You may already know about this and if you do, have you shared it with other people? Do so. Now.

The National Cancer Institute has an online tobacco cessation program for teens. It includes text to cell phone quit coaching and other helpful and private services.

These automatic communication programs have limited success rates whether using telephone, email, web, or now text and coming soon apps for smartphones. But, they do produce change in a few people and cost almost nothing to operate. Low effectiveness, sure, but incredible efficiency.

If you have a kid who smokes, point her to the website. If you know someone who knows a kid who smokes, point him to the website. WATtapping here can make a very small, but important change.

P.S. I found this at this JAMA article.

Bridget M. Kuehn. Texting Teens to Quit. JAMA. 2012;307(4):351.

doi: 10.1001/jama.2012.4

JAMA. 2012;307(4):351. doi: 10.1001/jama.2012.4

My former office mate, Tim Levine, weighs in on the never ending battle for Better Science with his latest attempt to right the wrongs in journal publication.  Tim’s focus is upon proper thinking with statistics and he provides a nice example ripped from the pages of journal research to illustrate.  Tim’s lesson is easy, fun, and popular, but does require counting, and alas, will probably leave the great unwashed only refreshed, but still filthy.

Tim’s right.  People don’t think correctly with statistics and merely default to the p < .0something as the Eureka moment when there’s more going on.  If only you’d count just a little longer, a little closer, a little harder.

Tim’s most important observation is this:

The null hypothesis is almost never literally true.

Stated another way, you can almost always invent good news even when there is really no news at all.  Null hypotheses are like dead snakes – you still think they could hurt you.  And, when you trip over a rock to avoid the dead snake, you have evidence that snakes are dangerous, even when dead, thus proving dead snakes are a threat which is publishable, but missing the science that fearful people are clumsy.

Tim offers five recommendations that generally counsel patience, replication, and objectivity.  Of course, none of those things will earn promotion or tenure and until we remove that from the equation we’ll continue with the fearful work that populates all journals.  Man, if you cannot break the Laws of Statistical Analysis, you will not publish, but will perish.  You must trip over rocks in your anxiety about the dead snakes in your data; that’s the path to success, but not to science.

Tim Levine (2011).  Statistical Conclusions Validity Basics: Probability and How Type 1 and Type 2 Errors Obscure the Interpretation of Findings in Communication Research Literatures.  Communication Research Reports, 28 (1), 115-119.

DOI:10.1080/08824096.2011.541369

P.S.  You can imagine how well Tim and I got along as office mates, both of us snickering over the foibles, follies, and failures of our colleagues.  It’s a wonder either of us got out of grad school alive.  Mad dogs, Englishmen, and methodologists!

If you’re a Do Gooder a key application of Do Gooding is with  intolerance, discrimination, and prejudice.  You want Other Guys to play nice regardless of race, religion, ethnicity, gender, sexual orientation, ice cream preference, and on and on with the litany of ways we differ from one another.  Hmmmm.  How do we get the Other Guys to Do Good?

The Sincere approach is to know that as a Do Gooder you’ve got Right on your side, so go forth boldly, Do Good speaking Truth to Intolerance and away you go, the Good, Done.  Except such Do Gooding rarely Does Good and indeed, as we will see, makes things Worse.  Almost like an ancient Greek tale, er, meme, where Do Good leads to Done Bad.

Consider this extended example from a test brochure given to Canadians to combat prejudice.

