Media Effects Research Lab - Research Archive
Exploring the Effect of Realism and Transformation of Augmented Reality (AR) Face Filters on Social Media
Student Researcher(s)
Maranda Berndt (Ph.D Candidate);
Qing Xu (Masters Candidate);
Hannah Smith (Ph.D Candidate);
Faculty Supervisor
(This paper was based on a project as part of the “Psychological Aspects of Communication Technology” course.)
INTRODUCTION
Augmented Reality (AR) face filters have gotten more popular on social media in the past decade, with about 600 million people using AR filters each month of Instagram and Facebook (Bhatt, 2020). Relying on augmented reality, a variety of face filters offer users 3D experience via incorporating digital elements in the real-time physical world (Keeports, 2022). Previous research has looked into motivation for an individual using AR face filters on Instagram in the realm of the version of themselves they wish to present (Javornik et al., 2022). However, there has been little to no research done on which social media platforms people choose for specific AR face filters and the motivation behind using AR filters.
AR filters can dramatically change an individual’s facial features (Appeal et al., 2020). Along with that, a wide variety of filters available mean there is a number of different changes that can be made to a person’s features (e.g., different eyebrow shapes, facial contour, dog ears, large eyes, etc.). Self-presentation theory discusses how individuals’ attempt to control and shape others’ impressions through manipulating their attributes, including setting, appearance, and behavior (Goffman, 1959). AR face filters is a way to control how individuals are viewed online. Therefore, the goal of this paper is to look at the relationship between the levels of realism and transformation in AR face filters and how the motivation behind using face filters could moderate and individual’s enjoyment of using the filter, their attitudes towards filters, and which social media platform they would choose to use filters on.
RESEARCH QUESTION/HYPOTHESES
H1: AR filters high in realism will increase (a) users’ likelihood of using Instagram filters, (b) increase negative attitudes towards AR filters, and (c) increase attitudes towards the use of the specific AR filter provided.
H2: Low transformation conditions will increase the likelihood of using filters on Facebook.
RQ1: Does transformation influence specific AR face filter attitudes?
H3: AR filters with high realism and low transformation will increase (a) the likelihood of using Twitter, (b) the positive attitudes towards AR filters, and (c) increase specific attitudes towards the AR filter provided.
H4: The interaction between low realism and high transformation will (a) increase the likelihood of using Snapchat, (b) decrease the likelihood of using Twitter, (c) increase positive attitudes towards using AR filters, and (d) decrease negative attitudes towards using AR filters.
H5: The interaction between low realism and low transformation will decrease the attitude of anxiety/depression towards using AR filters.
H6: The relationships in H1 will be moderated by (a) popularity and (b) appearance motivations.
H7: AR filters with high realism and high transformation will increase (a) the likelihood of using Instagram and TikTok, (b) increase the negative and dependent attitudes towards using AR face filters and (c) increase the attitudes AR filter attitudes when moderated by popularity and appearance motivations.
H8: AR filters with high realism and low transformation will increase (a) the positive attitudes towards AR filters, and (b) attitudes towards the specific AR filter provided when moderated by appearance motivation.
H9: The interaction between low realism and high transformation will (a) increase likelihood of using Snapchat, (b) decrease the likelihood of using Twitter, (c) increase positive attitudes and (d) decrease negative attitudes towards AR filters when moderated by connection motivation.
H10: The interaction between low realism and high transformation will decrease the use of Instagram and TikTok when moderated by appearance motivation.
H11: The interaction between low realism and low transformation will be moderated by connection motivation.
H12: The direct effects of transformation and realism on (a) platform preference, (b) overall social media preference, (c) attitudes towards AR face filters, and (d) attitudes towards use of the specific AR face filter provided will be mediated by enjoyment.
METHOD
After receiving IRB review approval, participants were recruited through social media platforms: Reddit, Facebook, and Twitter, and MTurk and completed a brief online survey distributed via Qualtrics after being randomly assigned to a condition. This study utilized a 2 (transformation: high vs. low) X 2 (realism: high vs. low) experimental factorial design. Participants were exposed to two different pictures simultaneously. One of these was an image of an individual (23-year-old woman) without a filter. This image was shown to all participants who took the survey. The other image was the same individual, pose and facial expression but with an AR face filter applied. The image was shown as dependent on the condition that the individual was allocated to: high realism/high transformation, high realism/low transformation, low realism/ high transformation, low realism/low transformation. The online survey included basic demographic questions, and measured participant’s enjoyment of the AR filter, their attitudes towards AR filters in general and the specific AR filter they were shown, and the social media platform they would use that particular AR filter on. The final sample (N = 239) was made up of 53.14% female participants, the average age was 34.13 and 83.26% of the participants were white.
