You are here
Home > Fattening Food > Boosting healthy food choices by meal colour variety: results from two experiments and a just-in-time Ecological Momentary Intervention

Boosting healthy food choices by meal colour variety: results from two experiments and a just-in-time Ecological Momentary Intervention


Study 1

Manipulation check

Within-subjects ANOVAs were conducted to compare the four meals regarding healthiness, energy content, and colourfulness. The participants more strongly agreed that they had chosen healthy foods when putting together the healthy meal (F(3, 240) = 46.09, p < .001, ηp2 = .37), low calorie foods when putting together the low calorie meal (F(3, 243) = 145.13, p < .001, ηp2 = .64), and more colourful foods when putting together the colourful meal(F(3, 243) = 72.77, p < .001, ηp2 = .47). Means and standard deviations are listed in Table 1. The manipulations were therefore successful.

Table 1 Means and standard deviations for manipulation check items for studies 1 and 2

Differences in food consumption

Results are summarised in Table 2. In a first step, total meal weight was compared between conditions, yielding significant differences. Meals in the low calorie condition weighed significantly less than the other meals (ps < .001).Footnote 2

Table 2 Total weight and proportions of food groups for the choice conditions in studies 1 and 2

In a second step, meals were compared regarding the proportions of food groups. Significant differences between conditions emerged for all food groups except protein sources. Specifically, colourful meals contained more vegetables than typical meals and more fruit than all other meals (ps ≤ .002). However, they contained less vegetables than low calorie meals (p ≤ .001), and more fats and sweets than healthy and low calorie meals (ps ≤ .005). Moreover, they contained more grains and starches and dairy than low calorie meals (ps < .001), more dairy than healthy meals (p = .021), and less drinks than all other meals (ps < .001). Paired comparisons for all food groups are listed in Additional file 2.

Evaluation of the choice strategies

Meals differed in the participants’ expectations of satiation (F(3, 246) = 53.62, p < .001, ηp2 = .40). Low calorie meals were expected to be less filling than the other meals (p < .001). Eating low calorie meals was also perceived to be more difficult (F(2, 164) = 66.14, p < .001, ηp2 = .45), more complex (F(2, 164) = 29.73, p < .001, ηp2 = .27), and less fun (F(2, 164) = 89.20, p < .001, ηp2 = .52) than eating healthy or colourful meals (ps < .001). Means and standard deviations are listed in Table 3.

Table 3 Evaluation of the choice strategies in studies 1 and 2

In the ranking task, healthy meals were ranked first for feasibility by 54.2% of participants, while 37.8% participants ranked colourful meals highest and 8.5% ranked low calorie meals highest. Regarding anticipated taste, colourful meals were ranked highest by 63.4% of participants, while healthy meals were ranked highest by 37.3% and low calorie meals were never ranked first.

Study 2

Manipulation check

Within-subjects ANOVAs were conducted to compare the three meals regarding variety and colourfulness. The participants agreed more strongly to have chosen a variety of foods when putting together the varied meal (F(2, 82) = 19.70, p < .001, ηp2 = .33), and more colourful foods when compiling the colourful meal (F(2, 82) = 45.70, p < .001, ηp2 = .53). Means and standard deviations are listed in Table 1. The manipulations were again successful.

Differences in food consumption

The results are summarised in Table 2. In a first step, total meal weight was compared between conditions, yielding no significant differences.Footnote 3

In a second step, meals were compared regarding the proportions of food groups. Significant differences were found for fruit, grains and starches, fats, and drinks. Specifically, colourful meals contained a higher proportion of fruit (ps ≤ .001) and a lower proportion of fats (ps ≤ .025) than the other meals. Moreover, colourful meals contained a smaller proportion of grains and starches than typical meals (p = .012), and a smaller proportion of drinks than both typical and varied meals (ps ≤ .019). Paired comparisons for all food groups are listed in the Additional file 2.

Evaluation of the choice strategies

Meals differed in the participants’ expectations of satiation (F(1.75, 71.89) = 5.68, p = .007, ηp2 = .12; Greenhouse-Geisser corrected). Participants felt that the typical meal would be less filling than the varied meal (p = .001). Colourful and varied meals did not differ in feasibility, simplicity, or fun (ts(41) ≥ |1.20|, ps ≤ .238). Means and standard deviations are listed in Table 3.

In the ranking task, colourful and varied meals were ranked first for feasibility equally often (50% of participants), while varied meals were ranked first more often regarding anticipated taste and healthiness (taste: 66.7% of participants; healthiness: 71.4% of participants).

Study 3

Relationships between perceived meal colour variety and food intake

Separate multilevel models were computed for all food groups. A significant positive relationship with perceived meal colour variety emerged for vegetables. When comparing the random slopes (b = 0.003, t(72.79) = 7.73, p < .001, quasi-R2 = .11) and random intercept models (b = 0.003, t(1132.30) = 9.55, p < .001, quasi-R2 = .07), the random slopes model assuming differences in the individual slopes was preferred (χ2(df = 2) = 9.82, p = .007). The participants therefore differed in the relationship between meal colour variety and proportion of vegetables consumed (see Fig. 3a). Ninety-four percent of slopes were positive, indicating that increased perceived meal colour variety was associated with a higher proportion of vegetables consumed, while 6% of slopes were negative, indicating that for a minority of subjects increased perceived meal colour variety was associated with a lower proportion of vegetables consumed.

