Impact of Body Inclusive Advertising
Lainey Kasian
4/27/2022
Impact of Body Inclusive Advertising
Social media marketing has changed the way marketing materials are created, viewed, and interacted with. This has been met with a shift from certain companies on their marketing strategies. One specific approach that is becoming more popular is social issue marketing, specifically issues surrounding women’s body image issues and increased representation of various body types in advertisements. Through a sentiment analysis of comments on social media posts, this research discusses the impact that this type of social media marketing has on consumers’ brand attitudes and purchase intentions.
Introduction
Entering into 2021, Athleta, a female athleisure brand, made some large promises to their customers. All of these promises focused on the phrase “Create for All, With All” 1 , meaning they are placing a high emphasis on increasing the inclusivity of their brand. There are pillars under this initiative around including more women of color, but also around including women of all sizes. Historically, the fashion and advertising industries have focused on the idea that thinness sells. 2 However, with modern research about the impacts of representation and more conversations being had about mental health, there is some discussion being had about the impacts this marketing strategy is having on the increasing prevalence of eating disorders and body dissatisfaction.
“The rate of development of new cases of eating disorders has been increasing since 1950.” 3 Eating disorders and body image issues have been plaguing society for decades. More and more contemporary research is being conducted and published to investigate the causes, influences and potential treatments for these diseases. One major theme that continually surfaces in these discussions is the role that social media is playing in this trend. Social media perpetuates an idealized narrative that so many users are exposed to and undoubtably internalize. From 2005 to 2021, the percentage of U.S. adults that reported using at least one social media site has risen from 5% to 72%. And looking closer at the demographics of the users, more females reported social media use (78%) than males (66%). 4 These rising percentages in conjunction with the already investigated connections between body image issues and social media clearly provide cause for concern. The vast array of posts that social media users are exposed to can also lead to some of these issues. People are not just seeing friends and family on social media; they are also following celebrities and influencers, being marketed to every step of the way. Social media has become an incredible tool for companies to reach their target audiences in a new format that allows more room for interaction. With this expansion, there has also been the emergence of different marketing strategies. 5
Previously, it was seen as taboo for companies to make political statements or appear to take sides on hot-topic issues. However, the interactivity of social media has proven to show more of a demand from consumers for companies to address certain things and companies are listening. Social issues are making their way into marketing materials significantly more than ever before.
One major issue that has been very prevalent lately is the unrealistic female ideals that are shown in advertisements. Several companies have come under fire for only depicting the ‘perfect woman’ in their advertisements and steering clear from featuring representative models. Since the initial backlash started, some companies have taken positive steps in trying to be more inclusive in their marketing while others seem to be holding firm on their previous modeling choices.
This research aims to explore the differences in public perception of companies that have chosen to make a change in their marketing and those that have not. Does this choice impact brand attitudes and purchasing intentions or are most consumers’ still blind to the issue?
Research Questions
Research Question 1 (RQ1):
Does social issue marketing surrounding women’s body issues on social media impact a company’s brand reputation among their customers?
Research Question 2 (RQ2):
Are purchase intention and sales impacted by how consumers view a brand’s stance on inclusive marketing for women?
Methods
In order to investigate the impacts that social issue marketing focusing on female body issues could have on a company, it is important to look at one company that is known for their inclusive marketing and another, in the same space, that is not. To further understand if the reputation of the brand among customers is impacted by these strategies, comments on Instagram posts will be analyzed for their sentiment. And to research one step further, revenue reports will be reviewed to understand if there are any implications for the sales of the company too.
Athleta, a female activewear company owned by Gap, recently announced a large expansion of sizing both and now sells sizes XXS-3X both in stores and online. Maggie Griswold, a writer for Women’s Health Magazine reacted to this announcement in a piece where she stated, “As someone whose sizes have often been relegated to a dark corner at the back of a store, I can’t overstate how much of a game-changer this is.” In addition to this announcement, they have introduced mannequins of various sizes in their stores and are vocal about being including women of all sizes in their advertisements. This brand was chosen for these reasons to represent a positive social issue marketing brand.