RESULTS
We found support for partial support for hypotheses 3C, 5, and 6C, and support for 7B. We found opposing support for 4B. Hypothesis three was measured with a factorial ANOVA; the main effects for transformation were not statistically significant. However, we can see a trend in the interaction plot suggesting that for lower levels of realism and higher levels of transformation there are increased opinions of the specific AR filter used, which provides partial support for our hypothesis. Partial support was found for H5, which we explored using a Factorial ANOVA. While the main effects for transformation, realism and the interaction term were insignificant, the plot suggests that for lower levels of realism and low levels of transformation there are decreased anxiety scores. We used Hayes Process Macro for hypotheses post 6 and found partial support for H6C, which stated that AR filters high in realism will increase attitudes towards the use of the specific AR filter type provided when moderated by appearance. While the interaction term was not statistically significant when we look at the interaction plot we can see that the slopes appear to be different with the low appearance line being steepest and those with high appearance having the steepest line. Hypothesis seven was also studied using Hayes Process Macro, specifically using two-step moderation models. The dual-interaction term in both appearance and popularity models was statistically significant. This therefore supports our hypothesis as high realism and high transformation increases anxious attitudes towards using AR filters and this is moderated by (a) popularity motivations and (b) appearance motivations as both heighten the difference between low and high realism.
We explored hypothesis four using a Factorial ANOVA; the main effect for transformation and realism were not significant for H4b, along with the interaction. However, upon further examination we do see an interesting trend in the interaction plot; namely, the plot appears to suggest that for lower levels of realism and high levels of transformation, there are increased intentions to use Twitter. This is the opposite of our hypothesis.
CONCLUSION/DISCUSSION
Theoretical contributions: Potential interesting contributions to the mental health and wellbeing literature as we found empirical evidence for certain types of filters affecting anxiety and negative attitudes towards AR filters more than others, and we demonstrate how motivations behind using filters can influence anxiety too. Additionally, we contribute to social media communications literature by providing initial empirical support for different motivations influencing opinions of filters dependent on how realistic they are.
Practical Contributions: We found some practical implications for Twitter and the stories platform; the site may want to embrace low realism and high transformation filters. The mental health and well-being related findings may be useful in policy making. Our findings may also have interesting implications for marketing practitioners when using techniques such as influencer or social media marketing - brands must consider how they wish to make their customers feel when encouraging them to try their filters.
Limitations: Our participant recruitment sources had a significant effect on our findings; we discuss two possible explanations for this: MTurkers make up a different demographic or the data from MTurk was of a low quality. Our model may have also been overly complicated; in trying to reduce error and account for numerous possible moderators and mediators we produced a model with a lot of moving parts. Additionally, this study focused on static face filters, while real-time video filters are a recent advance; TikTok is an excellent example worth examining. This study did not separate gender to see the differences between each gender. It would be interesting for future studies to separate gender to see if there are differences in each gender.
References
Bhatt, S. (2020). The big picture in the entire AR-filter craze. Retrieved from https://e
conomictimes.indiatimes.com/internet/brands-see-the-big-picture-in-ar-filter-craze
/articleshow/78266655.cms?from=mdr.
Keeports, A. (2022, April 7). What’s Behind Augmented Reality Face Filters? Quantilus. https://quantilus.com/whats-behind-augmented-reality-face-filters/
Javornik, A., Marder, B., Barhorst, J. B., McLean, G., Rogers, Y., Marshall, P. & Warlop, L. (2022). “What lies behind the filter?” Uncovering the motivations for using augmented reality (AR) face filters on social media and their effect on well-being. Computers in Human Behavior, 128, 1-15. https://doi.org/10.1016/j.chb....
Goffman, E. (1959). The presentation of self in everyday life. Anchor Books.
For more details regarding the study contact
Dr. S. Shyam Sundar by e-mail at sss12@psu.edu or by telephone at (814) 865-2173