come say hi at SPSP – coffee, sugar, and climate change – open and reproducible science

This week, at the society for personality and social psychology’s annual convention, I’m speaking in a symposium, rethinking health behavior change. I will talk about a study in which we* tested strategies to help people reduce the amount of sweetener added to their daily coffee (ideally without reducing enjoyment of it**). I’m also presenting a poster on how people talk to others about making behavioral changes that affect the environment.

One thing that excites me about these studies–they represent my first (admittedly clumsy) attempts at being completely reproducible and open with my science. Datasets, R analysis scripts, hypotheses, and all other study materials are publicly available***, and were preregistered****.

Openness and reproducibility in science fascinate me—both as a topic of research and as a guiding principle for my own research. Since starting graduate school, I have preregistered (nearly) all of my studies and have been working toward making the entire process transparent. I’ve also been learning how to write reproducible code in R. It has been challenging… you know, for the obvious reasons… misaligned incentives, human fallibility, complexity, and time. BUT, I’ve learned a lot (i think*****), and it has made me a better scientist (i think******). If nothing else, I can now make these cool graphs (below) for conference talks (and next time I won’t have to spend way too much time trying to make them look pretty*******).

Psych friends, come say hi at SPSP. Here’s the time and location for my talk and poster (and related scripts and files, here and here). Or, let’s just get a drink.

*me, Traci Mann, and Tim (our coffee connoisseur collaborator).

**that’s the hard part… sugar is yummy.

*** public project pages for the “coffee study” and “social message framing study” (the one climate change).

****an uneditable public archive of the study plan that is time-stamped prior to collecting (or looking at) data.

*****i welcome feedback and comments (particular on my R code). let me know if you find errors or have suggestions for improvement.

******hard to test empirically. though I’m pretty darn sure reproducibility and openness make Science better.

*******the beauty of reproducible code.

Sneak peek at SPSP presentation figures.

coffee

^Here’s the code (viewable in any web browser).

image-1-16-17-at-3-03-pm

^Here’s the code.

p.s. HT to Simine Vazire whose blog inspired the above footnote style. #usefulbloghack.

FDA’s Proposed Rules on Food Labeling

The Food and Drug Administration (FDA) of the US Department of Health and Human Services (HHS) has extended the commenting period to August 1st, 2014 for the proposed rules on food labeling (Docket ID: FDA-2012-N-1210).

I’ve written previously on the proposed rules. Here is a quick summary:

I applauded 3 major components of the ruling:

  1. Label added sugars in addition to natural sugars
  2. Addition of column on label to include both per serving and per package
  3. Highlight “calories”, “serving size” and “percent of daily value” through changes to the size, style, and position of font

And I questioned 1 component of the ruling:

  1. Revision of serving sizes of foods that can reasonably be consumed at one-eating occasion and updating, modifying, and establishing certain reference amounts customarily consumed. For example, this component would require that both a 12 oz bottle of soda and a 20 oz bottle of soda be labelled as a single serving.

I argued that this is a mistake for the following reasons:

  1. Research shows that people consume more when the container is larger. If we use the quantity of food or drink that is “reasonably consumed at one-eating occasion” or “customarily consumed” as a measure of what is safe, as is the implicit role of the FDA, then we fail to consider how a larger container may increase what is ordinarily consumed to levels that pose a health risk.

I suggested that new rule should reflect what is safe rather than what is “customarily consumed”. And, if the FDA insists that we use the quantity an average person consumes as a measure of what is safe, then we should at the least account for external factors (such as container size) that increase the point at which our body tells us we’re satisfied.

I pointed to 2 potential unintended consequences of this rule:

  1. Companies might discontinue smaller container sizes
  2. Consumers might choose larger containers over smaller containers with increased frequency

I’m revisiting this issue today after scanning the comments submitted thus far.

I found one submission from Behavioral Science and Regulation Group at Harvard that resonates with my concern. We both note that “serving size” is an implicit endorsement to consumers of what is an appropriate or healthy quantity. While I suggest the FDA use a different measure of “serving size” that is more in line with a healthful serving, the Harvard group suggests to change the wording of “serving size” to reduce implicit endorsement of healthfulness:

As the FDA acknowledges in its proposed rule, more than half of consumers perceive the term “serving size” to be a recommended serving size, not an amount customarily consumed. For those people that would, in the absence of a serving size, have eaten a small portion, the inclusion of a perceived serving size recommendation could lead them to eat more than they otherwise would. This is because these consumers believe that the FDA has implicitly endorsed the serving size as healthy. Consuming larger portion sizes is related to increased calorie consumption. While the rule’s revision of the serving size volumes and increased use of “whole package” labeling is appropriate and important, it will also exacerbate this problem, because the perceived recommendations will typically be for even larger portions […]

We suggest that the word “serving” and the phrase “serving size” be changed to avoid an implied endorsement. Changing “serving” to a word that does not suggest the context of a meal, like “unit” or “quantity,” may mitigate the endorsement effect.