The next step was to investigate Athleta’s competitors in order to choose a company that could be used in contrast. To draw comparisons between the two companies it is important that they are marketing to the same demographic and sales category. Lululemon was another company that came up when researching competitors of Athleta. Further research uncovered that Lululemon often gets backlash for not being inclusive in their marketing and therefore is a good choice for a comparison in this space. Lululemon’s history of body-shaming extends back to 2013 when the founder was quoted saying that their products “don’t work” on certain body types and continues today with women reporting feeling shamed by employees in the stores. ,
The first question that this research hopes to answer is: does social issue marketing surrounding women’s body issues on social media impact a company’s brand reputation among their customers? To investigate this connection social media posts and captions from Instagram were collected and run through a sentiment analysis program using RStudio. This gives data for the average sentiment captured in the comments left on the posts. The posts were chosen from February 9, 2021, when Athleta announced their size expansion and introduced their new mannequins, by using a random number generator and selecting 24 additional posts. This process will then be replicated on Lululemon’s Instagram by selecting 25 posts from the closet dates to the Athleta posts. These posts will also be coded to help create more meaningful connections back to the impacts of inclusive imagery. The codebook focuses on the popularity of the posts and the models chosen. All coded sections are outlined in Table 1.
| Post Metrics | |
|---|---|
| Likes | How many likes the post has |
| Comments | How many comments the post has |
| Sentiment Score | Calculated through coding with RStudio (NRC sentiment) and refers to the sentiment of the comments left on the post. Possible sentiments include: positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust |
| Model/Post Characteristics | |
|---|---|
| Number of Models | How many models are in the imag |
| Body Size Score | Defined further below, a scale of 1-9 based on previous research and a Female Body Scale |
| Activity Depicted | Refers to what the model is doing in the post. The posts will fall into one of the following categories: yoga, running, other athletic activity, daily errands, or other. |
Table 1. Description of coding variables for the content analysis of posts
The body size score will be coded by comparing the models in the posts to images on a female body scale chart used in previous similar research (Figure 1). 6
The images shown in the figure are coded with the values 1-9, with 1 corresponding to the figure on the far left and 9 corresponding to the figure on the far right. When coding the posts, if there is more than one model depicted, than the average body score of all the models will be documented.
The sentiment analysis will be done by collecting the Instagram comments from the posts using Apify and coding them in RStudio. A “nrc” sentiment analysis was conducted in R which is when each word is either coded yes or no for each of the following emotions: positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. The way this code works, each word can be included in the count for different emotions. In some parts of the analysis the raw scores from each of these emotions is analyzed, however in order to create a more uniform variable, an overall sentiment score was also calculated. This was done by categorizing each emotion as either positive or negative (Table 2), then the values for those emotions was either changed to a negative value (if the emotion was considered negative) or left as a positive value (if the emotion was considered positive). These numbers were then all added together resulting in an overall sentiment score for each post.
| Emotion Category | Inclulded Emotions |
|---|---|
| Postive | Positive, Anticipation, Joy, Surprise, Trust |
| Negative | Negative, Anger, Disgust, Fear, Sadness |
Table 2. Breakdown of how emotions were coded- positive emotions resulted in positive values factored into the overall sentiment score and negative emotions results in negative values factored into the overall sentiment score.
The second question that this research is investigating is whether this brand reputation translates to purchase intention and bottom-line sales. To answer this question the sales reports from Wall Street Journal (WSJ) were analyzed. WSJ publishes quarterly reports for Lululemon 7 and Gap 8 , the parent company for Athleta. Because Athleta reports under Gap, some estimation was needed to extract their financials. In a 2020 end of year report, it was noted that Athleta consisted of 7.3% of the net sales reported by Gap. 9 This percentage was then used to estimate the sales of Athleta specifically and not of Gap as a whole.
If the initial investigation into the first research question proves that there is a difference in reputation, this will allow the research to go one step further and look into if there are financial implications of that as well. If there are any trends uncovered in the sentiment analysis, then the financial report corresponding to that time period will be analyzed more closely. It is important to note that this will not prove causation but can give support for correlation of this question.
