Ultra-Processed Food and Plant-Based Meat — Instagram Sentiment Analysis

Table of Contents

Introduction: Ultra-Processed Foods and Plant-Based Meats

Mainstream conversations around plant-based meat products have become increasingly critical in recent years, particularly due to their perceived lack of naturalness and high level of processing. This narrative has been amplified by mainstream media, coinciding with growing scrutiny over ultra-processed foods (UPFs) and the inclusion of plant-based meats in the criticism of these products.

The debate surrounding ultra-processed foods is not without its contradictions. The paradox of the ultra-processed narrative is evident in that some ultra-processed items, like wholemeal bread, may offer health benefits, while “whole” foods, such as red meat, are linked to adverse health outcomes. Furthermore, despite research indicating that the processed/plant-based meat narrative has taken hold in mainstream media (and analysis of social media platforms like Twitter/X and Reddit reinforce these claims) academic studies on UPFs rarely include plant-based meats.

Even if plant-based meat is classified as ultra-processed (and many do meet the criteria due to the types of ingredients and level of processing), their nutritional value often defies this categorization. Unlike most ultra-processed foods that are high in saturated fat, sugar, sodium, and low in fiber, plant-based meats, when evaluated using nutritional frameworks, often score higher than their meat analogues. However, it’s important to note that nutrition scores can vary depending on the brand, ingredients, and product.

The ultra-processed food debate, and the subsequent inclusion of plant-based meat, holds significant implications from a consumer adoption standpoint. Plant-based meats are crafted to replicate the taste and texture of meat, with the intention that consumers reduce their meat consumption without sacrificing the familiar experiences associated with meat, such as certain dishes or cooking methods. This ability to serve as a bridge between conventional meat and plant-based alternatives is a pivotal objective for the burgeoning plant-based meat industry. Beyond Burger’s website states: 

“We’re not going to get people to shift away from eating animal meat by offering them a salad — we need to offer them a better meat option.”

Impossible burger has recently rebranded to reinforce the product as a meat replacement — which reinforces their goal to offer a meat substitute for meat eaters (90% of their consumers are omnivores). 

Transitioning away from meat-centric diets is crucial from both an environmental and human health perspective. Livestock emissions alone are responsible for 11.1-19.6% of all global greenhouse gas emissions. Diets high in meat, particularly processed meats, are associated with increased rates of chronic disease. Additionally, the livestock industry is responsible for biodiversity loss, the acceleration of zoonotic diseases, antibiotic resistance, environmental racism, and human and animal rights violations.

Plant-based meat alternatives, while not a cure-all for the problems associated with conventional livestock are designed to help reduce meat consumption and serve as a useful stepping stone toward reducing meat intake. Therefore, it is crucial to understand the criticism these products face, especially concerning their classification as ultra-processed foods.

This analysis aims to explore the narrative about plant-based meats and ultra-processed foods on Instagram over the past five years, providing insights into the concerns, challenges, and themes that emerge from this discourse.

Ultra-Processed Foods: Origins and Current Discourse

Despite what appears to be a recent rise in Google results for UPF (2019 – 2022 with a dropoff in 2023, Figure 1), Brazilian doctor Carlos Monteiro coined the origins of ultra-processed food terminology nearly two decades ago in 2009. 

The term ‘ultra-processed food’ is part of the NOVA system (not an acronym—NOVA means “new” in Portuguese), which categorizes food based on the extent and purpose of its industrial processing, not its nutritional quality. The system, composed of four distinct categories determined by processing level (Figure 2), is rooted in a critique of the industrialized Western food system and is considered a socio-political framework rather than a nutritional framework, as argued by Jenny Chapman (2022) in Processing the Discourse on Ultra-Processed Foods. This socio-political framework is reinforced by Monteiro’s research and his stance that UPFs are designated by their processing and the purpose of their processing.

The purposes involved in the processing of UPFs implicate the industrialized, global food system, reinforcing Chapman’s argument that the NOVA system functions more accurately as a socio-political critique than a foundation for nutrition science. According to Monteiro, UPFs are produced by multinational companies headquartered in the U.S. or Europe. These companies heavily market their products, including to children, with vivid packaging designed to be hyper-palatable (Figure #3). Monteiro notes the impact of Global North food systems on traditional food systems in his native Brazil. The blending of criticism toward the industrial food system with the implied health value of NOVA 1 and 2 over NOVA 3 and 4 results in a confusing approach to nutrition at best.