This group from Harvard also endorses the FDA’s changes that use visual cues to increase clarity for the consumer:

The result is a nutrition label that behavioral science indicates will decrease the time consumers spend finding information, improve readability, focus attention on the most important information, and make information easier to process and remember.

 

And they make a fabulous recommendation on how to help consumers “avoid too much” of ingredients that are known to pose a health risk:

The FDA can better communicate product healthfulness by grouping nutrients into mutually-exclusive evaluative categories and using color to highlight healthful ingredients or particularly high or low nutrient levels.

The comment is worth reading in its entirety (see comment  from Behavioral_Science and Regulation Group here).

It was not surprising to find that the comment period was probably extended in response to the numerous requests from industries that anticipate adverse affects from the ruling: Juice Product Association, Specialty Food Association, American Beverage Association, Council for Responsible Nutrition, Snack Food Association, and Grocery Manufacturers Association to name a few. Though there was one notable exception in the Academy of Nutrition and Dietetics. Many of these industry representatives jumped in to voice their concern. It seems the cranberry industry is particularly concerned about requirements to label added sugars. There were several representatives from various companies including Gary Dempze of Gaynor Cranberry Co., Inc. who says, “unlike other fruit, cranberries have little natural sugar and, therefore, have a uniquely tart taste. Cranberry products need to be sweetened so consumers can enjoy their health benefits.”

Other supportive comments come from Weight Watchers, the American Diabetes Association, the National Alliance for Hispanic Health, and the American Dental Association to name a few of the big ones.

All-in-all it is fun to read through comments and see where different institutions fall on the issue. Give it a whirl.

 

Temporal Self-Regulation Theory: Why we keep trying (and failing) to go for that early morning run.

keep-the-dream-alive Last night in a burst of optimism I set my alarm for 5:30 AM. I thought I would sneak in an early morning run around the neighborhood before work. But as bells rang at that un-godly hour, I cracked an eye to a dark, cold room and groped for the snooze button. Ten minutes later, with a slight increase in clarity, I delayed once more “today, sleep is more important”…snooze again. As you might have guessed, I didn’t wake up in time to run.

We’ve all had a similar experience. Our preconceived intentions to engage in healthy behavior too often fail to come to fruition when it’s time to act. But we also intuit that our intentions are somehow linked to our behavior.

Most of the prevailing theoretical models of health behavior such as the Theory of Planned Behavior (Ajzen & Madden, 1986, request pdf), posit that intentions in combination with a number of other factors, such as behavioral beliefs, can predict likelihood of behavior. And these theories do predict behavior reasonably well (see Godin and Kok, 1996), but they fail to explain why large increases in intention only lead to small changes in behavior (see review by Webb and Sheeran, 2006). In this way these theories fail to fully explain health behavior.

Hall and Fong (2007), developed Temporal Self-Regulation Theory to help explain why, when it comes to health-related actions, the intention–behavior link may break down. They postulate that perhaps our intentions sometimes fail to lead to behavior because,

[many health behaviors are] associated with a characteristic set of contingencies whose valence changes dramatically depending on the temporal frame.

I’ve added emphasis to the quote to help break it down. In generally when psychologists talk about behavioral “contingencies” they are referring to if-then conditions that create potential for the occurrence of certain behavior and its consequences. Using the running example above, one behavioral contingency could be stated, “if I run in the morning, then I might be healthier when I’m older”. The “valence” of this contingency is positive—who doesn’t want to live a long and healthful life? “Temporal frame” refers to the very human capacity to think not only in the present moment or short-term, but also to weigh long-term consequences of our actions. Our example contingency has a long-term orientation. The authors contend that valence of the contingency changes with temporal frame, so let’s say I am thinking in the short-term, the behavioral contingency could then be stated, “if I run in the morning, then I might be tired for the rest of the day”. This is, of course, negative in valence. So the theory predicts that I will be more likely to create an intention to run in the morning if I’m focussed on the long-term as opposed to the short-term. This helps explain why it’s so hard to engage in health protective behaviors (such as running) and dis-engage in health risk behaviors (such as smoking). It is hard to delay gratification and most health risk behaviors are satisfying in the short-term and unsatisfying in the long-term, while most protective behaviors are predominately unsatisfying in the short-term and satisfying in the longer-term .

So, back to why my intention to run in the morning failed to lead to running after the alarm went off.

Last night when I set my alarm for 5:30 AM I was thinking about my long-term health, “I’ll look and feel so good in my summer swimsuit after working out” or “I’ll be less prone to disease when I’m older”.  Further the immediate costs of setting the alarm were low—I only had to click few buttons. In contrast, while reaching for the snooze, the costs of running were more immediate and the short-term consequence were salient, “I’m tired now, and I’ll be too sleepy to be productive today if I run”.