Findings
The findings of this paper build upon each other and will be addressed in order within this section. The first area analyzed was representation of body size in the posts of the two companies. Next, a sentiment analysis was conducted on the comments on the posts. This was done in two steps, the first broadly characterized the comments and gave them a score and the second looked closer at the emotions assigned to the comments and drew conclusions based on that. These steps also connect back to the post representation and discuss how that could impact the comment sentiment. Finally, to answer the second research question there was a brief connection to the financial statements of the two companies.
Post Representation
As excepted, this research provided support that Athleta’s marketing style was more inclusive of different body types than Lululemon’s (Table 3). However it is important to note that the difference between the two was only .30 which suggests that Lululemon is more inclusive than previous research suggested.
| Company | Range | Mean |
|---|---|---|
| Athleta | 2-8 | 3.57 |
| Lululemon | 2-7 | 3.27 |
Table 3: Range and mean Body Size Score for Athleta and Lululemon
Out of the 25 posts analyzed for Athleta, six of them only included pictures of the clothing and didn’t have models featured. This left 19 posts to be coded for body size and of those 19, there were 4 that depicted models larger than 4.5 (which is the midpoint of the scale used). This means that 21% of the posts showed women who placed higher than average on the Female Body Scale (see figure 1).
When looking at the Lululemon posts, four of them didn’t include models and five of them only had male models. Since this research is focusing on women’s body issues the posts that only had males were not analyzed. This left 16 posts to be included in the body representation scores shown above. Out of those 16, there were two posts that showed models of a 4.5 or larger. This means that 12.5% of the posts showed women that can be considered to have larger than average body types.
Another important part to consider in this discussion is the number of posts used by both companies that did not feature a model. Athleta posted images of just merchandise in 24% of the images analyzed and Lululemon posted images without models in 16% of the images. The sentiment on these images will be analyzed as well but it is an interesting observation.
Overall Comment Sentiment
The main goal of this research was to investigate if the consumer’s attitude and sentiment towards the companies shifted based on the representation in their posts. Upon initial analysis there appears to be a fairly strong connection between average body score and the sentiment score of the comments (figure 2 and 3).
Figure 2: Sentiment score and Average Body Score for Lululemon posts
Figure 3: Sentiment score and Average Body Score for Athleta posts
To look a little closer, the average overall sentiment score of the posts that depicted models with an average body score of 4.5 or higher was compared to the average overall sentiment score of the posts where the models had an average body score of less than 4.5. Again, 4.5 was chosen as it is the midpoint on the scale used (see methods for further explanation). For both companies it was shown that posts with models of a 4.5 or higher body score received comments with a higher sentiment score, meaning the comments were more positive (Table 4).
| Company | Average Sentiment 4.5+ | Average Sentiment Under 4.5 |
|---|---|---|
| Athleta | 27.25 | 23.36 |
| Lululemon | 35 | 7.56 |
Table 4: Average sentiment score of comments for posts that scored 4.5+ or under 4.5 on the average body score scale.
Sentiment Broken Down
The sentiment analysis (nrc) conducted allows for understanding past simply positive or negative comments and gives information about the actual emotion conveyed by the words (see methods for further explanation). For this section of the analysis, the same group as above was used with the addition of a “no model” group. A count was then run on how many words in the captions related to each emotion (figures 4 and 5).
Figure 4: Sentiment score by emotion broken up by average body score (posts with an average body score of 4.5 or higher, posts with an average body score lower than 4.5, and posts with no models) for Lululemon posts
Figure 5: Sentiment score by emotion broken up by average body score (posts with an average body score of 4.5 or higher, posts with an average body score lower than 4.5, and posts with no models) for Athleta posts
Athleta
For Athleta, the under 4.5 group was lowest preforming in all emotion categories. In every category posts with models under a body score of 4.5 had the least amount of sentiment associated with the comments. While this lack of variation makes it more difficult to obtain meaningful information from this analysis there are still some key takeaways that can be understood from these data. First, trust is most prevalent in posts that did not depict any models. This could be a testament to marketing that is less model based, and more product based. Secondly, joy is largely the most common emotion identified across all model groups. This shows that overall, the sentiment of the comments left on Athleta’s posts reflect joy on the part of the consumer.
