What Do the Average Numbers of a Second Life Female Avatar Say About What We Want in Our Avatars?

Last Friday, I mentioned the meme of Second Life fashionistas publishing their customized avatar numbers. Second Life blogger Vaki Zenovka gathered together the numbers of 220 female avatars, averaged them out, and used the results to create this female avatar at left. (Click here to see the full version.) She also did a male version, but I cropped that dude out, because as Vaki notes, she only had 32 males to work with, and so that resulting average was less, well, average.

In any case, this brings up an interesting question: Ms. Zenovka's avatar has curvier hips than are typical with real world models, and is decidedly less busty than many female avatars you are likely to see in Second Life or in other virtual worlds. So what does this average say about what we want in our avatars? In my interpretation, what we're seeing is a desire to have an avatar who is fit and sexy, but closer in proportion to the real life women behind the avatar. (On average.)

Comments

[Professor Afterthought]

It would be interesting to compare avatar shape based on RL gender - how do men construct male and female avatars compared to how women construct male and female avatars? It might (or might not) lend support to the often heard line, "that's a dude behind the keyboard" when critiquing some of the hypersexualized female avatars we see across the grid.

[Vaki]

Thank you so much, Hamlet!

I have been, for the last two weeks, working on breaking down the data I compiled into some graphs, charts, and ratios that have actually been pretty revealing (and those are even useful with the relatively small sample size of male shapes I had to work with). I really hope to get that data up tonight; it should have already been up, but I've been fighting with Excel.

As a side note, you're right, the male data I had wasn't great in that first post, but for what it's worth, I have kept adding male data as it's added to Strawberry Singh's flickr group. It's still not much, but it's getting better.

[Hamlet Au]

Cool, Vaki! Please ping me when you put up those ratios!

[EnCore Mayne]

the subset (200/registered active users) shows an infinitesimally microscopic result. kinda like defining how similar snowflakes are huh?

[Stef Lohnna]

what we see above is the average SL fashion blogger av, not the average SL female av ... bloggers where the only ones who participated in the digit collection
... and it's in the nature of fashion bloggers to have a fit and sexy, but closer in proportion to the real life women av ...
if we would collect the digits for instant at the 10 most visited sims in SL the result would be COMPLETELY different.

[Scylla Rhiadra]

I too am unconvinced that the numbers given here are representative; I suspect that there is enormous variation in the average to be found from sim to sim, depending upon the nature of that particular place. For instance, the numbers found at, say, a sex site, or a club, are likely to be quite a bit different from those yielded by a survey of an art or fashion sim.

More questionable, however, is your conclusion, which reflects your usual preoccupation with demonstrating that residents are increasingly linking SL and RL identities: "what we're seeing is a desire to have an avatar who is fit and sexy, but closer in proportion to the real life women behind the avatar."

What on earth makes you think that these proportions are representative of "the real life women behind the avatar"? Or even that the person behind the avatar IS a "real life" woman?

Far more likely is that it represents an interpretation, in SL terms, of the current media-induced "ideal" for female proportions, as propagated in advertisements and popular movies and TV shows.

[Vaki]

For what it's worth, as I'm compiling all these numbers and averages, I have in no way intended to suggest that the avatars I created from the data were representative of Second Life as a whole. They aren't; they're representative of the participants in Strawberry Singh's digits project (and even then, I had to discard certain avatar data).

The participants in Berry's project are, in general, readers of fashion blogs, Plurkers, and SLU users. While there is a fairly good variety of avatars in the project (and I strongly encourage anyone interested to check out the Flickr group), the group is not a representation of SL as a whole. SL has a whole not only has a larger percentage of non-human avatars than are represented in the project, but -- let's be honest -- SL has a much, much larger percentage of very poorly-proportioned shapes. If we were to be truly representative of SL, we'd have to include all the noobs who are still walking around 8 feet tall with their breasts set to 100 and their body fat at 0.

