Development and validation testing of a weight management nutrition knowledge questionnaire for Chinese adults (C-WMNKQ)

Demographic information

Sample B consisted of 590 adults, predominantly female (80.0%), with age Median (M₂₅, M₇₅) of 28(24, 37) years, BMI Median (M₂₅, M₇₅) of 21.2(19.3, 23.4), 91.3% of the participants had a Bachelor’s Degree or higher. The general demographic information of the participants is shown in Table 1.

Content validity

In this study, except for the original questionnaire Question 12 (Do you think raspberries contain fiber?), which had an I-CVI of 0.64, and Question 13 (Do you think Bananas contain fiber?) and Question 28 (Which of these is a good strategy to help to lose weight?), which had an I-CVI of 0.71, the I-CVIs of the remaining 38 items were all greater than 0.78, and the S-CVI/Ave = 0.92, which was greater than 0.90. Considering the overall findings, the content validity index of the questionnaire was still acceptable, and the three items were modified. In Question 12, the word “raspberry” was changed to “strawberry”. In Question 13, some experts suggested that bananas are low in dietary fiber, but after discussion, it was considered that it is meaningful to know that bananas have dietary fiber even though they are low in content, so they were retained. In Question 28, in response to the experts’ question of whether the number or type of foods examined is not clear, the questionnaire was modified. The experts’ general perception of the types and amounts of food differed from the questionnaire content, while Michael Jensen pointed out that more food types in terms of quantity were important contributors to excess energy intake [16]. Retaining this item contributes to the understanding of the concept that food variety influences food intake.

Face validity

Sample A eventually included 55 adults. Most of the questionnaire items were fully understood, and some items that were semantically ambiguous or unclear were modified. The stem of the original Question 25, “Which of these is too large of a serving size for a protein food?” was changed to “Which of the following foods is too large a portion for the amount of protein it contains?”, and “2 tael” was replaced with “85 g” in option A. In option B of the original Question 40 (Which of these has the fewest calories?), we replaced “soda” with “soda carbonated beverage”. The modifications enhanced the face validity of the questionnaire by increasing the percentage of correct responses from 34.6% in the pre-survey to 44.7% for the original Question 25 and from 16.6% in the pre-survey to 44.4% for the original Question 40.

Item analysis

Item difficulty

The item difficulty value for all items ranged from 0.11–0.97, and none of the questions were considered too difficult (<0.1) except for the original Question 13 (Do you think Bananas contain fiber?) and Question 16 (Do you think broccoli contain fiber?), which had >0.9 and were considered too easy. These two questions were deleted and the mean value of the whole questionnaire after deletion was M = 0.57, SD = 0.20, indicating that the questionnaire was moderately difficult and had good discrimination.

Item discrimination

(1) The discrimination index (D-value) of each entry was −0.13–0.76, and all items were positive except for the original Question 28 (Which of these is a good strategy to help to lose weight?). Among all the positive items, D-value of Question 1 (Which of the following single, standard size cups of yogurt for breakfast would be a better choice for weight loss?), Question 13, Question 16, and Question 17 (Do you think green peas contain fiber?) were <0.2. (2) All items were significantly correlated with the total scores, except for the original Question 1. The coefficients of the point-two column correlations of the original Question 6 (If you are working to lose weight and are on vacation where two different breakfast buffet styles are available. Buffet A offers a variety of fruits and whole grain bakeries. Buffet B offers a variety of fruits, pastries, dairy products, deli meats, and whole grain bakeries. Which choice will make it easier to keep your calorie intake lower?), Question 17, and Question 28 were less than 0.2, and only Question 28 was negatively correlated with the total score.

Inter-item correlation

No correlation between the scores of any two items was greater than 0.9, and no items needed to be deleted.

