Gaming the Studies – Problematic Pictures
This post is part of a series on the research studies on sexism cited in the FeministFrequency video series Tropes vs Women in Video Games. The first post in that series covered my views on the FeministFrequency videos and can be found here.
This post in the series examines the two studies in the “Objectification Leads to Depersonalization” article by Loughnan et al (2010).1 I cover some relevant background material in two posts here and here.
Before getting into the details of the experiments, it’s probably worth making a few comments on the approach I’ll be taking. The primary purpose of these reviews isn’t to establish the academic merit of the studies or consider if they offer any novel insights into the human mind. The purpose is to consider whether the studies can be reasonably used to support the claims being made by FeministFrequency and by implication whether they can be reasonably used to justify calls for change within the video game industry.
I’ll give brief overview and cover a summary of the method and results of each experiment, focusing on details I plan to comment on. I’ll also include the most significant observations and comments made in the article.
After that I’ll cover my own views and consider the article against three different criteria:
- Relevance: How relevant is the study to gaming? i.e. How well does the experiment replicates circumstances related to video games in the real world?
- Significance: How significant is the study to the gaming community? i.e. Do the results indicate there is a potential problem or is this a case of streetlight psychology?
- Confidence: To what extent can we be confident the experiment actually supports the theories put forward by the authors? i.e. Are the any flaws in the experiment or any alternative theories that might explain the results?
Objectification Leads to Depersonalization:
The denial of mind and moral concern to objectified others
Loughnan, Haslam, Murnane, Vaes, Reynolds & Suitner (2010)1
I wasn’t exactly sure what sort of studies I would find when I started this review project. I have a reasonably healthy respect for the peer-review process, but I had concerns as I had heard many disparaging remarks about ideological bias present in the field of social psychology. Those concerns weren’t exactly alleviated by the first words of the first article:
“When objectification occurs, the person is depersonalized”
— Andrea Dworkin
The two experiments are about the way the objectified images of people influence the perceptions of mental attributes and moral status (“personhood”). The first experiment attempts to extend the work of Schwarz & Kurz (1989)2 on face-ism to cover photos where the head is excluded. The second looks at the impact on the mental and moral perceptions of people either fully clothed or wearing revealing clothing.
The studies are intended to look at the effects of objectification in media where research has shown different genders are presented with different amounts of face and body.3 The article doesn’t mention the priming effect however the experiments do involve manipulating a first stimulus (photo) and then measuring the response to a second stimulus (questionnaire). It does cover the harmful effects of self-objectification and cites studies they may support some of the other claims my in the FeministFrequency video.4 It also notes some research that identifies how people view moral agency and moral patiency as different aspects of personhood.5
This study used photos of three women (“targets”) in swimsuits from the internet, each modified into three different forms: head-only, body-only or full body. The subjects were 86 college aged people (54 f, 32 m; mean age 20.5). Subjects viewed a head-only, body-only and full-body photo with each one photo being of a different target. They rated the targets on a four of different scales. The ratings are all done scales (e.g. 1 = “not at all” to 7 = “very much so”) but I’ll be normalising results to percentage changes when discussing them.
What a full-body photo of a women in swimsuit
may look like (from the internet not the study).
The Mental State Attribution (MSA) scale consists of asking subjects about how much the targets possessed the ability to do 20 different things such as see, hear, feel fear, feel joy, reason, think, wish, plan, etc. The General Mind Attribution (GMA) scales directly asks “how much mind does this woman have?”. The blatant moral scale involves asking general questions about the target such as “how unpleasant would it be to harm this woman?”. The Experience Scale (ES) asks about the capacity of the person to have hunger, fear, plain, rage, desire, personality, consciousness, pride, embarrassment and joy.
Compared to the head-only photos, the full-body photos resulted in a marginal decrease (~3%) in the ratings of the MSA and ES (the more complex scales). Compared to the head-only photos the body-only photos resulted in a small decrease (~8% – ~15%) on the mental scales and a marginal decrease (~4% – ~6%) on the moral scales.
Discussion in the Article:
I think the article’s discussion of these results can be summed up by this quote:
“The results of Study 1 indicate that participants deny objectified women both proposed aspects of personhood: Mind and moral status.”
While a small decrease in perceptions of mind and moral status are not insignificant, the language appears to be rather indulgent. I’m not sure a shift in viewing a particular mental capacity from “moderately” to “somewhat” can be reasonably described as “denying” that capacity. The discussion does make note of the interesting lack of difference on the blatant scales (GMA and blatant moral scale).
This study used images of 2 men and 2 women from the internet. There were 2 images of each target: one “objectified” image where they were wearing a bikini (women) or shirtless (men) and one “non-objectified” image where they were fully clothed. The photos were controlled to ensure they all contained equal proportion of faces to remove any face-ism effect. They were also pretested to ensure there was no difference in attractiveness or emotionality between the objectified and non-objectified images.
The subjects were 80 college ages people (40 f, 40 m; mean age 19.2). They rated each target on the MSA, GSA, ES and moral patiency scale (PS). They also estimated IQ and rated competence at 4 jobs: lawyer, manager, stockbroker, scientist. In a final measure they were asked how many pain-inducing tablets they would give the target to induce a specific amount of pain.
The mental scales (MSA, GMA, IQ & Competence) were all lower for objectified images (~6% – ~21%). The gender of the observer (subject) had only one significant effect. Women rated objectified men lower on the MSA compared to non-objectified men, but did not rate objectified women lower compared to non-objectified women.