Despite its noted barriers to effectively grouping foods by their nutritional value (not to mention the explicit exclusion of nutrients in Monteiro’s early writings), the NOVA system has been adopted by nutritional researchers as a tool to categorize foods and study health implications. In an umbrella review of academic literature this year, researchers found diets high in ultra-processed food were connected with 25 adverse health impacts, including heart disease, cancer, and type 2 diabetes, reinforcing health concerns.

The use of the NOVA system, which spans multiple disciplines including socio-political and nutritional aspects, forms a somewhat precarious foundation for addressing potential health issues related to processed foods. Recent studies have highlighted a lack of consensus even among food experts about the classification of various foods within the NOVA system. This uncertainty is exacerbated by additional research linking ultra-processed foods (UPFs) to adverse health outcomes, presenting a confusing situation: ultra-processed foods are shown to be detrimental to health, yet their exact definition remains unclear. Chapman further encapsulates this lack of clarity in a podcast, where she shared that when people are asked about UPFs, they cite a range of concerns for avoiding them, from general health risks to their tendency to be packaged in plastics.

Other recent publications have further revealed the multi-disciplinary discourse about ultra-processed foods. Marion Nestle, a U.S. food researcher, has referenced the ultra-processed food debate in numerous writings and podcasts. Marion’s decades of work have focused on the food politics of our industrialized food system—the lobbying and marketing of politicians by ‘Big Food’ at the expense of public health. U.K.-based infectious disease doctor Chris Van Tulleken writes about the implications of UPFs from an environmental and social justice perspective (as well as health) in his book, Ultra Processed People (2023). The book is a New York Times bestseller.

Instagram Insights: Capturing the Pulse of Online Discourse

The increase in consumer awareness and media focus on ultra-processed foods has sparked a significant debate around plant-based and vegan options, with plant-based meats drawing particular scrutiny in the media and on social media platforms. Despite the low occurrence in media coverage—only 1% to 3.5% of headlines in major U.S. and U.K. outlets from 2023 to 2024 mentioned plant-based meat—the sentiment in these headlines was overwhelmingly negative, often critiquing their processing and expressing skepticism about added ingredients.

Our study focuses on the discourse around plant-based meats and ultra-processed foods, aiming to unravel the intricacies of themes, sentiment, and language used in online conversations. We analyze sentiment on Instagram to uncover evolving trends and delve into the language, themes, and primary concerns discussed over the past five years. We hypothesized that online sentiment would predominantly mirror negative trends observed in media headlines and social media, especially concerning processing and ingredients. By examining online media posts and their corresponding sentiments, we aim to enrich the dialogue surrounding plant-based meats and alternative proteins, addressing widespread concerns and misinformation about plant-based diets.

Descriptive Results

Utilizing the social media analysis tool CrowdTangle, we initially identified 541 Instagram posts (2019-2024) that met our keyword criteria for ultra-processed foods and plant-based meats (see Appendix). To delve deeper, we employed a data analysis tool, Google Colab, to sift through the text of these posts, cleaning and identifying posts with sentences containing keywords associated with both ultra-processed and plant-based products, as outlined in the appendix. While this narrowed our dataset to 302 posts, concentrating on sentences that incorporated both keyword groups ensured that our results were highly relevant to our research, enhancing the precision of our analysis.

Post Frequency

To effectively track and analyze the temporal dynamics of Instagram posts relevant to our study, we plotted 3-month moving averages from 2019 to 2024. As depicted in the graph, the number of posts began at a relatively low frequency in 2019, then showed a marked increase, reaching a peak in early 2022. Following this peak, there was a notable decline in post frequency (Figure #4).

Word Cloud Visualisation

The word cloud in Figure 5 summarizes our dataset’s keywords. The most common terms used are ‘Processed,’ ‘Food,’ and ‘Plant-based.’ ‘Meat’ and ‘Vegan’ are less common. Other keywords like ‘Environment,’ ‘Burger,’ ‘Soy,’ and ‘Ingredient’ are mentioned sporadically. While the word cloud only provides a surface-level analysis, it offers a visual summary of the underlying data trends. The results of the word cloud are expected given the search terms.

