conversation in the Head (and documents it) that doubtless follows the flow of the Arguments and that biasing treatment with the personal health score.
In this application, the decisional aid persuaded these Other Guys to reject an action they’ve been taught to accept for most of their adult lives. Think about that. These women were at least 75 which means when the big push for mammography began in the late 1970s, these participants were adult women. They’ve heard nothing but Get A Mammogram all the time for 30 or 40 years. Then they go through a decisional aid and reject expert advice, habit, and norm.
And just to reinforce that point, the researchers tracked the health records of these women for six months to see if they discussed mammography with their physicians. Read this interesting comparison.
Fifty-three percent (24 of 45) of participants had a PCP note documenting a discussion of the risks and benefits of screening within 6 months of participating compared with 11% (5 of 45) in the previous 5 years (P < .001).
Before the decisional aid, 11% had every had a noted discussion with their physician about mammography. After the aid, 53% had that discussion. That’s a near Large Windowpane increase. The actual screening rate also decreased from 83% before the aid to 60% in the year following – a Small+ Windowpane. Thus, you see changes in future behavior from this decisional aid. And, all those behavioral changes go against that 30 or 40 year history.
The persuasion point is not that this decisional aid is good or bad, but rather that is captures basic persuasion principles. If you changed the information from one cited study in decisional aid to a different source, you could change the slant of the decisional aid in a different direction. You could manipulate the scoring on that personal health questioning to either increase or decrease the score as with push surveying. You could drop either the Pro or the Con thought listing and increase the biased processing.
This thing works as persuasion because it puts Other Guys on the Central Route.
A research assistant (M.C.G.) administered the pretest survey. We then asked women to come to a routine appointment with their PCP early to read the DA. After reading the DA, which takes approximately 5 to 10 minutes, participants attended their scheduled visit.
You take each Other Guy one at a time, a research assistant meets the woman, takes her to a separate room, explains the procedure and then gives the woman the decisional aid. I’d argue that this is likely to produce a thoughtful high WATT consideration of the decisional aid.
With receivers who are so tuned in, the aid then provides Arguments, information that bears on the crucial merits of cancer and testing. It further encourages more thinking about those Arguments with that summary health score, but with a bias. If your score is low, the that graphic arrow tells you that you need a mammogram while if it is higher, you don’t. Consider that these are people over 75 those health scores will not be 0 or 1 or 2, but rather higher and the inevitable biasing conclusion provided in the graphic that mammograms aren’t helpful. Finally, the aid locks in that biased conclusion with that Pros and Cons task with drives the Long Conversation in the Head. Technically, this is called a “thought listing” and is used to measure elaboration activity in research. Here, it is a persuasion play.
Of course, among these researchers this thing is just a neutral, objective, and scientific presentation. Dispassionate. No one is telling you what to do. Think for yourself.
And, of course, this is nothing but persuasion. Change the contents of the information to other “scientific” sources, change the slant of the infographics or that health score or the thought listing task and I guarantee you will get different persuasion outcomes. Which one is the real science?
When you are thinking and doing what a medical researcher thinks you should think and do, it’s decision making. When you look at it with the cold heart of a panther, baby, it’s only persuasion. Everything about the design of a decisional aid shrieks with persuasion and manipulation.
Any decisional aid relies exclusively upon what the designers believe is the proper sequence of information and questioning. If you happen to have a large financial stake in the outcome of a decisional aid, the aid will look different compared to a decisional aid created by an NIH task force given a set of predetermined rules for evaluation. As we’ve noted before, Big Pharma runs “decisional aids” in the form of push surveys that have that sequence of information and questions that subtly guides the Other Guys to a desired decision. Also, a notable form of the Nudge, the thing called choice architecture, is a decisional aid that sequences information and questioning.
The really fun part of this play is how it gets hidden in the label, decisional aid. Call it that and you are scientific. Call it, instead, a push survey or a Nudge or a biased Central Route play and you are both scientific and persuasive.
As a practical persuasion play, consider a decisional aid. Give the thing to Other Guys in a physical setting that looks thoughtful, scientific, technical like a doctor’s office, but for your specific Local. Have a research assistant administer the decisional aid and make sure he or she is dressed to look smart, objective, but friendly. You must build an Authority Box for the Other Guy.
Now, design a decision aid with as much simple science as your TACT possesses. Don’t do any fancy Pretty Pictures like Tooth Fairies or fMRI pixies. Essentially use graphics instead of numbers to express exactly the quantity. Provide citations and references whether to the New England Journal of Medicine or the New York Times or the New New Thing. Include some kind of self report that the Other Guys score and count for themselves, then provide that biasing summary graphic that pushes the Other Guy to the high or low side as you wish. You can manipulate that three ways.
1. The specific questions you ask.
2. The score values you provide for each answer.
3. The label range of Low versus High on the summary graphic.
In this example, the self report of health status asked questions that were guaranteed to get “more” or “higher” responses that were then cut to count higher. The summary graphic at the bottom also provided less space for the “low” side of the graphic and more space for the “high” side of the graphic. By design, most women had to score “high” on this play which required them to count the change against mammography.
Finally, run everyone through that thought-listing play with Pros and Cons. You could further bias this by including, for example, Pro statements that no one would think and Con statements that everyone thinks, thus pushing a negative Long Conversation in the Head. Or reverse it and provide Pro statements that everyone thinks and no Con statements and get a positive Long Conversation in the Head.
You see the amount of effort this persuasion play requires, but remember, you’re pushing the Other Guys on the Central Route and that requires effort from both you and the Other Guys (although you have to hide your effort so the Other Guys doesn’t make an External Attribution to it). You expend effort on building the Authority Box, but you make it look natural, easy, and normal. Then that decision aid encourages High WATT processing with the biasing plays built into it.
The payoff from all this effort is not just with immediate change from the Other Guys, but their persistence, resistance, and future action. Central Route change is strong and elicits an internal structural change in beliefs and attitudes that the Other Guys maintain when they leave the Authority Box. They are now operating under their own steam. With Cue-based change on the Peripheral Route, you’ve lost the change as soon as they leave the scene and you have to hit them again with another Cue the next time. With this biased Central Route decision aid play, you work hard once.
This turned in ways you probably did not expect. Hey, this is the science of cancer and testing, right? What could be more scientific than a scientific presentation of the science? The researchers themselves display a blissful ignorance of the persuasion filling their sails and think this is the real Dr. Doctor MD.
But since we’re beyond good and evil here and beyond the perimeter with persuasion, we can see more widely. Properly constructed and executed, a decision aid in an Authority Box is a killer persuasion play.
Schonberg MA, Hamel M, Davis RB, et al. Development and Evaluation of a Decision Aid on Mammography Screening for Women 75 Years and Older. JAMA Intern Med. 2014;174(3):417-424.
P.S. So. Is there any science here? Well, you can still find guys in White Lab Coats who will testify like a prophet in the desert about Screening Now, Screening Tomorrow, Screening Forever. And they publish in JAMA, too.
My read of the research literature is that unless you have some kind of diagnosed problem, screening as a population tool (Screening Now Tomorrow and Forever), is almost always useless and dangerous; healthy people will get killed from the “false positive” and the unhealthy will survive based on treatment quality and not the screening. Catch It Early is pretty much a persuasion meme and not a well demonstrated piece of science.
P.P.S. And even with the nice graphical presentation of numbers, we know that whether you use numbers or icons, people still don’t understand the thing better. Remember this experiment? Smart people think you can explain complex ideas with simple pictures and you can’t. I’d argue again that the effect of this decisional aid is persuasive, not informational.