So I’ve got this thorn in my side.
The Canadian University Survey Consortium (CUSC) recently published the results of their first year student survey and reports on differences between visible minorities and non-visible minorities. All well and good. It seems as if analysts arrive at these two groups through back-coding; that is, based on respondents’ self-reported data they group respondents into categories that better suit their analyses. Again, all well and good. This is common practice and is by no means the only time this type of demographic data has been reported in this way.
They report some pretty impressive participation rates, so naturally I took a look at the survey instrument. This is what gets me. Their questions about ethnicity ask about ethnicity… and every other identity under the sun.
I may be overreaching here, but I think it’s safe to say that in survey research, demographic questions are a bit of a snake in the grass. They are included in almost every survey you will take, but seem to bite you in the most unpleasant places while you’re developing the instrument. Demographic questions about sensitive personal characteristics even more so.
The thing is, how can we--the big societal we--ask a question when we haven’t even decided on what we’re actually measuring?
There is real value in being able to point to socially constructed categories and say, “we’re discriminating against people based on x, y, and z and the results are a, b, and c.” It’s neat. It’s tidy. We need good evidence like this for recommending good public policy changes.
But the way we ask demographic questions about race has to change. These questions want to be asking about race but by doing so with “sensitivity,” really end up asking about something else entirely and ultimately avoid the actual issue: racism.
The question they use is phrased like this (and again, CUSC is not alone. This question seems to be based on a similar one used in Statistics Canada’s General Social Survey):
Are you...? (check all that apply)
- Aboriginal (e.g., status, non-status, Metis, Inuit)
- Arab (e.g., Saudi, Egyptian, etc.)
- Black
- Chinese
- Filipino
- Japanese
- Korean
- Latin American
- South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.)
- Southeast Asian (e.g., Cambodian, Indonesian, Laotian, Vietnamese, etc.)
- West Asian (e.g., Afghan, Iranian, etc.)
- White
- Other (please specify)
You’re going to find similar questions littered throughout the surveys administered by government and other public institutions. When I read this, it raised so many questions:
Are they trying to determine if the respondent would be considered part of a racialized group? Are they trying to determine the respondent’s identification with any diaspora? Their citizenship or immigration status? What will the lines of analysis be after this information is collected?
And these just come from underneath my researcher’s hat. From a more personal perspective, I have just one question:
How do I answer this question?
If I don’t have an answer, chances are there a ton of other people that won’t either.
So why do we keep using this setup? I think the issue comes down to institutional-powers-that-be avoiding tough questions that are typically used to interrogate every other survey question. So let’s do some interrogating, shall we?
Generally on a survey you want to group numerous response options in ways that scale together. That is, we want to find factors or indicators of a particular thing. Based on the analysis, in the case of the question above that thing is ostensibly race… or is it ethnicity? Or is it a family’s national origin? It should throw up a red flag that this question is not asking about one single thing. Yes, “Japanese” can be interpreted as a culture, a nationality, a diaspora, and loosely a race. That’s just the problem. No two survey respondents will answer this question in the same way.
These response options are just not comparable. So we’re going to ask about several permutations of having some type of genetic (or is it just geographic?) connection to the Asian continent, but not of the African continent? Not of the South American continent? Not of several other subregions like the Caribbean? Sure, the question offers a number of diverse options of ways that individuals may choose to describe their identity, but it is also offering a number of options that conflate geographic, diasporic, and cultural descriptions with the colour of one’s skin.
Oh no! I’ve said it. When we’re asking about race we’re really asking what colour you are! Ick! Ack! Ugh.
I understand the need for respondents to see parts of themselves in the survey instrument; there needs to be signals that show respondents that their experiences are valued and that the researchers have some sense of who they are. This question instills no such confidence. This question does not encourage a respondent to engage with the survey at hand.
This question is a clumsy attempt at inclusivity that only serves to betray itself in the end. With this many options and the high probability of respondents answering in many different ways, the analysis is going to come down to this: is the respondent in a dominant group, or a non-dominant group? Does the respondent occupy a place of power and privilege in society? Is the respondent white or not?
Because sometimes self-identification is only used in service of others’ perceptions.
So sometimes I just want to answer “Other,” without specifying.
We know that social perceptions of race are a crucially important factor in the way that people experience social hierarchies and institutions. We know that race has a sometimes devastating influence on experiences of violence. Race surely impacts access to opportunity. So, how do we measure it?
That’s a question I do not pretend to have the answer to, but I know that the question above is not getting at it. When we ask about race on surveys, we’re not only asking about the race (or races) an individual identifies with, but also about the ways that others’ perceptions of their race influence the way they experience and move through life. This is not an easy thing to measure with just one survey question!
I don’t think it will ever be enough to expand response options until everyone feels included because we haven’t stopped to really consider what we want to know. I’ll repeat this again, and maybe later we can have a more constructive discussion about how to measure the answers:
Is the respondent in a dominant group, or a non-dominant group? Does the respondent occupy a place of power and privilege in society? Is the respondent white or not?
Danielle Pierre
Research & Policy Analyst
Ontario Undergraduate Student Alliance