Lululemon
For Lululemon, the findings were very similar to that of Athleta. In this case as well the under 4.5 body score group had the lowest count across all emotions but there are other insights we can gain. There was not a clear highest performing with Lululemon as there was Athleta however three of the positive emotions (anticipation, joy, and trust) all performed comparably. It is also interesting to note that the “no model” category outperformed the 4.5+ group in both anticipation and trust and was a close second in joy. A final key observation is that in anger, the 4.5+ and the under 4.5 group had the same count.
Comparisons Across Companies
When considering how the sentiment on Lululemon comments compares to the sentiment of Athleta comments there are a few key things to note. First, the sentiment on Lululemon’s posts is much more evenly spread than that of Athleta’s. Joy is very clearly a top emotion identified in Athleta’s comments whereas the emotions identified in Lululemon’s comments are more uniformly spread. Additionally, for both companies trust was most highly associated with posts that did not include female models. A final noteworthy comparison is that when looking solely at the positive emotions (anticipation, joy, surprise, and trust) the 4.5+ group performs better on more of Athleta’s posts than it does Lululemon’s posts (Table 5). This shows evidence that when models that are considered above average body size are depicted in posts, Athleta’s posts get more of a positive response than Lululemon’s.
| Company | Anticipation | Joy | Surprise | Trust |
|---|---|---|---|---|
| Athleta | No Model | 4.5+ | 4.5+ | No Model |
| Lululemon | No Model | 4.5+ | No Model | No Model |
Table 4: Table 5: Highest scoring group for each of the positive coded sentiments
Summarizing Representation and Sentiment
The initial part of this research was interested in seeing if there was a notable difference in the sentiment of comments on advertising posts depending on the size of the model used. The following section then aims to take that one step further and identify if this representation and consumer sentiment also translates into sales or financial penalties. However, before addressing that it is important to summarize the key takeaways from above.
First, it was identified that the average model size in Athleta posts was larger than that of Lululemon posts. Then when looking at the overall sentiment of the comments left on these posts, it was found that when women above the median body size (4.5) were depicted, the sentiment score of the comments was higher, meaning the comments were more positive. Lastly, the sentiment was broken down further into specific emotions. During this investigation it was found that the comments on Lululemon’s posts were more evenly spread across all emotions whereas joy was clearly the most commonly identified sentiment on Athleta posts. Another interesting discovery of this step was that trust was more often reflected in the comments when no female models were used.
In summary, Athleta’s posts include a larger range of body sizes and when women with a body score of 4.5 or higher are shown in advertisements, they tend to receive comments with more positive sentiment.
Financial Connection
The final step of this research was to see if there were any financial implications for the companies involved. In order to do this the financial reports were pulled for both companies (see methods section for details). These financials (by quarter) were then compared to both the sentiment score of the comments and the average model size of the posts from each quarter (Figure 6 and 7).
Figure 6: Average body size score and sentiment by quarter and sales for Athleta. Highest in each column is highlighted
Figure 7: Average body size score and sentiment by quarter and sales for Lululemon. Highest in each column is highlighted
The results from this investigation are inconclusive as they do not provide evidence for a clear connection to be drawn in either way. For Athleta, the quarter that had the comments with the highest average sentiment score also had the highest sales numbers. However, for Lululemon the quarter with the highest sales was not the quarter with the highest average sentiment. This fails to show a connection between the sentiment of comments and actual income.
Discussion
This paper aims to address two important research questions through an investigation of Instagram posts and captions as well as connections back to reported financials of the two companies.
It was found that there seems to be a positive correlation between posting images that include women with a larger body type. This conclusion held true for both Athleta and Lululemon and is based off of a calculated average sentiment score of posts that include women above and below average. In the case of this research, average body size is defined by the Body Representation Chart which ranges from 1-9, making a score of 4.5 average.
The second research question dealt with the financial implications of representation and comment sentiment. There was no clear connection found between the sentiment or body size of the models and the sales for each quarter that data was collected for. This suggests that while there may be a connection between attitudes of consumers, there is no evidence that these attitudes translate into purchase intention or losses for the companies.
