2.1. Sampling

This study is part of a larger project that aims to understand the relationship between digital platform usage and childhood obesity in China. We focused on Chinese teenagers aged between 11 and 16 for the following considerations. Firstly, children that are too young may not have enough pocket money to purchase food independently from their parents. Thus, it is their caregivers who make food choices. Moreover, many high schools in China require students to live in dorms on campus. These adolescents tend to have enough money for food and receive limited parental intervention on their usage of digital platforms. Consequently, although they have not reached adulthood, their behaviors regarding digital platform usage and food choices can be similar to those of young adults.

Taken together, our population of interest should not be too young or too old. In China, ten is considered a milestone in one’s childhood. Once they are over 10 years old, children are expected to take more responsibilities, and parents also allow more independence for them. Additionally, as there can be large regional differences in terms of the age when children enter the school system, most children should finish middle school and start high school at 16 years old. Hence, 11–16 is the age range of our target population.

Prior to the content analysis, we conducted interviews to understand what digital platforms our target population used to search for food-related information. A total of 28 interviews were conducted. Five digital platforms were mentioned the most: Bilibili, Douyin, Kuaishou, Xiaohongshu, and Pinduoduo.

These five platforms vary in their major affordances and target users. Douyin, Kuaishou, and Bilibili provide similar affordances that enable users to share videos. However, Bilibili allows for longer videos, whereas videos via Douyin and Kuaishou are usually limited to two minutes. Moreover, Kuaishou is targeted at residents of relatively low socioeconomic statuses and rural areas, whereas Douyin is targeted at residents of large cities.

Xiaongshu affords a wider range of media content sharing, including text-based messages, photos, and videos. Notably, commercials are allowed, and purchase links are made available via all four platforms mentioned above.

Finally, Pinduoduo is an e-commerce platform similar to Amazon. Most product information is presented as text and images, with fewer videos. As the current content analysis only involved the data from publicly available Internet services, institutional approval was not required.

We collected F&B content data from these platforms through web crawling techniques between January 2022 to February 2023. Given the differences between the platforms, we adjusted our data collection. As there is a food section via Bilibili and Pinduoduo, we searched food-related posts in this section. Specifically, we accessed 14,436 videos through Bilibili. We deleted similar videos and employed a stratified random sampling. Specifically, we calculated the percentages of the five categories of videos in the food section of Bilibili (i.e., cooking, taste test, food exploration, picnics, and live records of food). Next, we randomly selected videos from these categories and adjusted their numbers based on their percentages. In total, 600 videos were included for formal coding.

As for Pinduoduo, we accessed the top 600 food-related posts by employing MobDuos data analysis software [39]. After deleting duplicate and irrelevant posts, 500 were kept for formal coding.

The other three platforms do not have a food section, so we used different data collection techniques. Specifically, we accessed 12,880 videos via Kuaishou using 45 food-related hashtags, which were selected through an exhaustive search by two graduate students. Again, we calculated the percentages of these hashtags and randomly selected 606 videos for formal coding based on the percentages of these hashtags.

The data collection methods used for Douyin and Xiaohongshu were similar. Following two major social media indices in China [40,41], we accessed the top 100 food-related influencers on Douyin and Xiaohongshu. Then, we randomly selected six posts from each influencer, leading to 600 posts for each platform. Therefore, the final sample size was 2906.

Notably, as our goal is to conduct a systematic investigation on the digital food environment in China, we did not distinguish the types of food-related posts. Any posts about the food were sampled, whether offered by commercial companies or ordinary users.