Cracking Down on Prejudice in Our Society

In today’s society, you must control prejudice. In other words, being Canadian means having an anti-prejudiced attitude.  For instance, The Human Rights, Citizenship and Multiculturalism Act prohibits discrimination in employment based on the grounds of race, color, ancestry, place of origin, religious beliefs . . .  Employers have an obligation to create a ‘no prejudice’ workplace, and companies face legal liability for workplace prejudice or discrimination.  The same standards are being set in the education domain.  In fact, a recent government policy initiative by Human Resources and Skills Development Canada requires that educators demand anti-prejudice classrooms. Teachers and students caught displaying racist attitudes and behavior can face serious consequences, such as termination or expulsion . . . There are also social perks to controlling racism – for instance, low prejudiced people tend to be better liked than racists.  The better we are at reducing prejudice, the more we are likely to fit in with today’s antiprejudice norms.  Research studies reveal that people with prejudiced attitudes are at risk of being excluded or ostracized.  In one recent study, most people reported that their social groups at work and at school disapproved of prejudice and racism, and people feared being looked down upon if they made prejudiced or racist remarks . . .  In today’s multicultural society, we should all be less prejudiced.  We should all refrain from negative stereotyping.  It is, after all, the politically and socially correct thing to do, and it’s something that society demands of us.

Now, you may chose to rewrite portions of this excerpt that you find heavy handed, charmless, or bleak, but if you don’t change the persuasion force behind the words, you will fail at Do Gooding and make things worse.  The key problem with this brochure is the Attributional impact it has.  The motivation for tolerance comes not from an Internal Attribution to ourselves, but rather from an External Attribution from social norms and potential punishment from other people.

Legault, Gutsell, and Inzlicht (pdf) randomly assigned Canadian students to different brochures in their persuasion experiment that tested Attributional force – Internal versus External – for its effect on attitudes.  Participants either got that External Attribution like the excerpt above or an Internal Attribution brochure like this.

Why it’s Important to Reduce Prejudice in Our Society

As a society, we hold the virtues of tolerance and nonprejudice in a very special place – they are important because they increase open-mindedness and social justice.  Social justice is  the vital ingredient in a free, fair, and peaceful society.  When equality and equity among human beings are achieved, there is less reason for any group or individual to be unhappy . . . It is also important to be nonprejudiced because it is so     interesting to interact with and learn about people from other cultural and social groups.  We live in a wonderful and diverse cultural community.  That diversity makes our society great because it brings a wealth of knowledge and experience together. When we let go of prejudice, the rich diversity of society is ours to enjoy . . . Not to mention, being open-minded is a real advantage to our mood and well-being.  When there is less racial and cultural tension, people are happier and healthier, and better able to do the things they enjoy . . . You are free to choose to value nonprejudice.  Only you can decide to be
an egalitarian person . . . In today’s increasingly diverse and multicultural society, such a personal choice is likely to help you feel connected to yourself and your social world . . .

Again, you may rewrite what you find saccharine, effete, or silly, but as long as you maintain that Internal focus on self motivation for behavior, the manipulation will produce the desired effect.  Consider this bar chart from just one experiment.

Now, more importantly, the statistical outcomes.

As illustrated in Figure 1, participants in the autonomy-brochure condition displayed significantly less prejudice than did those in the no-brochure condition, F(1, 66) = 14.49, p < .001, eta2 = .18. Conversely, those who read the controlling brochure actually demonstrated greater prejudice than those in the no-brochure condition, F(1, 66) = 4.34, p < .04, eta2= .07. As hypothesized, using control to motivate prejudice reduction backfired, and was more detrimental than not motivating participants at all. The support of autonomous motivation to regulate prejudice, however, caused a reduction in prejudice.

Sure, it’s statistically significant – here’s your sign – but note the effect sizes in those eta2 values.  They translate into Medium+ Windowpanes, around 30/70.  And, see the detail.  The Internal Attribution brochure produces an obvious benefit over the No Message Control and the External Attribution brochure produces a less obvious, but still near Medium Windowpane harm compared to the No Message Control.  Thus, attribution moves people in opposite directions compared to Control, one producing greater benefit, the other greater harm.

Legault et al. ran a second experiment to replicate and extend the basic finding and also conducted useful moderator studies with path models to refine the conceptual model, but I want to hit the main point.  Doing Good requires more than a pure heart, social courage, and a brochure.  Sincerity alone can make things worse, casting yourself into the ancient Greek nightmare where you kill what you love.  If you do not understand and apply persuasion principles properly you will kill your father, marry your mother, but live forever in tale, narrative, or meme!

