Everyday Research Methods

Can you eat dessert at breakfast to lose weight?

The causal headlines are bearing some great news--

"How Dessert for Breakfast Can Help You Lose Weight"

or "Dessert at breakfast may help dieters."

The press is buzzing about a recent study from Tel Aviv University. Here's the abstract from the journal article. The researchers randomly assigned people to one of two low-calorie diets. Women ate 1400 calories on both diets; men ate 1600 calories. The one difference was that on the first diet, breakfast was a low-carb 300 calorie meal. On the other diet, breakfast was high carbohydrate, high protein meal that included dessert such as cookies or cake.

Unlike those in the low-carb breakfast group, the dieters who ate dessert at breakast were more likely to keep their weight off during the follow up period. In fact, they continued to lose an average of 13 extra pounds, while the other group gained all their weight back.

a.) What do you think--does the study described in the abstract support the causal statement that "eating dessert at breakfast can help you lose weight"? Consider the research design and go through the three causal rules in this example.

b.) What causal claim could you make about this study?

Suggested answers:

a) The study is an experiment, so it can potentially suport a causal claim. The results did show covariance: dieters who ate dessert at breakfast did lose more weight than dieters who did not. Because the dessert aspect of the diet was manipulated by the researchers, we know that the dessert came first and the weight loss came later (therefore, there is temporal precedence). What about internal validity? There's a potential problem here, because the dieters who ate dessert for breakfast also had a high-protein breakfast. The comparison group did not eat dessert, but they also ate a low-carb breakfast, too. So we don't know if the weight loss was caused by the dessert, by the high protein, or by some combination of these two.

b) Because the study appeared to have manipulated two variables, journalists would be more accurate if they said that "eating a high-protein, high carb breakfast that includes dessert can help dieters lose weight."

What makes someone an angry drunk?

What makes someone an angry drunk?

This MSNBC article discusses the difference in life focus and how that affects one’s level of retaliation while intoxicated.

The average age of the study's 495 volunteers was 23, all of whom described themselves as social drinkers and none of whom had any past or present drug, alcohol, or psychiatric-related problems. They each took a questionnaire designed to measure which of the participants were future-focused and which were more impulsive.

a.) Why do you think the researchers made sure the participants did not have any previous drug, alcohol, or psychological issues?

b.) There were 495 volunteers. Do you think they used a biased or unbiased sampling method?

The researchers gave half the volunteers alcohol and the other half no alcohol. Each participant then played a speed reaction game with a confederate where the winner shocked the opponent. “As the game wore on, the shocks got longer and more intense, making it seem like the opponent was getting meaner and meaner with every win.” Those who were drunk retaliated more than those who were not drunk. And finally, those who were drunk and impulsive had the highest likelihood of retaliation.

The article quoted the study’s author, Brad Bushman, who summarized the results:

"The less people thought about the future, the more likely they were to retaliate, but especially when they were drunk. People who were present-focused and drunk shocked their opponents longer and harder than anyone else in the study,” Bushman explained. "Alcohol didn’t have much effect on the aggressiveness of people who were future-focused."

c.) What are the independent variables and the dependent variable? What is the design of this study?

d.) Sketch a graph of the results, according to Bushman’s quoted description above. Could you say if there are main effects or interactions?

 

Suggested answers

a.) For one thing, it’s probably not a good idea to get people drunk when they have a history of substance abuse. But also, such participant variables could possibly create unsystematic variability. This unsystematic variability would get in the way of the researchers' ability to detect a true difference between groups.

b.) It’s difficult to tell from the journalist’s description alone, but it seems like the participants self-selected for the study, a method that leads to external validity issues. This is a biased sampling method. However, the study was focused more on internal validity (comparing the drunk and sober conditions) than on external validity.

c.) IV1: Life focus (future focused vs. present focused (that is, impulsive)). This is a participant variable.

IV2: Drinking alcohol (drunk vs. sober).  This is a manipulated variable.

DV: Retaliation with shocks

There are two independent variables, so it is a 2x2 factorial design.

d.) Your graph would have “degree of retaliation” on the y-axis. One of the IV’s would be on the x-axis, and the other would be the color of the lines or bars.

It seems that there is a main effect for drinking alcohol such that being drunk leads to greater retaliation than when sober. There is a main effect for life focus such that those who are present focused (impulsive) have a greater rate of retaliation than those who are future focused. And it seems that there is an interaction. For future-focused people alcohol does not make a difference, but for those who are present focused (impulsive) being drunk causes greater retaliation.

 

Guest post by Kate Banford, University of Delaware

Three different takes on a single journal article

In the journal-to-journalism cycle, the details of a study can get distorted on their way to the web. Here are three different takes on a recent article published by Canadian researchers. The study had apparently investigated links between spanking and child outcomes.

Notice that the headlines in each study make different claims and focus on different variables.

Spanking kids can cause long-term harm: Canada study

Children Who Get Physical Punishment Tend Toward Aggression: Survey

Spanking makes kids more aggressive, should be illegal: Report

b) Identify the variables in each headline. Why do you think the different sources emphasized different variables?

c) Do you think the Canadian study at the source of these headlines was able to support a causal claim? Why or why not?

d) Is it possible to conduct an experimental study on spanking ? Why or why not ?

e) You might notice that the first headline describes the project as a "Study." The second headline describes the project as a "survey". The third headline describes the project as a "report."

Use your psycINFO skills to find the original publication and identify it as an empirical journal article or a review journal article.

 

Thanks to Dr. Carrie Smith of the University of Mississippi for sending these headlines!