Word Frequency: Unpacking “Fake”

We then counted the most frequent words. Given our search terms (Appendix), the results were as expected and aligned with the keyword search (ultra-processed, meat, plant-based meat, etc.). However, we found that the word “fake” appeared 236 times and was among the top ten most frequent words in our dataset, indicating a potential negative bias in discussions surrounding plant-based meats.

When viewed temporally with the count and percentage of posts, the results indicate a sharp increase in posts containing the word “fake” in discussions about plant-based meat and ultra-processed foods from 2019 to 2023, with the highest percentage of “fake” posts (55%) occurring in 2021 (Figure #3).

Sentiment Score and Statistical Analysis

To analyze sentiment trends over time, we applied the VADER sentiment analysis tool to sentences from Instagram posts containing relevant keywords. We focused on analyzing sentences rather than entire posts to capture the specific context in which relevant keywords (Appendix) were used, ensuring more precise sentiment analysis. We aggregated sentiment scores by month and calculated the median for each. If a post included multiple relevant sentences, their scores were combined. We then plotted these monthly median sentiment scores, using a simple linear regression to add a trend line for clearer visualization of the trends.

Statistical Analysis

The regression analysis yielded the following parameters:

• Slope: -0.0036
• Intercept: 0.1216
• R-squared: 0.0793
• P-value: 0.03077

These results indicate a statistically significant, albeit minor, downward trend in aggregate sentiment over the observed period. The negative slope of -0.0036 suggests a gradual decrease in median sentiment, corroborated by the statistical significance indicated by the p-value of 0.03077. Although the change is slight, it is statistically significant and suggests that sentiments on the topics have become slightly more negative over time.

The relatively low R-squared value of 0.0793, however, highlights that the model explains only a small portion of the variance in sentiment scores. This suggests that other unaccounted-for variables may significantly influence sentiment trends.

Negative sentiment towards plant-based and ultra-processed foods is on the rise. Further research is needed to identify factors influencing this trend and to design more effective communication strategies to address the negative sentiment. The negative trends are not unexpected based on prior research conducted in 2023. A comprehensive study, mostly on Twitter/X and Reddit, conducted by the Changing Markets Foundation on plant-based diets found that nearly one-quarter of the posts attacked plant-based meat and dairy as ultra-processed.

To provide a deeper analysis of themes within these posts, we conducted a Qualitative examination of the Instagram posts.

Qualitative Analysis: Themes and Language

Thematic Analysis Using ChatGPT-4

To delve deeper into the discourse surrounding plant-based meats and ultra-processed foods, we employed ChatGPT-4 to conduct a thematic analysis and subsequent sentiment analysis by theme. Our analysis focused on identifying themes and assessing the overall sentiments associated with each theme.

Identified Themes:* 

Health Concerns: This theme was predominant with 274 sentences* referencing terms like “processed,” “additives,” “preservatives,” “synthetic,” and “unhealthy.” The discourse within this theme typically reflects skepticism about the health benefits of processed foods and meat alternatives, underlining apprehensions about their ingredients and potential health consequences.

Dietary Choices: Comprising 192 sentences, this theme revolves around terms such as “vegan,” “plant-based,” “meat substitute,” “diet,” and “nutrients.” Discussions in this category often explore the choices individuals make, such as adopting plant-based diets or using meat substitutes, and are frequently framed in terms of health and ethical considerations.

Food Authenticity: With 120 sentences, this theme focuses on the debate between “real meat” and “fake meat,” and the authenticity of products marketed as substitutes. It captures the significant dialogue concerning consumer perceptions and the marketing of these products as true alternatives to traditional meat options.

*As some of these sentiments overlap, these sentences are not all unique (i.e. sentences about health may also be categorized under diet and authenticity depending on the content). 

In a visualization of the themes over time, the graph reveals trends in each of these themes, with health concerns showing a peak in 2022 (Figure #8). 