As we’ve noted several times in the Persuasion Blog and the Primer, Attribution Theory is a powerful persuasion play that is almost always available as either an active tool or an important element in understanding the Local.  How do you want the Other Guys to explain their world?  Failure to ask this question or to answer it correctly dooms change whether from Do Gooders or Do Badders.

Realize here that the External force of that first brochure not only activates an external locus of control, but could also easily elicit Reactance, that Like Hell!, knee-jerk response people display when they perceive an unfair restriction on their actions.  Scan over the comments at the Free Republic website, a conservative net community, about this very study and see examples (most ironic – gee, conservatives with a sense of humor?) of that reactant effect.

Now, if you are a zealot, you find satisfaction of some sort when your political adversaries react against you, but, hey, you aimed at changing people, you knucklehead, and all you succeeding in doing was alienating not only adversaries but a bunch of Other Guys as well.  You may sniff about the Evil Opposition, but then, of course, you are running afoul of the Rule:

Great Persuaders Don’t Need Rich Uncles, Kindness from Strangers, or Third Party Vote Splitters.

Here the Rule means only muggles complain about their opponents as if you can win only when you have no competition.  But, then, that’s the basic tension between Sincerity and Persuasion.  When you know you have Truth or Beauty or Justice it is difficult hide it under Peitho’s robe.  You don’t hide your light under a basket!  Yet, as this report demonstrates, along with other examples from literature as described in the Blog and Primer, beacons of light are often unpersuasive and worse still can easily activate more dissent, confusion, and conflict.

You need to understand the difference between what you value and how you persuade on those values.

Legault L, Gutsell JN, Inzlicht M.  Ironic Effects of Antiprejudice Messages: How Motivational Interventions Can Reduce (but Also Increase) Prejudice.  Psychol Sci. 2011 Dec 1;22(12):1472-7. Epub 2011 Nov 28.

doi:  10.1177/0956797611427918

A valued network pointed me to a study by Facebook engineers investigating the diffusion of information through the Facebook social network with an experimental design!  Yes.  Experimental.  The engineers randomly assigned nearly 250 million account holders (about half of the Facebook universe) to either receive a news story with a link from a friend or not.  They then tracked how that information diffused through the network.  Here’s a visual example of this t-test design.  Click to enlarge.

Over a two week period, Facebook engineers would assign a Feed story with link to one person while randomly assigning two other people as a no Feed control.  They did this repeatedly for two weeks over the 250 million selected accounts.  They then tracked how the link got diffused through the friend networks.

Talk about a big data set.  Huge, Jerry.  Huge!  Here’s the key finding.  Click to enlarge.

Look again at those underlined values.  Those randomly assigned to get a news story with URL in their feed showed a diffusion rate of 0.191% while those in control showed a rate of 0.025%.

Make sure you understand those little percentages.  About two tenths of one percent diffused the link in the treatment condition.  Thus, not even 1% of treatment people passed the link along.  Not even one half of 1%.  Barely 2/10 of 1 percent.

You can read the paper which I found at scribd.com.  Given all the kinds of variables you can generate from a database this large, there’s more to the paper with lots of graphs, but they depend upon relative ratios rather than absolute effects.  When you read that Outcome paragraph just above what probably struck you first was that huge relative risk ratio of 7.37.  A Large RR would be 4.50, so this is absolutely Stupendous.  The rest of the paper focuses upon those RRs while blithely skipping past the absolute diffusion rate.  I remind you:

0.191%.

Please think like a maven here and not a muggle looking to move your 401k into the Facebook IPO. What’s the TACT here?  What’s the Change you want from the Other Guy? Clicking on a link with a mouse and forwarding it to your Facebook network.

I’ve derided the social media TACTs before as the smallest units of communication you can create.  Realize that this huge database quivers with the digital signals of the twitch, the click, the WATtap.  That’s all they collected.  A click.

