Avatar Skin Tone Experiment Reveals Strong User Preference for Whiteness

In virtual worlds and MMOs where most aspects of avatar appearance can be instantly changed by a few button clicks -- including gender and skin tone -- do we leave our real world attributes and preferences behind? After noting that avatars with lighter skin seem to get more likes than darker ones on Flickr, I enlisted the help of Mandy Smith, a popular SL fashion photographer on Flickr, to take two photos of the same avatar in the same pose in the same location, with the only difference being her skin tone:

She posted these to her Flickr feed without additional commentary, separated by the span of a couple weeks or so, and counted the total views and Likes by her followers.

Were the totals each significantly different? Well, yes, I said they were in title of this post, but here's the actual numbers:

The light-skinned avatar got roughly 2.5 times more views and 2.5 times more Likes. So yes, a strong preference for whiter skin, even when all else is equal. Now in this particular case, Mandy Smith's default avatar is white, so I suppose some could argue her fans were really responding more favorably to the avatar appearance they associate with her more. Even then, it's striking how strong the reaction differs.

If you have an avatar-based Flickr account, I encourage you to try this out for yourself, and report results back here. (Or get in touch with me on Twitter.) People often say "I don't see race", but experiments like this may suggest otherwise.

Comments

[Blended In]

There is simply less contrast in the photograph on the right. Try it again against a more neutral background.

[Clara Seller]

I would have to agree with Blended In. It all depends on which one you are setting up the photo for. If I changed the background contrast, softened the clothing, and upped the dramatic lighting, I could accentuate the detail of darker skin and make the blond look like a marshmallow.

[YsabelleStewart]

While I suspect there IS a racial bias, and it's depressing to think there is, nevertheless photographing darker skins means choosing different lighting, settings and backgrounds. I'd say the photographs on the right are defined by settings for a white person, and unsurprisingly it's not working out right. Maybe one of the things that's necessary for decent SL photography is more information, articles and tutorials on photographing people of colour properly. I think this knowledge is generally lacking not just in virtual world photography, but also in real world photography. There are articles and videos on this subject online.

[Jay Frost]

This was just a social experiment. In social experiments they always use the same parameters on the test to get results. If you changed one parameter in the experiment you would get skewed results. Which would ultimately ruin the experiment.

[Ciaran Laval]

There have long been discussions in Second Life about the difficulty of making darker skins that look good, Iris Ophelia covered some of the issues here back in 2011 :

http://nwn.blogs.com/nwn/2011/06/new-second-life-skin-gets-it-right.html

This would influence how an image on Flickr was received.

Strawberry Singh might be a good person to talk to about skin tones, her Flickr stream has many photos with differing skin tones.

[Tizzy Canucci]

I'm not sure how to respond to this without sounding hypercritical. I think reasoned and accurate claims are what's needed at the moment, and these aren't. It's not an experiment, for many reasons which I would be happy to email to you. In my opinion, looking at what's in the soft porn Flickr groups would give a more accurate view of what people like than a pseudo-experiment like this - but even then, liking an image is different from discrimination against real people.

[Ilsa Hesse-Heron]

Could part of it be that skin designers have more choices of "white"?

[Blended In]

(RE: In social experiments they always use the same parameters on the test to get results.)

Correct; use a more neutral background for BOTH pictures and repeat the experiment.

[Chic Aeon]

You tried this same argument before with marginal dark photos and better light skinned photos. This was a slightly more fair test but as your respondents so far have told you, you simply can't light the two shades of skin (or anything really) with the SAME lighting.

Each object you choose to photograph needs to be adjusted for contrast. ** if you would have taking a good photo of the dark skin using correct lighting and then taken the same photo for the light skin, it would have been washed out and would have lost it contrast.

A simple facelight would have made a huge difference in the second photo.
And as someone mentioned the quality of the skin matters also. There needs to be more contrast in the darker skin --- which is more difficult.

There are some gorgeous dark skinned gals out there that get rave reviews on Flickr. Why not have THEM do a test with a white skin and see what happens. Again though, setting lighting for one and using the same lighting will not be fair unless you "split the difference" and light both of them in a less than optimal manner.

I personally think your premise is very flawed anyway and that higher Flickr rates don't equate at all to prejudice.

Really.

[JKRaven]

The major problem in the SL virtual reality isn't so much a preference for "Whiteness" but the fact that there are so few darker skin tones that look "realistic" or natural. They're usually flat in tone. That said, the avatar shapes also don't lend themselves to ethnicity either. They're designed to be thin, tall and mostly European in form. There are also very few hair creators that make ethnic hair very well.
So, to get an amazing looking avi, many fall back on an average "white" European look. Why? Because it's the best way to guarantee a more aesthetically appealing look.
If creators, designers and even LL's themselves would be more willing and able to produce more ethnic shapes, skins, hair....that didn't look so flat and lifeless, I am willing to bet there would be a market for it.

[JKRaven]

Chic Aeon also has a great point as well. Lighting is everything when it comes to SL photography.
There aren't as many extremely talented artists like Skip Staheli when it comes to Photoshop painting so many of us have to rely on in world windlight settings. In that you can't light a fair skinned avatar the same way you would a darker toned avatar and vice versa.

[Annie Brightstar]

I admit I mostly take landscape pictures, but I do sometimes step in front of the camera and this article for me is a non-story, with the lack of contrast the only surprise is that it picked up 85 likes, If Mandy Smith had posted the identical picture but used another flickr account I doubt the picture would got many views at all.

[Clara Seller]

In the end, the winner of the most popular looks in SL is a template for "sheep".

When going to an inworld event, most of the avatars are just background noise to me like monotonous copies of copies. Same goes for bloggers. There's a real opportunity here to make something unique and beautiful if you're the kind of person who likes to write their own rules and find their own solutions.

[Avatar Designer]

As a creator, I can tell you creating ethnic skins and products is a waste of time and money-THEY SIMPLY DO NOT SELL. I've tried to make diverse products but my bottom line is important to me, time vs. cost to make and also profit. I can't ass away tons of time on a product I know is not going to sell. Sure it makes me feel good to offer diverse choices but I have a mortgage to pay, bills , and food to buy. I make what I know sells and it's not ethnic products. That's just how it is in Second Life.

[LagOh]

Where da White women at?

[Lorraine]

Time of day, and day of week of posting on Flickr also influence views, because more users are on at certain times of day than at other times. I have had drastically different amounts of views depending on these factors. Were they equal in these pictures? Just curious.

[Zoey Neville]

not only that but all my experiements on FLickr have resulted in if you post 2 pics at the same time the one highest on the feed with get the most likes and views , im running my own experiment on this and will be blogging about it soon , with new standards for the testing after reviewing all the flaws within this experiment and testing them myself.

[Blue Dot]

I know this article is from a while ago, but in case someone still happens to land here:
1708 views and 211 faves = 12.4%
685 views and 84 faves = 12.3%
I expected a difference of at least a few percentage points, instead the views of the two pictures generated the same percentage of faves. And by same I mean essentially identical. The only substantial difference is the number of views, but as Hamlet Au wrote, those pictures were posted 2 weeks apart, thus the first one had more time to collect views, unless she counted the views after the same amount of time. I don't mean there isn't racism, but to be fair there are these couple of points to take into account, besides what the other readers said.