Health Concerns and Dietary Choices peaked in mid-2022, reflecting increased consumer awareness and concern about the health implications of processed foods and meat alternatives, and growing interest in plant-based and vegan diets. Food Authenticity followed a similar pattern. Overall, the graph suggests that discussions around these themes have started to normalize after a period of heightened interest in 2021 – 2022. 

Analysis of Themes

The Health Concerns theme is predominantly negative. The statements express skepticism and distrust towards processed and imitation meats. Many people believe that consuming these products can pose potential health risks and that they are unnatural. Comments often use phrases such as “highly processed,” “inflammatory seed oils,” and “fake” to describe these products. People feel that these alternatives are not just inferior to real meats in terms of health benefits but are actively harmful. The negative sentiment is also reflected in statements criticizing the ingredient quality and the industrial processes used to manufacture these foods.

The sentiments within the Dietary Choices theme are mixed, reflecting a balance between the benefits and drawbacks of plant-based diets. Positive sentiments are evident in discussions about the ethical and health motivations behind choosing vegan or plant-based diets. However, there is also a significant critique of plant-based products that are perceived as overly processed or unhealthy, indicating that while the choice to avoid animal products is seen positively, the alternatives offered are not always viewed favorably. This dichotomy suggests that there is appreciation for the intent behind dietary shifts but criticism of the execution, particularly regarding food processing and product composition.

The Food Authenticity theme carries a strongly negative sentiment, focusing on plant-based meats’ authenticity and perceived quality. The dialogue is heavily centered on the notion that these products are “fake” and inferior to real meat. Statements often argue that plant-based products fail to offer their traditional counterparts’ nutritional and sensory benefits and critique the heavy processing involved in their production. There’s also a notable distrust of marketing practices in this area, with suggestions that consumer perceptions are being manipulated.

Examples of Themes

Below, we list 10 examples from each theme. These sentences were pulled at random and represent the overall categorization of each theme. 

Health Themes

1. “It doesn’t matter what type of diet you follow, opting for fake meat—highly processed and full of inflammatory seed oils—is absolute nonsense.” 2. “The proliferation of fake cheese, meat, milk, and butter, which are marketed as healthier despite their extensive processing and long ingredient lists, is alarming.” 3. “The fake meat market seems propped up by marketing rather than genuine consumer demand.” 4. “Plant-based ‘meats’ contain the same harmful ingredients found in many processed foods, yet they are marketed as healthy.” 5. “The Food Pyramid and My Plate promote a diet heavy in refined grains, seed oils, and sugars, prioritizing industrial ingredients over natural fats.” 6. “Plant-based alternatives require excessive energy for production, contradicting their perceived sustainability.” 7. “The second and third ingredients in many fake meats, like pea protein and seed oils, are less nutritious and potentially harmful.” 8. “While it’s important to support sustainable practices and reduce processed food consumption, the overwhelming processing involved in plant-based meats cannot be ignored.” 9. “Despite their popularity, vegan and plant-based processed foods may undermine health due to their high processing levels.” 10. “Plant-based meats, though often branded as health foods, are typically highly processed and packed with unhealthy ingredients.”

Food Authenticity

1. “Mock meat, referred to in various terms such as fake, imitation, or vegan meat, is essentially an engineered substitute for real meat.” 2. “Critics argue that despite popular belief, fake meat is inferior to real meat in terms of health benefits.” 3. “Consumers are urged to resist the persuasive marketing and influencer endorsements that portray processed foods as healthy.” 4. “Seitan, a vegan meat substitute made from wheat dough, offers a less processed alternative within the plant-based spectrum.” 5. “Recent regulatory changes in SA have banned the use of traditional meat-related terms like nuggets and ribs for plant-based alternatives, reflecting concerns over misleading labeling.” 6. “The search for meat alternatives that aren’t laden with additives has increased, with some consumers preferring options that taste more like traditional meat.” 7. “Lab-grown meat is criticized for not being natural or sustainable, indicating that it fails to address core issues within the food industry.” 8. “The stark contrast between eating real, freshly prepared meats and consuming processed plant products designed to mimic meat underscores the authenticity issue.” 9. “Real food advocates point out that genuine food items don’t require extensive ingredient lists full of unpronounceable components.” 10. “Dissatisfaction with the taste of faux meat products, compared to their authentic counterparts, is common among consumers.”