Wouldn’t it be nice to see if this clicking behavior drove other behaviors?  Like going outside of the Facebook network and using email, phone, or, gasp, face to face contact with other people about the information?  Like making a financial contribution?  Like volunteering.  Like calling law enforcement with a tip?  Like any real world, practical TACT?

And, shootfire, even fooling yourself briefly into thinking a click is as fundamental as kiss or a sigh as time goes by, what diffusion rate do you get?

0.191%.

Mavens, recall the fabulous Al Gore Facebook day.  He made a great scarcity play with a one day extravaganza on Climate Change through Facebook and inveigled everyone to donate their network to the cause.  How’d that go?

0.004%.

All folks had to do was click once and Gore’s machine would do the rest to then propagate the Gore message to your friends.  One little click.  And he got 0.004%.

Sure, we are mixing different variables in the comparison between Al Gore’s project and the one reported here.  The Facebook experiment created a forced exposure while you have to go find Mr. Gore on Facebook that day.  But we don’t have good research on social media, so we’re stuck doing these mental comparisons of apples and oranges, bytes and packets.

Social media has proven itself to be a toy, a pastime for killing time, phatic communication par excellence, a quick and easy way to be quick and easy, social without being sociable.  Sure, imagine a big event like the assassination of JFK or the Challenger explosion and doncha know that Facebook and twitter would light up.  But, so what?  Where’s the TACT?  What’s the Change in the Other Guy?

WATtapping clicks.

P.S. For what it’s worth, this paper apparently has not been published in a peer review source.  It’s been submitted to the Cornell University Library at arXiv.org.  I suspect it will get published in a good peer review outlet given the experimental design in a natural, if virtual, setting.  My practical persuasion concerns remain.  Facebook is a failure for Changing the Other Guys.  Focus on the kiss and the sigh, mavens.

As you are walking through a university commons area, I approach you, clipboard in hand, maybe wearing a blue polo shirt, definitely smiling. “Excuse me, would you be willing to participate in a short marketing study. We need help rating pictures of people that we’ll use in a later study.” I’ve got really white teeth and it’s a good day, so you agree.

I give you the clipboard with the survey. The cover sheet has some questions about demographics, then you turn the page and see the picture. You are randomly assigned to see one of three images: a lamp (Control), a fit young woman (Positive), a fat young women (Negative). The women look like this.

You then rate the picture on a variety of dimensions. That’s it; I told you it was short! Now, please go over to that other researcher, yes, the one in the blue polo shirt with the big white smile. That smiley guy thanks you and holds out a bowl of candy as a little appreciation for your participation. You take as many wrapped candies as you please and move on.

Mavens, this is so easy. You’re testing the priming effect of those images, Fit or Fat, on candy taking. When people see the Fat image, they take fewer candies because they don’t want to look like that! Easy-peesey. Let’s get technical and read what the researchers found.

Results revealed only a significant effect of prime (F(2, 53) p 3.88, p = .03). In support of our prediction, planned contrasts revealed that those who saw the image of an overweight person took significantly more candies (M = 2.2) than those in the healthy weight (M = 1.4; F(1, 53) = 7.03, p = .02) or in the neutral control condition (M = 1.5; F(1, 53) = 4.77, p = .04). There was no difference between those in the healthy weight and neutral conditions (F(1, 53) p < .25, NS).

Great Cazart! Read that again. People took MORE candy when they saw the FAT image?!? That can’t be right. Or if it is we’ve got another one of those puny Observational Fairy Tale effect sizes. Do the math. Take that F(1, 53) = 7.03, derive the square root of 7.03 which converts that F to a t, then get the formula and convert it to a Windowpane and find . . . jeepers, a Medium effect, 35/65.

Huh. Fat images prime more candy taking. And obviously, practically so. Campbell and Mohr provide an extensive, Full Cycle, five study package to demonstrate why this is true and how difficult it can be to change Other Guys, especially with common sense. I will not detail each study (and the several pretest projects) Campbell and Mohr report and will leave that as a useful and highly recommended activity for any and every maven out there. This is excellent research. But to our immediate purpose.

