Does knowing what vending machines options are healthier impact sales?: Experimental design practice

[sue_frantz]

Will visitors to vending machines make healthier purchases if they know which purchases are healthier?

After discussing experimental design, ask your students to work in small groups. First, they are to take the above question and reword it as a hypothesis. As part of their design, they should identify the independent variable and its levels and identify the dependent variable—providing operational definitions of the levels of the independent variable and the dependent variable. After discussion has waned, ask a volunteer from each group to share their group’s design.

Conclude this activity by sharing the design and results from this freely available experiment conducted in Philadelphia (Gibson et al., 2024).

Researchers tested four different messaging systems in 267 vending machines; 150 machines sold beverages and 117 sold snacks. The study began with a 9-month period where researchers just monitored sales at these machines. The intervention period began immediately after this baseline period and lasted 13 months.

All of the materials used in this study are available with the article. Click on “supplemental content,” and select “Supplement 2.”   

Message 1: Beverage-tax poster only. Next to the beverage machines, the poster read, “Philly bev tax is here. Starting January 1, 2017, a 1.5-cents-per-ounce tax will be applied to sweetened beverages.” Next to the snack machines, the poster read, “Are you drinking fewer sweetened drinks because of the cost? Keep up the healthy choices by also choosing healthy snacks.”

For the remaining three conditions, a poster placed next to the vending machine explained how to interpret the labels.

Message 2: Green labels. These labels were placed on items that were deemed a healthier choice. For example, seltzer water and popcorn got green labels. In the corresponding poster’s small print were the criteria the researcher’s used for the green label. This was their operational definition of “healthy.” Healthy beverages had 5g or less of sugar per 12 ounces. Healthy snacks

Message 3: Traffic lights. Green, yellow, or red labels were placed on each vending machine option. Again, green labels were on healthier items like seltzer water and popcorn; the poster explained these were “choose often” options. Yellow labels were on so-so items like diet drinks and trail mix; the poster explained these were “choose sometimes” options. Lastly, red labels were on unhealthy items like sweetened teas and chocolate bars; the poster explained that these were “choose rarely” options. In the poster’s small print were the criteria used for determining what was green, yellow, and red.

Message 4: Physical activity. Each item had a label noting how many minutes of brisk walking it would take to work off the calories in the beverage or snack. The accompanying poster noted that the healthier items would take less than 45 minutes to walk off.

The primary dependent variable was product sales; data were provided by the vending machine company. Researchers compared sales data from the green label, traffic light, and physical activity conditions vending machines with the beverage-tax poster only vending machines. Here are some of the key results.

• “Traffic light machines were 25% more likely to sell a yellow (eg, taxed diet soda) compared with a red (eg, taxed soda) beverage compared with beverage tax machines.” While purchasers didn’t go green, some were willing to go from red to yellow. If you go to the machine to get a Coke, you might be willing to go Coke Zero, but declining the Coke for a seltzer water is asking a bit much.  
• “For snacks, green-only machines were 10% more likely to sell a green snack (eg, baked chips) compared with a red snack (eg, candy) compared with beverage tax machines.” If you want Doritos, you might be talked into popcorn.
• “At the machine level, physical activity compared with beverage tax decreased the expected monthly number of beverages sold by 24%.” If you go to the machine to get a Pepsi, and you see that it’s going to take 62 minutes of brisk walking to walk off the calories, you might choose to fill up your water bottle instead.
• “Traffic light labels significantly decreased total calories sold per customer trip compared with physical activity.”

The researchers did a manipulation check. They asked the visitors to the vending machines if they noticed “any posters or labels displayed on the machines.” In the beverage-tax poster group, 26% reported seeing the poster. In the green-only group, 40% reported seeing the green labels. In the traffic light group, 65% noticed the green, yellow, red labels. In the physical activity group, 58% noticed the brisk-walking-minutes labels. Interestingly, “Reported noticing of messaging did not moderate condition effects on calories sold per trip.”

Are any of your students interested in replicating this study at your institution? What would they need to do in order to conduct this experiment? Would they make any changes to the original study? Why or why not?

Reference

Gibson, L. A., Stephens-Shields, A. J., Hua, S. V., Orr, J. A., Lawman, H. G., Bleich, S. N., Volpp, K. G., Bleakley, A., Thorndike, A. N., & Roberto, C. A. (2024). Comparison of sales from vending machines with 4 different food and beverage messages: A randomized trial. JAMA Network Open, 7(5), e249438. https://doi.org/10.1001/jamanetworkopen.2024.9438