Can serving-size labels reduce the portion-size effect? A pilot study

Abstract: Research has shown that the bigger the portion that people are served, the more food they eat; this phenomenon is referred to as the portion-size effect. Providing objective serving-size information on food products has been shown to reduce the influence of external food cues on people's eating behavior. The current study examined whether providing objective serving-size information would also reduce the portion-size effect. 100 female participants were served either a small or large portion of pizza in the context of a taste test. The large portion was either unlabeled, labeled as “Contains 2 servings,” or labeled as “Contains 4 servings.” Food intake was lower when the large portion was labeled “Contains 4 servings” compared to when it was labeled “Contains 2 servings.” Moreover, participants' intake in the large portion/4 servings condition was statistically similar to the intake of participants in the small portion condition. Thus, the standard portion-size effect was observed when the large portion was unlabeled or was labeled as “Contains 2 servings,” but not when the large portion was labeled as “Contains 4 servings”. These findings suggest that providing serving-size information can reduce the portion-size effect, but that the specific content (and not just the presence) of serving-size information is important in determining food intake.


Alex’s Notes: Portion sizes have increased dramatically over the past decades, and unfortunately, more food available to eat increases how much we eat. A recent meta-analysis found that doubling the size of a portion results in a 35% increase in consumption. And being mindful doesn’t seem to help. So why not instead focus on the food environment? Serving sizes of foods are often ambiguous and leave consumers to draw their own conclusions about how much to eat.Providing objective serving-size information, then, should eliminate the need for people infer the appropriate amount to eat and should reduce the reliance on external cues (i.e. portion size). This is supported by the work of Brian Wansink, showing that participants who were given a bag of granola that was labeled “contains 2 servings” ate approximately 30% less than did participants who were given a bag of granola labeled “contains 1 serving.” If you haven’t heard of Wansink before, look him up, his work is great.

To build on the above, the current study recruited 100 female undergraduate students and assigned them to either a small portion condition or large portion condition. Those allocated to the small portion condition were served a plate containing 200 g of cheese pizza presented as 12 bite-sized pieces, and those allocated to the large portion condition were served a plate containing 400 g of cheese pizza presented as 24 bite-sized pieces. The same plate size was used in all conditions. The girls’ average age was 21 years and average BMI was 21.5 kg/m2.

Before the girls were allowed to chow down, however, some deception was in order. The main outcome of the study was to see how serving size information influences the amount of food consumed. This couldn’t be let known to the participants else they screw up the whole thing by monitoring how much they eat. Therefore, the researchers made up study conditions to fool the girls.

To start, the girls were told that they were going to taste and rate the pizza on a variety of dimensions (e.g., how salty, how crunchy). Experimental sessions took place between 11 am and 3 pm and the girls were asked not to eat three hours prior but told they to eat as much of the pizza they wanted in order to make accurate taste ratings. Second, they were provided a pizza package before being brought the actual pizza and asked to evaluate it on various dimensions such as color, font style, and esthetic appeal. The researchers told the girls this was the container of the pizza they would be eating later. Of course they didn’t really care about responses to the pizza container; rather, the researchers used it as a way to subliminally implant the serving size information for the study. That is, for the small-portion condition and one of the large-portion conditions, the packaging included no information about the number of servings. For the other two large-portion conditions, the packaging either included the statement “Contains 2 servings” or included the statement “Contains 4 servings.” I tried to lay out the conditions in the table below.


Small (S)

Large (L)

Large 2 (L2)

Large 4 (L4)

Amount of Pizza





# of Servings



2 servings

4 servings

The girls were also rated on their levels of hunger before beginning, and initial hunger was significantly correlated with total food intake. However, age and BMI were not, although BMI did trend (p=0.1) towards a positive correlation with total food intake (bigger girls have bigger appetites).

The Eating Environment Wins Again

As would be expected, the girls in the S condition ate significantly less than the L condition (by 17%) and the L2 condition (by 30%). However, there was no significant difference between the S and L4 conditions. Moreover, the L4 condition ate significantly less than the L2 condition (by 20%) and ate slightly but not significantly less pizza than the L condition.

So in a nutshell, the fact that the participants ate more in the L than the S condition supports the idea that food size influences how much we eat. The fact that the participants ate less in the L4 condition than the L2 condition supports the idea that the nature of the information provided influences how much we consume. But not everyone reads the labels, and the potential for food labels to influences people’s eating behavior relies on them noticing and understanding them.

We could argue day and night about how to implement effective labels, but regardless of the choice, one thing that is overlooked (in my opinion) is the huge disconnect between consumers’ actual food beliefs and the serving sizes. For instance, a 20oz bottle of soda is not “2.5 servings.” No one drinks the damn thing in 8 ounce portions. Or what about cereal being labeled with servings of ¾ cup and pints of ice cream with 4 servings? None of these are realistic. People drink the entire bottle, eat bowls of cereal (at least 1.5-2 cups), and devour the whole pint. By using the unrealistic and small serving sizes, the nutrition information doesn’t appear as “bad” and the consumer will eat the whole thing under the assumption that this information applies to entire package rather than to the more realistic feeding regimes of the people these foods cater too. Would you really eat a pint of Ben & Jerrry’s New York Super Fudge Chunk® ice cream if the label listed the thing as having 1200 kcal, 80g of fat, and 100g of sugar (yes these are real numbers off the website)? Trick Question! Of course you would, that stuff is awesome. But seriously, I hope you get my point.


Network Affiliates

Quick Links I

Our Location

2908 Brownsboro Rd
Suite 103
Louisville, KY 40206
(502) 690-2200

SHR Newsletter

Subscribe to our FREE newsletter
to receive the latest updates in your inbox!
SHR Newsletter
Internet Radio