Campbell and Mohr combine the theoretical and the practical in this research working with stereotype activation and healthy eating. They explore how information about the Bad Behavior can have the perverse effect of increasing rather than decreasing the Bad Behavior. We’ve seen recurring examples of this failure among the Lifestyle Drum and Bugle Corps with their rancid warning labels that elicit Reactance rather than Compliance. And, like virtually all human action, our behavior is complex and the result of many factors. Campbell and Mohr’s work adds stereotype activation to the list. Let me quote them on their key conceptual discovery.

Increased accessibility of the countervailing health goal and increased accessibility of the link between the behavior and membership in the stereotyped group both limit the effect of stereotype activation on stereotype-conducive behavior.

De-geeked, this means that prosocial persuasion perversely does two things: 1) it activates a goal for the Bad Behavior and 2) it activates a feeling of membership in the Bad Behavior group. Thus, if you see a picture of a Fat person, you prime the goal of Getting Food and also activate a link between your Self and belonging to the Fat Group.

While Campbell and Mohr may not agree with this, I interpret the outcomes as entirely consistent ELM and Low WATT processing. The results from their Experiment 4 run this basic finding through what I would consider High WATT versus Low WATT conditions. Under Low WATT that basic stereotype activation occurs and people take more candy following the Fat image; under High WATT conditions, there’s no difference.

People are simply not High WATT Objective processors of all the relevant information about their weight, their health, and their normal behaviors. They are floating through the mess of life when, as in Experiment 1, somebody with a bright white smile glides up beside with a clipboard and a request to help out very quickly. Sure, they are thinking enough to accurately complete the survey and examine the picture, but they avoid all that cognitive work in the Long Conversation in the Head over calories and exercise and ideal weight and on and on. In that Low WATT state, Cues like stereotypes, group membership, and priming effortlessly guide thinking and, if the Local provides it, easy, almost automatic, action. Here, look at this picture; want a candy?

Like virtually all public persuasion, most prosocial persuasion hits Other Guys in a wide variety of Locals. Thus, the same message about health may elicit Reactance in these Guys while producing this boomerang Stereotype in those Guys. Often times mavens think the problem is competition from other sources. Hey, the Food Police struggle against Big Food and Big Marketing with their Big Bucks, but if you know what’s going on you realize the larger problem is the Other Guys themselves. They are so variable in their WATTage and the Local changes so rapidly between and across Other Guys that finding one message or type of message that makes a practical difference is just damn difficult.

Some people think that Tailoring is the way to go, but even this kind of specificity misses the mark. Usually Tailoring aligns the persuasive content with demographic or psychological characteristics of the Other Guys. What’s missing is the persuasion tailoring where you match the message to the WATTage and the Local the Other Guys variably confront. That’s the essence of a Persuasion Engine. Sure, you can swap out pictures in a message that match the skin color, age, sex, gender orientation, and on and on, but those qualities are considerably less important and powerful than understanding the Other Guys’ WATTage and whether their Local predisposes them to Stereotype Activation, Reactance, or something else. A Persuasion Engine approach aims to match message and Other Guys on the basis of persuasion principles.

Well, now I’m babbling just a bit. Let’s get out of here.

Campbell and Mohr shine a light on the Low WATT problem of Stereotype priming wherein you make things worse with your prosocial effort. They also display a great illustration of that Full Cycle research that takes a concept and runs it through the research wringer, testing a basic X-Y relationship and demonstrating its mediators and moderators. This is yet another example of persuasion science.

Persuadin’ ain’t easy, mavens. And Campbell and Mohr explain why.

Margaret C. Campbell and Gina S. Mohr (2011). Seeing Is Eating: How and When Activation of a Negative Stereotype Increases Stereotype-Conducive Behavior. Journal of Consumer Research , Vol. 38, No. 3 (October), pp. 431-444

DOI: 10.1086/659754