In summary, for all of the above items that did not meet the criteria for item analysis (the original Questions 1, 6, 13, 16, 17, and 28), Question 13 and 16 were deleted, and the remaining items were retained: (1) Question 1 examined energy density through the form of food labels. Food labels provide information about the nutrient content of a food, and the ability to understand this information somewhat influences dietary choices [6]. The Dietary Guidelines for Chinese Residents 2022 also recommends “learning to read food labels” [44]. (2) Michael Jensen was consulted about the possibility of combining the green peas in the original Question 17 with the lentils in Question 18, and he pointed out that it was necessary to understand the dietary fiber content of each of the two types of legumes separately, so Question 17 was retained, and it can also be contrasted with Question 18. (3) The original Questions 6 and 28 both belonged to the conceptual dimension of food variety influencing food intake, which was proven to be relevant to weight management and should not be deleted.

Construct validity

Sample C consisted of 168 adults (91 nutrition professionals and 77 finance professionals) whose demographics are shown in Table 2. There was no statistically significant difference in demographic variables except for sex and BMI. Both nutrition professionals and finance professionals were predominantly female, with a majority of normal BMI. The results showed that the overall percentage of correct scores was 72.5% (95% CI = 69.5%–75.6%) for nutrition professionals and 46.2% (95% CI = 43.1%–49.2%) for finance professionals, with an overall difference of 26.3% (p < 0.001). Except for the dimension “Variety of food affects food intake”, nutrition professionals scored significantly higher (p < 0.001) than finance professionals on the total questionnaire score and other dimensions, as shown in Table 3.

Criterion validity

The C-WMNKQ and C-GNKQ were measured simultaneously for the 590 respondents in Sample B. The results of the Pearson correlation analysis showed that the correlation coefficient between the total C-WMNKQ score and the total C-GNKQ score was 0.641(p < 0.001, 95% CI = 0.594–0.684), and that the correlation coefficients with the scores on the four dimensions of the C-GNKQ: dietary recommendations, source of nutrients, daily food choice, and diet-disease relationship, were 0.491(p < 0.001, 95% CI = 0.425–0.551), 0.609(p < 0.001, 95% CI = 0.558–0.654), 0.433(p < 0.001, 95% CI = 0.372–0.495), and 0.487(p < 0.001, 95% CI = 0.428–0.542), all of which showed significant positive correlations. The criterion validity of the questionnaire was good.

Reliability

Internal consistency reliability and split-half reliability

The overall internal consistency Cronbach’s alpha coefficient was 0.808, the ω coefficient was 0.81, and the coefficients for the overall and dimensions are shown in Table 4.The items above the overall reliability after deletion of the items were the original Question 1 (0.814), Question 6 (0.810), and Question 28 (0.816), and the reasons for the retention of these three items have been discussed before. The overall split-half reliability Guttman coefficient was 0.788.

Test–retest reliability

A total of 168 people (91 nutrition professionals and 77 finance professionals) completed the pre-test questionnaire, of whom 129 (65 nutrition professionals and 64 finance professionals) completed the post-test questionnaire 2 weeks later, a shedding rate of 23.2%. The Spearman correlation coefficient for the total score of the pre- and post-tests was 0.843, p < 0.001; the overall ICC value of the questionnaire was 0.908 (95% CI = 0.870–0.935), which met the criteria for retest reliability, as shown in Table 5.

Convergent validity

Gender

Except for Alcohol intake and sugar sweetened beverages, there was a significant difference between genders on the total score and the other dimensions. The independent samples t-test showed that on Variety of food affects food intake, females (M = 0.77, SD = 0.66) scored significantly lower than males (M = 0.97, SD = 0.71), t = −2.964, p = 0.003, 95% CI = 0.069–0.342, d = 0.291. On Total score (Female: M = 22.57, SD = 5.99; Male: M = 20.63, SD = 6.65. t = 2.892, p = 0.004, 95% CI = 0.616–3.265, d = 0.307), Energy density (Female: M = 10.57, SD = 2.69; Male: M = 9.75, SD = 3.02. t = 2.886, p = 0.004, 95% CI = 0.262–1.378, d = 0.287), Portion size (Female: M = 7.58, SD = 2.74; Male: M = 6.75, SD = 2.75. t = 2.945, p = 0.003, 95% CI = 0.277–1.384, d = 0.302), Reliable nutrition information source (Female: M = 1.26, SD = 0.75; Male: M = 0.94, SD = 0.89. t = 3.623, p < 0.001, 95% CI = 0.147–0.498, d = 0.389), females all scored significantly higher than males. See Fig. 1.