These tables and figures from Hall and Fong (2007) demonstrates how protective and risky health behavior have the opposite contingency valence with respect to time orientation. As depicted in the table 1, participants in this study estimated the point in time at which they would notice the benefit/cost of health protective behaviors (e.g. exercise and dieting), and health risk behaviors (e.g. smoking and drinking).

temporal proximity measure Hall and Fong (2007) Sticking with our morning run example, Figure 1 below demonstrates that people don’t notice the cost of running when thinking about rising at the crack of dawn for a run (question #1) or when deciding to run by setting the alarm an hour early (question #2). We start to feel the cost when the alarm goes off and we have to get out of bed and dress (questions #3). The perceived cost continues to grow as we run and after we’ve successfully run once (questions #4 and #5). We start to feel the cost less once we’ve made this morning run a regular routine for a week (question #6). As we continue to engage in our morning run routine the perceived cost continues to decrease, completely disappearing after a several years (question #9).

Now, what about the benefit of running early in the morning? Figure 1 indicates that we don’t feel the benefit of our run until we’ve done it regularly for a week (question #6), at which point the benefits grow exponentially for a year (question #8) and then decreases toward zero as we approach a decade (question #10).

These results provide evidence that the perceived benefit of running occurs well after the initial behavior occurs, while the perceived cost is felt just before, during  and a short while after the behavior initiates.

So when we are making the decision to set the alarm early for tomorrow’s run costs are low and abstract, so we are focusing on the long-term. When the alarm goes off and we are engaging in the behavior the costs are high and concrete, so we are focusing on the short-term. 

Before looking at Figure 1 below, notice that numbers 0 through 9 on the x-axis correspond to questions 1 through 10 in Table 1 pictured above. This is because academics like to make things more complicated than they need to be :). Screen Shot 2014-03-20 at 9.00.55 AM Figure 2 shows that the same trend holds for another health protective behavior (dieting). Screen Shot 2014-03-20 at 9.01.08 AMAs expected, the authors found the opposition result for health risk behavior—costs come after engaging in behavior and benefits occur before/during, see Figures 3 and 4 below. Screen Shot 2014-03-20 at 9.01.19 AM Screen Shot 2014-03-20 at 9.01.31 AM

So how does Temporal Self-Regulation Theory help me running in the morning? It suggests that on thing that might help is to try to minimize the short-term costs and maximize the short-term benefit. This can be hard, but may be as simple as rewarding yourself with a favorite breakfast if you complete the morning run.

Obviously, perceived temporal proximity with regard to behavior is only part of the picture. The authors introduce a working model (below) to illustrate Temporal Self-Regulation Theory more fully, which I’ve enhanced with definitions of each component. The model introduces two factors, behavioral prepotency and self-regulatory capacity that (1) influence (or moderate) the link between intentions and behavior; and (2) directly influence behavior in the absence of intentions. Health behaviors are complex and theories require continuous testing and refinement but Temporal Self-Regulation Theory adds an interesting new component to existing theories that is surely worth further consideration and testing.

Enhanced schematic representation of Temporal Self-Regulation Theory

References

Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes, intentions, and perceived behavioral control. Journal of Experimental Social Psychology, 22, 453-474.

Godin, G., & Kok, G. (1996). The theory of planned behavior: A review of its applications to health-related behaviors. American Journal of Health Promotion, 11, 87-98.

Hall, P. a., & Fong, G. T. (2007). Temporal self-regulation theory: A model for individual health behavior. Health Psychology Review, 1(1), 6–52. doi:10.1080/17437190701492437

Webb, T. L., & Sheeran, P. (2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin, 132, 249-268. doi: 10.1037/0033-2909.132.2.249

 

Links That Tickled Me

  1. Nate Silver’s revamped FiveThirtyEight blog made it’s debut this week. His post, What the Fox Knows, explaining the impetus behind their mission and the value of “data journalism” is worth a read.
  2. @econtalker had a heated, but interesting conversation with Jeffery Sachs on the Millennium Villages project. Jeff responds to episode with Nina Munk on her book critical of the project’s impact, The Idealist: Jeffery Sachs and the Quest to End Poverty. Healthy skepticism of foreign aid is a good thing, but my take from these episodes is that the Millennium Villages project has measured, positive impact. This said it is clear that further evaluation from impartial party is required.
  3. I’m drudging through a couple thick and juicy papers on self-regulation:
  • Hall, P. a., & Fong, G. T. (2007). Temporal self-regulation theory: A model for individual health behavior. Health Psychology Review, 1(1), 6–52. doi:10.1080/17437190701492437
  • Mann, T., de Ridder, D., & Fujita, K. (2013). Self-regulation of health behavior: social psychological approaches to goal setting and goal striving. Health Psychology : Official Journal of the Division of Health Psychology, American Psychological Association, 32(5), 487–98. doi:10.1037/a0028533