Development and validation of an online tool to assess perceived portion size norms of discretionary foods

Perceived portion size norms, the typical perception of how much food individuals consume at one sitting, significantly influence food consumption and portion control. The ubiquity of large serving sizes has likely shifted these norms towards larger portions, particularly for discretionary foods (high in saturated fats, added sugars, salt, or alcohol). This is concerning because overconsumption of these foods contributes to excessive energy intake, poor diet quality, and chronic diseases. Reducing inflated portion size norms is a potential strategy to improve dietary habits. Previous research has identified different types of perceived portion size norms, such as personal and social norms, and proposed models like the 'norm range model' that suggest portion sizes within the lower end of the perceived normal range may subtly influence intake. However, validated tools to assess portion size norms for discretionary foods are lacking. Existing assessment methods, including self-selected portion size tasks, normality judgment tasks using images, and estimation tasks, lack consistency and validation. This study aimed to develop and validate a user-friendly, accessible online image-based tool to address this gap, minimizing biases like unit bias and social desirability bias by presenting a wide range of portion size options.

The literature review highlights the importance of perceived portion size norms in shaping food consumption behaviors and their potential distortion due to the prevalence of large serving sizes in the modern food environment. It discusses the various types of perceived portion size norms (personal and social) and the 'norm range model' which suggests that serving sizes at the lower end of the perceived normal range may unconsciously nudge individuals to consume less. The review also addresses the limitations of existing tools used to measure portion size norms, pointing to inconsistencies in food presentation, limited portion size options, and a lack of validation. The need for a consistent and validated tool, specifically for discretionary foods, is emphasized. The potential of online, image-based tools to overcome some limitations of real-food assessment methods is acknowledged.

This study developed and validated an online image-based tool to assess perceived portion size norms for 15 commonly consumed discretionary foods in Australia. The tool, created using Qualtrics, features image series of each food showing eight (or six for bottled/canned drinks) increasing portion sizes. Participants (114 adults aged 18–65) participated in a within-person crossover validation study. In an in-person laboratory session, participants selected their perceived portion size norm twice for each food: once using the online tool viewing food images and once using equivalent real food portions at food stations. The order of methods and food presentation was randomized. The portion size options ranged from small to large, based on package sizes, typical portion sizes from previous literature, and pilot study feedback. A demographic section collected data on gender, age, height, weight, postcode, physical activity level (PAL), and cooking confidence. Data analysis involved cross-classification to determine the proportion of correct and adjacent matches between the two methods and intra-class correlation coefficients (ICC) to assess agreement. The proportion of over- and underestimation was analyzed, and the relationship between cooking confidence and agreement was investigated using the Chi-square test.

A total of 114 participants completed the validation study. Cross-classification analysis showed that 91% of comparisons were classified as correct or adjacent matches (within one portion size option of the selection using the other method). The ICC across all foods was 0.85, indicating good agreement between the online tool and real-food assessments. Good to excellent agreement (ICC > 0.75) was observed for 12 out of 15 foods. Three foods (chocolate bar, muffins, banana bread) showed moderate agreement (ICC between 0.5 and 0.75). The median energy intake for perceived portion sizes varied depending on food type. For most foods (13 of 15), the median energy reported was similar for both methods, though interquartile ranges varied. Participants with high cooking confidence had a significantly higher proportion of correct matches than those with low cooking confidence, specifically for four out of 15 food items. No significant effect was found for presentation order.

The high agreement between the online image-based tool and real-food assessments validates its use in future studies. The findings are consistent with other studies using image series, though variations exist due to differences in test foods and study design. Moderate agreement for some foods (banana bread, muffins, chocolate bars) might be due to the wide variation of portion sizes available for these foods or difficulty distinguishing details from two-dimensional images. The positive association between cooking confidence and agreement suggests that individuals with higher food literacy might be better at estimating portion sizes. However, this aspect requires further investigation. The study's strength lies in the comprehensive design, large enough sample size, pilot testing, and inclusion of a range of portion sizes to minimize bias. However, limitations include the convenience sample predominantly consisting of young, female university students, the potential influence of one method on the other within a session, and that the food was presented on white plates, which might not reflect real-life eating scenarios.

This new online image-based tool for assessing perceived portion size norms of discretionary foods shows good agreement with real-food assessments, making it a valuable tool for future research. Its ease of use and accessibility make it suitable for various settings. Understanding portion size norms can inform public health messages, food labeling, and industry practices to create food environments that encourage healthier portion choices, thus helping reduce overconsumption of discretionary foods. Future research could explore the tool's generalizability across diverse populations and expand it to include healthier food types.

The study's limitations include the convenience sample which primarily comprised young female university students, not fully representing the general population. The within-session design, despite randomization, might have introduced some bias. The presentation of foods on white plates may not reflect real-life eating habits. Future studies should explore the influence of participant characteristics and use a more representative sample.