Berry's goal in creating the group was not to seek the "average" shape in SL. Berry was, initially, just wondering about her own proportions, and it sparked a lot of community contribution. The community contributions led me to create an average of the participants. I really caution against trying to extrapolate beyond that.

(P.S., I'm not trying to be defensive, just trying to clarify what I mean by "average.")

[pepys ponnier]

Oh for god's sake. Anyone who has seen any of the Brazilian women in SL knows a butt size of 35.xx just doesn't cut it. Which begs the question: What do various cultures value in a woman's figure?

[Markus Breuer (Pham Neutra)]

I always love an analysis like this. And while this one is certainly funny, it probable tells nothing about either the RL person behind the avatar, avatar shapes in SL or beauty ideals of male or female persons behind SL avatars in general. As others have mentioned, this meme was one which spread nearly exclusively in the SL fashion community - and fashionistas have different ideals of female (or male) beauty than many other subgroups in society.

As we all know, the "models" preferred in the western RL fashion industry are thin - usually unhealthily so - and consequently have a nearly androgynous chest. The only odd finding here is the relatively wide (average) hip - which is rather uncommon in RL models.

One could only wish that data about avatar shape in general would be available outside of Linen Lab's servers. This would make for some really interesting analysis ...

[Aliasi Stonebender]

Indeed. My personal human-form avatar is quite tall, for example, and intentionally so. (Even taller are some of the full-prim avatars I wear these days, hoping to revamp them once we get full-on mesh support.)

I think this is an interesting project, but don't draw conclusions from it that aren't supported!

[Eloise]

Printing just averages of any set is dangerous. If half The population have size 100 breasts, the other size 0 you get an average of 50 (very close to the 48 you got).

Modes (and I'd expect more than one mode on many of these) would be more representative. We might, for example see clusters of common heights, clusters of common sizes of other things. Foot size 0 and hand size 20 are the Unsurprising ones, shoes all fit size 0 and nails, rings etc usually fit size 20.

Try to get the stats right - you're sounding more and more like a Tory politician.

[Arcadia Codesmith]

Eloise is spot on.

Averages tell us a great deal about the nature of statistics and very little about the underlying data.

It's a fun exercise, but if you're reading any deep meaning into the results, you might have better luck with tea leaves.

[Vaki]

Good point, Eloise and Arcadia -- and I agreed with you. I hope you'll forgive the shameless blog-whoring, but I went into a lot more depth on this here: http://insertfunnyname.wordpress.com/2011/04/22/going-beyond-average-female/ and here: http://insertfunnyname.wordpress.com/2011/04/25/going-beyond-average-male/. For those who don't want to slog through an insane number of charts, I'll sum up:

The "average" bodies are representative in some ways, and are tremendously unrepresentative in others. For instance, looking at the "Average Female's" height slider: the largest group (the mode) had Height 50, which is also the average (the mean). However, other than those who had Height 50, very few avatars had their heights set in the 50s. A large majority had their heights set in the 40s and the 60s, which averaged together to 49.95 -- 50.

A better example is in breast size (I didn't put up a chart for breast size, because my post was already overflowing with charts, but this makes the point well). The average breast size was 49. However, only two avatars had their breast size set to 49; this was not representative at all. There were three mode points: breast size 50, 55, and 60. While breast size 49 is pretty close to the first, it's not representative of the other two, and you'd think that the large number of avatars with breast size 55 and 60 would push the average higher. But so many avatars had their breast sizes set in the lower ranges and so few had them set in the upper ranges that the average stayed on the low-ish side. Regardless, especially with breast size, I think the mode is more accurate than the mean (in other words, I think the spikes in the frequency chart are more accurate than the average of all the numbers: you're more likely to find female avatars with breasts set to 50, 55, or 60).

I do sort of idly wonder what I might find if I created a shape from the modes of the data rather than from the means, but that'd have to wait a few weeks.