On 02 July 2003, Edward Castronova wrote:
> Surveys of opinion are bad sources of data.
Well, they may be sources of bad data, but I wouldn't say they were
bad sources of data. They're good enough that politicians listen to
them come election time.
Besides, if surveys of opinion are bad sources of data, what does
that make statements of opinion by single individuals? Just because
30 people say something, that doesn't make it true; however, it
carries 29 people more weight than the opinion of 1 person (assuming
their opinions are equal-value).
> Self-selected surveys of opinion are worse.
So we shouldn't actually be listening to any of the arguments going
on in MUD-DEV, then, as they're the opinions of people who have
actively decided to participate in the debate (yourself excepted, as
you were drawn into it by remarks made about your paper).
> Self-selected surveys of opinion about sensitive topics are worse
> than bad, they are lie collectors.
I should mention here that while Katherine Wright's (and Nick Yee's)
surveys are self-selected, the Parks & Roberts one wasn't. In their
first survey, they selected the biggest two social and educational
MOOs (LambdaMOO and DU), then 5 more MOOs chosen randomly from a
comprehensive list of MOOs (Gurk's MOO page). From the entire player
base of these 7 MOOs, they randomly selected candidates for survey
among those who had played in the previous 14 days. 1,200 characters
were solicited, of which 233 responded. There was some redundancy
due to people having multiple characters (and therefore being
solicited twice), and some failure from people who were disqualified
from entering by reason of age (for legal reasons, players had to be
18 or above). Parks & Roberts reckon they got a response from about
30% of the people who could possibly have responded.
So although there is some of self-selection here, in that people
didn't HAVE to reply, it doesn't allow for people with fiery
opinions to charge in unless they were invited. A 30% (or even a
19.5%) response rate is sufficiently large to make a survey be
considered general, rather than purely self-selected.
> Go up to a male gamer, identify yourself as a person doing a study
> on men playing women avatars, ask him a bunch of questions about
> that. Of course - of course - he is going to say a bunch of bland
> things. I'm sorry, but in my opinion, much of that kind of
> research isn't citeable. It doesn't get at the truth of people's
> motivations any more than my study did.
Whoa, there!
You said "But no man has said that they actually do play a female
avatar because, on the whole, they like how gameplay is when you're
female.". I produced evidence that actually, 60% of players in a
survey did say that.
It doesn't matter whether you think the survey was flawed (which of
course it was) or that the sample set was small and self-selected
(which of course it was): the fact is that about 20 people DID say
what you say you haven't seen people say. Now either you have to
accept that some people actually DO think that they get a gameplay
advantage by playing as a male (and that you were merely unaware of
it earlier), or you accuse them of lying.
Accusing people of lying is a fairly strong allegation; if you did
so, I'd want to know why you think these guys would all want to give
the impression they perceived a gameplay advantage by playing as a
female when actually they don't.
I have some sympathy with you in that there is often a frustrating
lack of rigour among social scientists in their surveys, not
necessarily always their fault. However, that does not mean that
none of them can ever tell us anything. Some of them can tell us
very useful things indeed, even if not what the surveyor was
intending that we be told.
I quoted Wright's survey because it gave counter-examples to your
"no man has said..." statement. Men HAVE said it; you either can't
proceed with that line of argument, or you have to change it to "no
man has said without lying..." and justify why you believe that
those who said it were indeed lying.
I quoted Parks and Roberts because you said "I keep wondering, if
the two avatars do the exact same things, why should anyone care
what gender they are?". Well, now you know why they care. You can
completely ignore the replies of the 202 Survey 2 respondents on the
grounds that they were, as one, hiding their true reasons. On the
whole, though, given the thoroughness of Parks and Roberts, the
chances that everyone was lying are somewhat remote.
> Evidently people disagree about what it all means. We only have
> our own opinions based on our own experiences.
We only have our own opinions if we refuse to believe those
solicited by trained professionals who spend their whole lives
conducting surveys.
I don't believe that Wright's survey reflects in its numbers the
reality of why people cross-gender play. In quoting it, though, I
didn't need to believe this; I only needed one statement from it,
which could have a margin of error of 59% and still do its job.
I believe that Parks and Roberts' survey IS a very good reflection
of the reality of cross-gender play at the time it was undertaken
(1999) in MOOs. Their case is strengthened by the fact that some of
their results coincide with those of earlier, less formal surveys
(that they weren't aware of) which came up with simular figures (for
MOOs). The numbers will be different for EQ, and there will be other
categories due to EQ's more gameplay-oriented approach; however,
that doesn't mean that any of the explanations for cross-gender
playing that Parks and Roberts cite are inapplicable, and it was to
list these explanations that I cited their survey.
> We can't advance the discussion any more without trying to unpack
> what proper treatment of males and females is or would be, and
> that's a topic for another list I think.
I agree, but it looks like it's going to continue on this list for
some time anyway! <grin>
Richard