Fig. 3

Associations between perceived meal colour variety and proportion of food groups consumed in the meal in Study 3. Each thin grey line represents a regression line for one participant. The thick black line represents the overall regression line. a Proportion of vegetables. b Proportion of fruit. c Proportion of grains and starches. d Proportion of sugary extras

A significant negative relationship emerged between perceived meal colour variety and the proportion of fruit consumed. When comparing the random slopes (b = −.001, t(81.17) = − 2.81, p = .006, quasi-R2 = .10) and random intercept models (b = −.001, t(1130.00) = − 3.96, p < .001, quasi-R2 = .01), the random slopes model was preferred (χ2(df = 2) = 66.36, p < .001), indicating that the relationship between perceived meal colour variety and the proportion of fruit consumed differed between participants. Sixty-seven percent of slopes were negative, indicating that a greater perceived meal colour variety was associated with a lower proportion of fruit consumed, while 33% of slopes were positive, indicating that a greater perceived meal colour variety was associated with a higher proportion of fruit consumed (see Fig. 3b).

A significant negative relationship also emerged between perceived meal colour variety and the proportion of grains and starches consumed. When comparing the random slopes (b = −.002, t(65.54) = − 3.87, p < .001, quasi-R2 = .07) and random intercept models (b = −.002, t(1130.25) = − 5.18, p < .001, quasi-R2 = .02), the random slopes model was preferred (χ2(df = 2) = 15.03, p < .001), indicating differences between participants in the relationship between perceived meal colour variety and the proportion of fruit consumed. Seventy-six percent of slopes were negative, indicating that a greater perceived meal colour variety was associated with a lower proportion of grains and starches consumed, while 34% of slopes were positive, indicating that a greater perceived meal colour variety was associated with a higher proportion of grains and starches consumed (see Fig. 3c).

Lastly, a significant negative relationship emerged between perceived meal colour variety and the proportion of sugary extras consumed. When comparing the random slopes (b = −.001, t(74.75) = − 2.05, p = .044, quasi-R2 = .11) and random intercept models (b = −.000, t(1126.25) = − 2.93, p = .003,, quasi-R2 = .01), the random slopes model was preferred (χ2(df = 2) = 72.37, p < .001), indicating that the relationship between perceived meal colour variety and the proportion of sugary extras consumed differed between participants. Sixty-two percent of slopes were negative, indicating that a greater perceived meal colour variety was associated with a lower proportion of sugary extras consumed, while 38% of slopes were positive, indicating that a greater perceived meal colour variety was associated with a higher proportion of sugary extras consumed (see Fig. 3d).

For fried foods, a significant negative relationship emerged for perceived meal colour variety for the random slopes model (b = −.000, t(141.01) = − 1.98, p = .049, quasi-R2 = .01). However, the deviance test (χ2(df = 2) = 4.21, p = .122) comparing the random slopes to the random intercept model preferred the random intercept model (b = −.000, t(1123.69) = −-1.76, p = .079, quasi-R2 = .00), which did not reach significance.

No significant relationships with perceived meal colour variety were found for protein and dairy (see Table 4 for a summary of all models).

Table 4 Results of the multilevel models to analyse the relationship between perceived meal colour variety and the consumption of seven food groups

Impact of the prompt to eat a colourful lunch on food consumption

Differences between baseline and intervention weeks

A significant difference between the baseline and intervention weeks emerged for vegetables consumed. When comparing the random slopes (b = 0.04, t (548.83) = 2.16, p = .031, quasi-R2 = .02) and random intercept models (b = 0.04, t (768.21) = 2.20, p = .028, quasi-R2 = .02), the random intercept model assuming no differences in the individual slopes was preferred (χ2(df = 2) = 0.69, p = .709). Thus, the difference between baseline and intervention weeks regarding the proportion of vegetables consumed was comparable between participants. Results indicate that the participants consumed a greater proportion of vegetables during the intervention week compared to the baseline week.

A significant difference between baseline and intervention weeks emerged for dairy consumption. When comparing the random slopes (b = − 0.04, t(81.50) = − 3.16, p = .002, quasi-R2 = .02) and random intercept models (b = − 0.04, t (766.80) = − 3.17, p = .002, quasi-R2 = .02), the random intercept model assuming no differences in the individual slopes was preferred (χ2(df = 2) = 0.79, p = .675). Thus, the difference between baseline and intervention weeks regarding the proportion of dairy consumed was comparable between participants. Results indicate that the participants consumed a smaller proportion of dairy products during the intervention week compared to the baseline week.

For all other food groups, no significant differences emerged between the baseline and intervention weeks (see Table 5).

Table 5 Results of the multilevel models to compare differences in food consumption between baseline and intervention weeks

Differences between baseline and follow-up week

Between baseline and follow-up weeks, no significant differences were found (b |0.02|, ts(≥ 74.40)  |1.31|, ps ≥ .190), indicating that food consumption during the follow-up week returned to the baseline level when prompts were no longer sent.

Evaluation of the prompt

Participants indicated that they found eating colourfully is something that is rather easy for them to do (M = 57.96, SD = 24.87). They also indicated that eating colourfully is pleasant (M = 70.79, SD = 27.95), and self-rated compliance was satisfactory (M = 60.36, SD = 26.89).



Source link

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Top