Dietary Choices

1. “Fake meat may be plant-based, but that doesn’t mean it grows on trees!” 2. “Navigating the grocery store aisles doesn’t have to be a challenge, with fresh, whole, and healthy foods always being a great option.” 3. “Plant-based doesn’t always mean healthy, especially when it comes to highly processed foods like some vegan meats.” 4. “Vegan diets are great, but let’s not pretend that all vegan foods are automatically healthy.” 5. “Choosing a plant-based diet doesn’t have to mean relying on processed foods.” 6. “It’s not just about being vegan, it’s about being health-conscious.” 7. “The rise of vegan diets has brought attention to plant-based nutrition.” 8. “Making dietary choices can be tough, but focusing on whole, unprocessed foods is key.” 9. “Veganism isn’t just a diet, it’s a lifestyle choice that involves careful consideration of food sources.” 10. “Plant-based eating is more than just a trend; it’s about making healthier dietary choices.”

Posts with Top Engagement: Comment Section

The final step in our analysis involved examining the comments from top-engaged posts to better understand community engagement and sentiment regarding plant-based meats and ultra-processed foods. We focused on engagement rates (the total number of interactions a post receives divided by the total number of followers, multiplied by 100%) as a metric, considering the variability in follower counts across accounts. An engagement rate above 3%, deemed good by industry standards, was used as our benchmark, filtering our dataset to 38 posts with engagement rates ranging from 3% to 12%. We manually copied comments from the top 38 posts and utilized ChatGPT-4 to clean and review the text.

Data Summary: 38 Posts

Overview Instagram Profiles: Top Engaged Posts

We manually reviewed each profile, associated business, or links to better understand the profiles of the individuals/entities in the top most-engaged posts. 

Profiles of Posters

1. Health Advisors: Individuals offering general health advice, often advocating for specific dietary approaches.
2. Carnivore Diet Advocates: Promoters of a meat-only diet, which includes high-profile advocates like MD Robert Kiltz and influencer, Kaylor Betts
3. International Waste Management Company: Engages in discussions likely around sustainability and environmental impacts of food production.
4. Mental Wealth Project: Focuses on the psychological or lifestyle aspects of dietary choices — pro meat. 
5. Online Meat Purchasing Store: Emphasizes the benefits of meat consumption, possibly with a focus on quality and sourcing.
6. Paleo and Gluten-Free Advocates: Promote diets that exclude certain food types considered modern additions to the human diet.
7. Regenerative Farming Advocates: Discuss agricultural practices aimed at restoring soil health and enhancing ecological balance.
8. Primal Health Coaches: Offer guidance based on the primal lifestyle, focusing on diet and fitness principles that mimic ancient human practices.
9. Healthy Cooking Enthusiasts: Share insights on preparing healthy meals, often highlighting natural and whole ingredients.
10. Business Owners in the Meat and Health Industry: Such as Heat and Soil supplements, focusing on holistic health and nutrient-dense meat products.
11. Digital Nomads and Parents: Share personal and lifestyle content that may intersect with dietary choices and family health.

Yearly Engagement and Sentiment Analysis of Top-Engaged Posts

We then applied the previously used VADER sentiment analysis to determine if sentiment of these top engaged posts over time aligns with the earlier statistical findings, which showed a small but meaningful decline. 

Results suggest that over time average engagement has risen, while average sentiment has declined (Table #1)

Year	Total Interactions	Average Engagement	Average Sentiment
2019	43	4.05%	0.40
2020	7392	4.62%	0.08
2021	45,082	5.67%	-0.14
2022	31,845	6.59%	-0.25
2023	11,857	7.17%	-0.52

Visualizing the engagement and sentiment analysis of top-engaged Instagram posts related to plant-based meats and ultra-processed foods from 2019 to 2023 reinforces these findings (. Total interactions (blue bars) show a significant increase from 2020 to 2022, peaking in 2021, followed by a decline in 2023. The average engagement rate (green line) steadily increases, indicating growing interaction relative to audience size. The average sentiment score (red line), analyzed using VADER sentiment analysis, shows a decline, becoming increasingly negative over time, with the lowest point in 2023. This analysis, focusing on highly engaged content, reveals increasing community interest but rising negativity in sentiment, highlighting the need for further exploration of underlying factors driving these trends.