The gender of the target made a significant difference on the moral scales. On the ES and PS only men were rated lower when they were objectified (~6% – ~9%). On these two scales women were not rated lower when they were objectified. There objectified targets given slightly more pain-tablets on average (~12%) with no gender effect present.
Discussion in the Article:
The discussion covers these results reasonably well, albeit with the same use of the word deny:
“The main effect of objectification emerged for all comparisons with objectified targets denied both mind and moral status, although some effects were qualified by interactions. Interestingly, participant gender appeared to have little effect on ratings. This perhaps suggests that both men and women depersonalize the objectified. … This result surprisingly suggests that objectification may have an equal or even greater effect on the perception of males as it does on the perception of females.”
These studies are not particularly relevant to video games. There is no indication if video game images cause objectification of video game characters that the objectification caries over into real life (that is dealt with in later studies). However, it is perhaps interesting from consideration of character or avatar design. Developers could leverage the face-ism and revealing clothing elements of their appearance to bias player perceptions in both normative and subversive ways. It is also perhaps of interest in players’ selection or modification of avatars in multiplayer games and the degree to which it may bias the way other players perceive them.
Given the limited relevance to video games it’s not that clear how significant this study really is. To the extent that it is significant, it does highlight that objectification of men is at least as significant as the objectification of women. Here it is important to consider the difference between video games and other popular media in terms of the extent of gender representation. It seems uncontested that men are far more frequently depicted as characters in video games. That means that many of those men, from the shirtless barbarian wielding a sword to the shirtless slave building a pyramid, are going to be the most significant examples of objectification in the medium.
Video games – Objectifying men since at least 1987.
There are a number of issues with both of the studies. In both studies it is not clear that the level of face-ism or clothing is the only or even the dominate factor in influencing outcomes. I discussed a number of studies in my second background post that identified many elements of appearance that can influence perceptions of people. These include appearance of health, facial expression, stance, make-up and clothing brand. While there was an effort in the second study to ensure no over-all difference in emotion or attraction, there was not systematic effort to ensure all other factors were eliminated. The low number of photos, less than a dozen over both studies, doesn’t given confidence that any variations would balance each other out (compare the Naumann et al study which used over 200 photos).6 Clothing is of particular concern with the competence measure as all the jobs are white-collar and would have clothing expectations beyond simply wearing some.
Which non-objectified Brad Pitt is the most intelligent and competent?
I found the use of the pain measure in the second study is a little confusing. The study cited as a source of the measure uses it as a comparison between two targets.7 That is the subject is told there a 5 tablets that must be distributed between 2 targets. This would necessitate the mean result overall is 2.5 tablets per target. However, the results from the second study have both non-objectified targets (2.19 tablets) and the objectified targets (2.47 tablets) are below this value. The wording in the article is “they must decide how many tablets to give to the person in the image to induce a specific amount of pain”, which suggests that it is a measure of how much the target can tolerate pain rather than how much they do or don’t deserve pain relative to another.
Which objectified Brad Pitt has the most moral status or is most able to endure pain?
While I clearly chose the pictures above to best illustrate my point, I doubt the researchers would have selected pictures as obviously biased. However, as we saw in the Doyen et al (2012)8 study in my first background post it can be quite possible for even diligent experimenters to allow bias to creep into a study. The esteem these researchers appear to hold for Andrea Dworkin and her theories does not give rise to confidence in the neutrality of the sample pictures. This leaves the statistical confidence in the results as simply reflecting a confidence that if the same photos (or perhaps photos selected by the same researchers) were shown to a similar sample of college aged students that the rating results would be match. It does not provide confidence that the result would apply across “objectified” generally.
Overall I do not have much confidence that the results seen in these studies reflect a causal relationship with objectification. That said, it is important to note this conclusion does not imply that the objectification theories are wrong, just merely that they are not well supported by these particular two studies.
It is probably worth highlighting the actual conclusions of the study to anyone using it in support of a claim that objectification of women in video games is problematic while also claiming the objectification of men is not problematic. It might also be tempting to cite this study in support of claims that objectification is a greater problem in video games for men than it is for women. However, the lack of confidence I have in the design and execution of the studies means I will not be doing so.
- Steve Loughnan, Nick Haslam, Tess Murnane, Jerosen Vaes, Catherine Reynolds & Caterina Suitner (2010). Objectification leads to depersonalization: The denial of mind and moral concern to objectified others. Journal of Experimental Social Psychology, 40(5), 709–717.
- Norbert Schwarz & Eva Kurz (1989). What’s in a picture? The impact of face-ism on trait attribution. European Journal of Social Psychology, 19, 311-316.
- Dane Archer, Bonita Iritani, Debra Kimes, & Michael Barrios. (1983). Face-ism: Five studies of sex difference in facial prominence. Journal of Personality and Social Psychology, 45, 725–735.
- I’m not planning on covering those studies as part of this series, but may investigate it later.
- Heather Gray, Kurt Gray & Daniel Wegner (2007). Dimensions of Mind Perception. Science, 315, 619.
- Laura Naumann, Simine Vazire, Peter Rentfrow & Samuel Gosling (2009). Personality Judgments Based on Physical Appearance. Personality and Social Psychology Bulletin, 35, 1661-1671.
- Kurt Gray & Daniel Wegner (2009). Moral Typecasting: Divergent Perceptions of Moral Agents and Moral Patients. Journal of Personality and Social Psychology, 96(3), 505-520.
- Stephane Doyen, Olivier Klein, Cora-Lise Pichon & Axel Cleeremans (2012). Behavioral Priming: It’s all in the Mind, but Whose Mind? PLoS ONE, 7(1): e29081.