Using a descriptive social norm to increase vegetable selection in workplace restaurant settings.

Design

A pretest–posttest quasi-experimental design was used with three consecutive phases, each lasting 2 weeks: Preintervention Phase, Intervention Phase, and Postintervention Phase. During each phase cash register purchases made by participants dining at each of the three restaurants were recorded. During the intervention phase only, posters containing the social norm message were displayed in all restaurants. General posters on healthy eating that were normally displayed in the restaurants were present in all sites throughout the study.

Sample Size

Previous work using a social norm intervention in a pretest–posttest observational design yielded small effect sizes with phi values of 0.1 ( Burger & Shelton, 2011 ). A power analysis was conducted (G-Power 3.1) revealing that at least 785 observations were required to detect a small effect (assuming an alpha of 0.05 and power of 80).

Social Norm Message

Posters were used to communicate the same descriptive social norm message in each restaurant. On average, five posters (210 mm × 297 mm) were placed in each restaurant, near to the entrances and on top of the food counters at the point of selection. In addition, approximately 10 smaller (148 mm × 210 mm) posters were placed at each site in table-top holders, such that approximately half of the tables in each site featured one. The posters were printed in color, featuring a brown wood effect background with text superimposed, varying in color (gray, beige, orange, blue, and white) and font type and size. Small leaf/floral motifs (white) were incorporated into the top and bottom of the poster design (above and below the text). The message stated, “Most people here choose to eat vegetables with their lunch.”

The message was based on data acquired over a 2-week period prior to the start of the study at each site, using the same approach used for the study phases. The majority of participants (62%) purchased meals containing vegetables across all three sites. The poster message and design of the poster were selected on the basis of the responses of a focus group with 12 participants.

Restaurants and Meals

Three restaurants were recruited to this study. Two were based in the South of England and one in South Wales, U.K. Data collection was carried out between February and August, 2015; none of the 6-week data collection periods included any public holidays. All three sites were workplace restaurants, with two serving meals seven days a week (from early morning to late night), and the other restaurant serving meals Monday to Friday only (from morning to midafternoon). All three catering sites were run by the same external catering company. Each site was a self-service restaurant serving hot and cold food and drink. Diners queued to select food at a food counter and then purchased their selection at the cash register before sitting to eat. All sites offered a variety of main meals that included meals served with and without vegetables. These included hot meals (e.g., fish and fries) and cold meals (e.g., salads). Most meals cost between £3 and £4 (Great Britain Pound—GBP). All sites also offered side portions of vegetables, and it was possible for people to add vegetables to a meal that might otherwise not include them. They could also purchase other food items (e.g., cakes and chips) and drinks (e.g., water, soft drinks, and hot drinks). Communal tables were provided for dining. The tables seated groups of around 2–6 people.

Observation of participant characteristics

For one day during each phase at each restaurant a pair of researchers observed participants purchasing meals in the restaurants to estimate basic demographics (gender, age, and weight status). Researchers were blinded to each other’s ratings. Observations were conducted from 11:00 a.m. to 2:00 p.m. Using previously established criteria ( Eves, Webb, & Mutrie, 2006 ; Kerr, Eves, & Carroll, 2001 ), participant gender was categorized on visual appraisal, weight status was categorized using body silhouettes (nonoverweight vs. those who were overweight or greater), and age (under 60 vs. over 60) was categorized based upon presence of gray hair and general appearance.

Meal data

The number of meals purchased with and without vegetables during each 2-week period were recorded at each site and combined across all sites. Pearson’s chi-squared test was used to compare (a) preintervention to intervention, (b) intervention to postintervention, and (c) preintervention to postintervention (to examine meal selections across the study). Odds ratio ( OR ) and confidence intervals (CI) were also estimated.

Cold drinks data

Purchases of water (as a percentage of the total number of cold drink purchases) were also examined as a comparator. This was to check whether purchases that were not expected to change across phases (because they were not the target of the social norm messages) changed, possibly because of a general change in purchasing patterns unrelated to the intervention. The data were extracted using the same approach used for the meal data.

Meal Purchases

A total of 9445 meal purchases were recorded (further details are outlined in Table 1 ). The overall number of meals purchased remained stable over time, but the intervention and postintervention phases were associated with increased purchase of meals with vegetables.

Pearson’s chi-squared test revealed that the introduction of the posters was associated with an increase in purchase of meals with vegetables from 60% during preintervention to 64% during the intervention phase; χ 2 (1, N = 6,357) = 11.32, p < .01, Φ = 0.04 ( OR 1.2, 95% CI [1.1–1.3]). From the intervention phase to the postintervention phase there was a further increase in purchase of meals with vegetables from 64% to 67%; χ 2 (1, N = 6,267) = 7.27, p < .01, Φ = 0.03 ( OR 1.2, 95% CI [1.0–1.3]). Overall, preintervention to postintervention was associated with an increase in the purchase of meals with vegetables from 60% to 67%; χ 2 (1, N = 6,266) = 36.35, p < .001, Φ = 0.08 ( OR 1.4, 95% CI [1.2–1.5]) (See Figure 1 ).

Percentage of meals purchased with vegetables, split by test phase. The introduction of the poster was associated with a significant increase of meals purchased with vegetables from the preintervention to intervention phase. Removal of the poster was associated with a further increase from the intervention to postintervention phase. Overall, baseline to postintervention was associated with an increase in the purchase of meals containing vegetables. ** p < .01. *** p < .001.