Visualizing the engagement and sentiment analysis of top-engaged Instagram posts related to plant-based meats and ultra-processed foods from 2019 to 2023 reinforces these findings. Total interactions (blue bars) show a significant increase from 2020 to 2022, peaking in 2021, followed by a decline in 2023. The average engagement rate (green line) steadily increases, indicating growing interaction relative to audience size. The average sentiment score (red line), analyzed using VADER sentiment analysis, shows a decline, becoming increasingly negative over time, with the lowest point in 2023. This analysis, focusing on highly engaged content, reveals increasing community interest but rising negativity in sentiment, highlighting the need for further exploration of underlying factors driving these trends.

Discussion, Conclusions, and Recommendations

Per our review, over the past five years, sentiment and post content about plant-based meat and ultra-processed foods have varied, with a statistically significant (though slight) downward trend in overall sentiment. 

Based on our thematic review of Instagram posts, there is a discernible pattern of unfavorable attitudes towards plant-based meats and ultra-processed foods, predominantly driven by concerns over their authenticity and potential health impacts. These sentiments may foster demand for more natural and minimally processed dietary options and reveal skepticism towards the current offerings by the food industry. Despite an appreciation for the shift towards plant-based eating, there appears to be dissatisfaction with the current processing methods over the past five years. This trend is reflected in Boston Consulting Group (BCG) research from 2023, which noted that the narrative around alternative meat as processed or artificial has “taken hold.” BCG argued that plant-based companies should be more transparent about ingredient lists and emphasize the “plant” in plant-based products.

Some of the top-engaged posts included carnivore influencers, meat industry proponents, and advocates for regenerative farming. This suggests that some of the discourse about plant-based meat and ultra-processed food is influenced by those with vested interests in undermining trust in plant-based alternatives. These posts often underscore a preference for traditional dietary choices and emphasize the benefits of meat consumption and natural foods over plant-based alternatives.

Noteworthy — Culture Wars and The Carnivore Diet

Alternative meats are currently embroiled in a cultural debate. In our research, we prominently featured influencers such as Dr. Kiltz, Kaylor Betts, Nicolas Gustafson, and Brian Sanders (total followers = 932,000 as of May 2024). These individuals, active over the past five years, are vocal critics of plant-based meat and contribute to a narrative that links traditional meat consumption with masculinity. They account for 10.4% of the most engaged posts. We manually searched their accounts for mentions of meat and masculinity, often finding content about increasing testosterone or disparaging ‘feminism.’ 

Our findings reflect results from the Changing Markets Foundation, which found that 11% of a dataset of disparaging posts about plant-based diets (83,790 posts, or 9% of the total) relates to culture wars. These are polarizing, identity-driven conversations aimed at sowing division, such as questioning the masculinity of men who choose plant-based diets. This perspective is underscored by articles like the New York Times’ “Meet the Men Who Eat Meat (and Only Meat)” from April 30, 2024. The continued popularity of the carnivore diet and the high engagement with these influencers’ posts indicate that the association of meat with masculinity remains a potent cultural force.

Limitations and Further Research

Our analysis faces several limitations, primarily due to the reliance on a relatively small dataset of Instagram posts spanning five years. This dataset’s scope prevented the examination of sentiments by demographic details such as age and location, which would provide a deeper understanding of consumer attitudes. The analysis also disproportionately reflects sentiments from 2021, as over half of the top-engaged posts originated from that year, potentially biasing our findings. Additionally, the focus on highly engaged posts, often from profiles with specific interests, may not fully represent broader public opinion. Certain accounts may be more highly featured, leading to overrepresenting specific viewpoints.

Furthermore, using specific keywords (Appendix) in our thematic analysis with ChatGPT-4 may exclude more nuanced discussions and subtle perspectives. Lastly, the absence of geographical data in our tool, CrowdTangle, means we cannot analyze the regional distribution of sentiments, highlighting the need for a more geographically diverse data collection approach to understand regional variations in sentiment better. 