Age

There were significant differences between participants in different age groups on Variety of food affects food intake (F = 3.764, p = 0.024, η² = 0.013) and Alcohol intake and sugar sweetened beverages (F = 3.509, p = 0.031, η² = 0.012). The Bonferroni correction method showed that on Variety of food affects food intake, participants aged 35–54 years (M = 0.92, SD = 0.72) scored significantly higher than those aged 18–34 years (M = 0.76, SD = 0.65), p = 0.019, 95% CI = 0.02–0.31. On Alcohol intake and sugar sweetened beverages, participants aged 35–54 years (M = 2.54, SD = 1.29) scored significantly higher than 18–34 years (M = 2.25, SD = 1.25), p = 0.026, 95% CI = 0.03–0.56. See Fig. 2.

Education

There was significant difference in Total score (F = 5.649, p < 0.001, η² = 0.037), Energy density (F = 6.287, p < 0.001, η² = 0.041), Portion size (F = 3.152, p = 0.014, η² = 0.021), Reliable nutrition information source (F = 2.840, p = 0.024, η² = 0.019) by education level. The Bonferroni correction method showed that on Total score, master’s degree or above (M = 23.77, SD = 5.65) was significantly higher than junior high school or below (M = 16.25, SD = 3.88, p = 0.007, 95% CI = 1.30–13.74), college (M = 20.16, SD = 5.12, p = 0.028, 95% CI = 0.22–7.00), undergraduate (M = 21.93, SD = 6.34, p = 0.02, 95% CI = 0.17–3.51). In terms of Energy density, master’s degree or above (M = 11.15, SD = 2.63) was significantly higher than junior high school or below (M = 7.50, SD = 2.07, p = 0.002, 95% CI = 0.86–6.45), college (M = 9.39, SD = 2.45, p = 0.011, 95% CI = 0.24–3.29), undergraduate (M = 10.29, SD = 2.79, p = 0.012, 95% CI = 0.11–1.61); undergraduate was significantly higher than junior high school or below, p = 0.043, 95% CI = 0.05–5.54. On Reliable nutrition information source, master’s degree or above (M = 1.33, SD = 0.75) was significantly higher than junior high school or below (M = 0.50, SD = 0.54), p = 0.039, 95% CI = 0.02–1.63. See Fig. 3.

Income

Income was significantly different on Portion size (F = 4.378, η² = 0.022). The Tamhane’s T2 method showed that participants with a per capita monthly household income of >8000RMB (M = 7.77, SD = 2.89) scored significantly higher than participants with <3000RMB (M = 6.46, SD = 2.75, p = 0.041, 95% CI = 0.03–2.57) and 3000–5000RMB (M = 6.91, SD = 2.27, p = 0.014, 95% CI = 0.12–1.60). See Fig. 4.

Residence

Place of residence was significantly different on Total score and Portion size. The independent samples t-test showed that on Total score, participants in cities and towns (M = 22.47, SD = 6.24) scored significantly higher than in countryside (M = 20.30, SD = 5.38), t = 2.922, p = 0.004, 95% CI = 0.710–3.622, d = 0.373. On Portion size, participants in cities and towns (M = 7.58, SD = 2.77) scored significantly higher than in countryside (M = 6.35, SD = 2.45), t = 3.726, p < 0.001, 95% CI = 0.581–1.876, d = 0.471. See Fig. 5.