Conclusion
While this research is a good start to understanding these connections, there are some key limitations that inhibit the findings.
Limitations
There are two main areas where limitations are present in this study. The first has to do with the collection of the comments. Instagram comments are not easily accessed unless you are the owner of the account. An outside tool was used to scrape the comments however it was unable to collect all the comments from each post. There were about 15-30 random comments per posts that were accessible which is about 10% of the total comments per post. Since multiple posts were looked at for each company, there was still plenty of data used in this study, but not all.
The second area where limitations were present was in the financial component. First, the time period from which data was collected only included quarters 1-3 of 2021. The posts were specifically chosen to begin when Athleta announced their size expansion, however it would be helpful to continue collecting data over several years to allow for any trends to appear and be analyzed. Additionally, another limitation in this section was that Athleta is part of a parent company, Gap, where Lululemon is its own entity. This affected the accessibility of their financial records. Gap reports as a whole so some calculations were made to estimate Athleta’s sales. These calculations were based on 2020 reports but were still only estimates.
In Summary
As the impact of body representation becomes increasingly studied and understood the importance of this type of research will also increase. It is crucial that companies understand the role they play in how girls and women see themselves. The unfortunate reality is that businesses are profit driven and therefore unless there are financial implications to their choices, they are unlikely to change.
Footnotes
1 “Athleta,” accessed November 8, 2021, https://athleta.gap.com/browse/info.do?utm_campaign=likeshopme&cid=1160634&utm_content=www.instagram.com%2Fp%2FCRRhh93lty4%2F&mlink=1157681%2C19225547%2CCommunityHub_Action_0922_BLM&utm_source=dash%20hudson&utm_medium=instagram&tid=atsm000160&clink=19225547.
2 Emma Halliwell and Helga Dittmar, “Does Size Matter? The Impact of Model’s Body Size on Women’s Body-Focused Anxiety and Advertising Effectiveness,” Journal of Social and Clinical Psychology 23, no. 1 (February 1, 2004): 104–22, https://doi.org/10.1521/jscp.23.1.104.26989.
3 “Get the Facts | National Organization for Women,” accessed October 10, 2021, https://now.org/now-foundation/love-your-body/love-your-body-whats-it-all-about/get-the-facts/.
4 1615 L. St NW, Suite 800 Washington, and DC 20036 USA202-419-4300 | Main202-857-8562 | Fax202-419-4372 | Media Inquiries, “Demographics of Social Media Users and Adoption in the United States,” Pew Research Center: Internet, Science & Tech (blog), accessed October 10, 2021, https://www.pewresearch.org/internet/fact-sheet/social-media/.
5 Efthymios Constantinides, “Foundations of Social Media Marketing,” Procedia - Social and Behavioral Sciences, 2nd International Conference on Strategic Innovative Marketing, 148 (August 25, 2014): 40–57, https://doi.org/10.1016/j.sbspro.2014.07.016.
6 Maggie Griswold, “Athleta’s Latest Launch Is The Inclusivity Push In The Fitness Industry I’ve Been Waiting For,” Women’s Health, March 26, 2021, https://www.womenshealthmag.com/life/a35744863/athleta-inclusive-sizes-launch/.
7 “Athleta’s Competitors, Revenue, Number of Employees, Funding, Acquisitions & News - Owler Company Profile,” Owler, accessed October 27, 2021, https://www.owler.com/company/athleta.
8 “Lululemon Founder To Women: Your Thighs Are Too Fat,” Throughline Group, November 12, 2013, https://www.throughlinegroup.com/2013/11/12/lululemon-founder-to-women-your-thighs-are-too-fat/.
9 Julie Mazziotta Updated December 08 and 2020 11:11 Am, “Woman Who Lost 80 Lbs. Body Shamed By Lululemon Employees: ‘I Felt Less Than,’” PEOPLE.com, accessed October 27, 2021, https://people.com/health/woman-lost-80-lbs-body-shamed-by-lululemon-employees/.