Additionally, while not included in this study due to insufficient data, our observations suggested some negative sentiments among vegans or vegan activists. More studies are needed to understand better how sentiments about plant-based meat are influenced by members of the vegan movement who oppose ultra-processed foods.

Our findings highlight the growing negative sentiment towards plant-based meats and ultra-processed foods on Instagram, driven by concerns over authenticity and health impacts. Future research should focus on broader, more diverse datasets to capture a more comprehensive range of consumer perspectives, including demographic details such as age and location.

Methodology and Appendix

Step 1: Data Collection and Keyword Selection

To gather data on public discourse surrounding ultra-processed foods (UPF) and plant-based meats (PBM) on social media, we utilized the social listening tool, CrowdTangle. CrowdTangle is a content discovery and social monitoring platform owned by Meta Platforms, Inc., that tracks the spread and engagement of content across Facebook, Instagram, and Reddit.

We focused on Instagram posts from the past five years (2019-2024), using specific keywords related to UPF and PBM:

Ultra-Processed Foods Keywords	AND	Plant-Based Meat Keywords
‘ultra processed foods’ (OR) ‘ultra processed food’ (OR) ‘ultra-processed foods'(OR) ‘ultra-processed food'(OR) ‘highly processed'(OR) ‘overly processed'(OR) ‘processed food'(OR) ‘processed foods'(OR) ‘UPF’ (OR)	AND	‘plant-based meat’ (OR) ‘plant based meat’ (OR) ‘alternative meat’ (OR) ‘fake meat’ (OR) ‘vegan meat’ (OR) ‘plant-based meats’ (OR) ‘plant based meats’ (OR) ‘alternative meats’ (OR) ‘fake meats’ (OR) ‘vegan meats’ (OR) ‘meat substitute’ (OR) ‘meat alternative’ (OR) ‘synthetic meat’ (OR) ‘plant-based burger’ (OR) ‘plant based burger’ (OR) ‘plant based burgers; (OR) ‘plant-based burgers’ (OR) ‘vegetarian meat’ (OR) ‘Vegetarian meats’ (OR)

Final results: 541 Instagram Posts

Step 2: Sentiment Analysis and Data Sorting

To analyze the sentiment of the collected Instagram posts, we followed these steps:

Steps:

1. Data Upload: We uploaded the dataset, consisting of 541 posts, into Google Colab for processing.
2. Sentiment Analysis Tool: We utilized VADER (Valence Aware Dictionary and sEntiment Reasoner) for sentiment analysis. VADER is a lexicon and rule-based sentiment analysis tool specifically attuned to sentiments expressed in social media.
3. Initial Sentiment Analysis: VADER sentiment analysis was initially applied to all 541 posts to gain a general understanding of the sentiment distribution.
4. Data Filtering: To focus on more specific insights related to our research topics, we filtered the dataset to include only posts containing sentences with both keywords “ultra-processed foods” (UPF) and “plant-based meat.” This filtering resulted in a subset of 302 posts.
5. Targeted Sentiment Analysis: We then applied VADER sentiment analysis exclusively to these specific sentences within the 302 posts. If multiple sentences in a post contained both keywords, we aggregated the sentiment scores to obtain an overall sentiment for that post.
6. Sentiment Score Aggregation: For posts with more than one relevant sentence, sentiment scores were aggregated by calculating the mean sentiment score. This approach ensured that the overall sentiment of the post was accurately represented.
7. Data Sorting: Posts were then sorted based on their aggregated sentiment scores, categorizing them into positive, negative, and neutral sentiment groups for further analysis.

Rationale:

• Google Colab: We chose Google Colab for its collaborative features, ease of use, and ability to handle large datasets with extensive computational resources.
• VADER Sentiment Analysis: VADER was selected due to its effectiveness in analyzing social media text, which often includes slang, emoticons, and varying punctuation.
• Filtering and Aggregation: Filtering the data ensured that our analysis was focused on posts most relevant to our research topics. Aggregating sentiment scores provided a more comprehensive view of each post’s overall sentiment.

Step 3: Thematic Analysis Using ChatGPT-4

To gain deeper insights into the language and comments related to ultra-processed foods (UPF) and plant-based meat (PBM), we employed ChatGPT-4 for thematic analysis of the 302 posts with relevant sentences. This approach allowed us to uncover core themes, capture contextual nuances, and identify recurring patterns in the discourse.

