Thinking about polls
Some stuff I've figured out while designing polls online (specifically on lj clones). This is not about Real Scientific Polls, which one hopes are designed with Real Scientific Ethics etc in mind. I'm still figuring this stuff out, so please tell me what I've missed/screwed up etc.
I guess overall the most important thing I've learned: Never make people feel excluded, even if it messes with your data a bit. You're not doing Proper Research here anyway. And unless your poll is on a locked post and you know your flist really well, don't assume everyone taking it cisgendered/American/Western/straight/young/etc.
Specifically, unless it's a joke question always have an "other" option, and where possible use checkboxes. That way people who fall between the cracks of your questions (and there'll always be someone) have a better chance of finding an answer that better describes them.
Taking this further is the "Other (please specify below)" option, with a textbox. I must admit I rarely do this but I've seen it suggested as a standard for all polls where people's identities are at stake.
It's also important to avoid ambiguity, consider all the ways the question could be read and whether or not it makes sense to all the possible people who could take it.
EDIT: As![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
pippin points out, a more subtle problem (which I fell into!) is including other options but making your own central (this is the danger of over-reliance on "other" as a catch-all) For example, listing the "normal" options first and the "unusual" ones later. I've edited my post to mix them up a bit.
Second: People's identities are very important, so be really careful when asking about them. If you say "What's your favourite icecream, vanilla or chocolate?" the pistachio fans will be justly irritated but will live. The vegans will be more justifiably annoyed, wondering whether you include vegan icecream. But if you say "Are you male or female?" you're buying into really hurtful transphobic oppression. A follow on to this is that unless it's really relevant to your poll, categorise people by how they self identify, not how others identify them. People spend enough time being told others get to categorise them.
Specifically, gender questions should (afaict!) at the very least say
"What is your gender identity?
-female
-trans/genderqueer/intersex
-male
-other"
and these MUST be tickboxes, not bullets.
Sexuality should include (both explicitly and implicitly, eg when talking about relationships or attraction): asexual, bisexual, pansexual, trans, lesbian, poly, gay, genderqueer, straight, intersex, queer, other.
Nationality is a much thornier question than you'd think. There's where someone lives, where they happen to be geographically right now, where they were born, where they grew up, where they feel included/at home, their ethnicity..
Similar issues apply to religion, ethnicity etc.
Finally, respect people's right to privacy. If the topic is something they might not want aired in public, make the poll results only viewable to yourself. This still leaves the aggregate data publicly viewable, but not people's individual responses. Some example topics I've encountered (and screwed up on :( ) are experiences of sexual assault, gender identity, and sexuality.
I guess overall the most important thing I've learned: Never make people feel excluded, even if it messes with your data a bit. You're not doing Proper Research here anyway. And unless your poll is on a locked post and you know your flist really well, don't assume everyone taking it cisgendered/American/Western/straight/young/etc.
Specifically, unless it's a joke question always have an "other" option, and where possible use checkboxes. That way people who fall between the cracks of your questions (and there'll always be someone) have a better chance of finding an answer that better describes them.
Taking this further is the "Other (please specify below)" option, with a textbox. I must admit I rarely do this but I've seen it suggested as a standard for all polls where people's identities are at stake.
It's also important to avoid ambiguity, consider all the ways the question could be read and whether or not it makes sense to all the possible people who could take it.
EDIT: As
pippin points out, a more subtle problem (which I fell into!) is including other options but making your own central (this is the danger of over-reliance on "other" as a catch-all) For example, listing the "normal" options first and the "unusual" ones later. I've edited my post to mix them up a bit.Second: People's identities are very important, so be really careful when asking about them. If you say "What's your favourite icecream, vanilla or chocolate?" the pistachio fans will be justly irritated but will live. The vegans will be more justifiably annoyed, wondering whether you include vegan icecream. But if you say "Are you male or female?" you're buying into really hurtful transphobic oppression. A follow on to this is that unless it's really relevant to your poll, categorise people by how they self identify, not how others identify them. People spend enough time being told others get to categorise them.
Specifically, gender questions should (afaict!) at the very least say
"What is your gender identity?
-female
-trans/genderqueer/intersex
-male
-other"
and these MUST be tickboxes, not bullets.
Sexuality should include (both explicitly and implicitly, eg when talking about relationships or attraction): asexual, bisexual, pansexual, trans, lesbian, poly, gay, genderqueer, straight, intersex, queer, other.
Nationality is a much thornier question than you'd think. There's where someone lives, where they happen to be geographically right now, where they were born, where they grew up, where they feel included/at home, their ethnicity..
Similar issues apply to religion, ethnicity etc.
Finally, respect people's right to privacy. If the topic is something they might not want aired in public, make the poll results only viewable to yourself. This still leaves the aggregate data publicly viewable, but not people's individual responses. Some example topics I've encountered (and screwed up on :( ) are experiences of sexual assault, gender identity, and sexuality.
no subject
no subject
no subject
- trans/genderqueer/intersex
avoid this kind of lumping maybe?
most trans people aren't third-gendered. some genderqueer people are, some intersex people are.
no subject
no subject
no subject
no subject
That's true, and why I said if possible. I think the textbox option works better if the survey taker is going to report on and discuss the results, making their own charts (which I've seen people do for fun/curiosity), or if a person is writing their own survey script.
Personally, if its a checkbox/button style survey, I'd rather click 'other' than try to puzzle out what the survey writer and other survey takers are assuming when they list trans/genderqueer/intersex (together or separately, and if under questions about sex, gender or both). Of course, I'm sure others will disagree; as long as the survey taker is defining the choices on sensitive subjects like this its not possible to please everyone.
no subject
no subject
Even if the choices given are confusing to me... if a survey creator thinks to put ANYTHING other than just the binary M/F option, I'm usually pleased to see that they are aware and trying to be inclusive.
no subject