Steps:

1. Data Preparation: We uploaded all relevant sentences from the filtered dataset, consisting of 302 posts, into ChatGPT-4. These sentences were selected based on their inclusion of both UPF and PBM keywords.
2. Framework for Analysis: We provided ChatGPT-4 with the keywords listed in Step 1 as a framework for the thematic analysis. These keywords included “ultra-processed foods,” “plant-based meat,” “vegan meat,” and “processed vegan food.”
3. Thematic Analysis Process:
   • Keyword-Based Categorization: ChatGPT-4 was instructed to conduct a thematic analysis based on the provided keywords. This involved identifying and categorizing comments according to the key themes related to UPF and PBM.
   • Contextual Understanding: ChatGPT-4 analyzed the contextual nuances of the comments, helping to reveal the sentiments and attitudes expressed by users in their posts.
   • Pattern Recognition: By examining recurring patterns in the data, ChatGPT-4 was able to highlight key concerns, common topics, and other relevant themes.

Rationale:

1. ChatGPT-4: ChatGPT-4 was chosen for its ease of use and advanced natural language processing capabilities, which allowed for a detailed analysis of the social media posts
2. Keyword Framework: Using the keywords as a framework ensured that the analysis was focused on the most relevant themes and topics related to UPF and PBM.
3. Thematic Analysis: This method provided a structured approach to understanding the underlying patterns and sentiments in the data, offering richer insights than a simple sentiment analysis.

Step 4:  Analyzing Top-Engaged Posts

To gain insights into community engagement and sentiment, we focused on comments from the top-engaged Instagram posts related to ultra-processed foods (UPF) and plant-based meat (PBM). This step was divided into two parts: thematic analysis of comments and an overview of sentiment based on prior sentiment analysis.

Part 1: Thematic Analysis of Comments Steps

1. Identifying High-Engagement Posts:
   • Engagement Rate Calculation: We calculated the engagement rate for each post using the formula: (total interactions divided by followers) multiplied by 100%.
   • Selection Criteria: Posts with engagement rates above 3% were included, filtering our dataset to 38 posts with engagement rates ranging from 3% to 12%.
2. Data Extraction:
   • Manual Copying: Comments from the top 38 posts were manually copied to ensure accuracy (total comments = 3,618).
   • Data Cleaning and Preprocessing: We used ChatGPT-4 to clean and preprocess the text. This involved removing irrelevant content, such as spam or promotional messages, and standardizing the format for consistent analysis.
3. Thematic Analysis:
   • Keyword-Based Categorization: ChatGPT-4 categorized the comments based on the key themes related to UPF and PBM (as identified in Step #1).
   • Identification of Core Themes: The thematic analysis identified core themes.
   • Contextual Overview: This section provides a richer understanding of discussions on top engaged posts. 

Part 2: Overview of Sentiment Steps

1. Previous Sentiment Analysis:
   • Sentiment of Top-Engaged Posts: The sentiment scores of relevant sentences from the previous sentiment analysis (Step 2) were reviewed to provide an overview of the sentiment for the top-engaged posts.
2. Sentiment Distribution:
   • Positive, Negative, and Neutral Sentiments: The posts were categorized based on their sentiment scores into positive, negative, and neutral groups.
   • Sentiment Insights: This categorization helped in understanding the overall sentiment expressed in the most engaging content, offering insights into how the community feels about UPF and PBM.

Rationale

• Focus on Highly Engaged Content: Analyzing comments from posts with high engagement rates ensured that the analysis focused on content that resonated strongly with the community, capturing the most impactful discussions.
• Normalized Metric: Using engagement rates allowed for a fair comparison across accounts with varying follower counts, ensuring that both large and small accounts were evaluated on an equal footing.
• Accurate and Clean Data: Manual copying of comments ensured the accuracy of the data, while ChatGPT-4’s preprocessing capabilities ensured that the text was clean and standardized, providing a reliable dataset for analysis.
• Comprehensive Insights: By separating thematic analysis and sentiment overview, we gained a deeper understanding of both the qualitative themes and the quantitative sentiment expressed in the community’s most engaging posts.

---

Footnotes