Evaluation of the Hong Kong Government's Consultation Paper Equal Opportunities: A study on Discrimination on the Ground of Sexual Orientation

Sean M. Burke
Department of Linguistics
Northwestern Univeristy
Evanston, Illinois, USA


Because of unclear, vague, and presuppositional wording in the survey questions, exacerbated by serious implementational shortcomings, the survey whose results are reported on in Equal Opportunities: A study on Discrimination on the Ground of Sexual Orientation fails to meaningfully measure the most important public attitudes toward gays, lesbians, and bisexuals.

Problems in Survey Techniques

There are two basic technical problems with the survey whose results are reported on in Equal Opportunities[...]

The first problem is the issue of bias; the survey at several points presupposes ("presupposition" is a linguistic term whose technical meaning is, roughly, "implicate" or "overtly assume") and asserts notions about homosexuality which are not established and which could influence subject's responses; these presuppositions and assertions will be discussed in rest of this paper, after this section.

The second problem is a graver one: the issue of the circumstances of the administration of the test. When dealing with stigmatized or controversial subjects, it has been shown time and time again that the best way to get true indications of what subject think is to ask them through anonymous written forms. The more a subject has to worry about issues of confidentiality and the more a subject has to interact with another person in the administration of the test questions, and the greater the status of the person or organization in question, the more likely survey subjects are to "tow the party line" (or to use another popular metaphor usually employed in elections, "hop on the bandwagon") instead of give their own opinion, resulting in survey findings which are grossly exaggerated in favor of what the subjects perceive as the mainstream position.

With this in mind, the methodological design of this Consultation Paper survey is almost a worst-case scenario in terms of the biases it introduces. The interview is conducted entirely verbally, maximizing interaction with the pollster. What's more, in the script that pollsters are given to read (see Appendix III, p. 1, "Introduction"), there is no assurance of confidentiality. More threateningly, the pollster explains his affiliation with the Hong Kong Government. If for no other reason at all, I personally would be very cautious in responding to a survey conducted by a government that will be absorbed, to some degree and in some fashion, by the People's Republic of China, the latter government not being, by any stretch of the imagination, known for progressive views on homosexuality or bisexuality, nor known for permitting or fostering wide ranges of public opinion on controversial social and political issues.

If I were a survey subject, unless I were assured that, say, my responses would be anonymous and that record of them would be destroyed immediately after the final tabulation (as is done in elections in many parts of the world), my immediate response would be to watch what I say, lest record of it be looked at with an unkind eye in uncertain years to come.

Linguistic Problems

In this section I will illustrate the ambiguities in the first (and most important) question in the survey, in terms I hope will be clear to non-linguists. I will note the problems these ambiguities pose for the validity of the survey results. I will then note some of the additional cases of ambiguous wording throughout the survey.

The primary error in this document is lack of clear wording; this survey presents the subject with concepts he is asked to offer opinions on. When the survey doesn't express a concept precisely, then what the subject is reacting to could be something quite different from what the survey writers intended.

Question 1. Have you heard about ... [het./homo./bisexuality]?
Use of the preposition "about" here (as opposed to the more expected "of") form carries the strange implication that the speaker has something in mind relating to the topic he's naming, which he's presupposing that the hearer is familiar with. Compare this "of" example:
--Have you heard of Jake Salopec?
--No, who is he?
with this "about" example:
--Have you heard about Jake Salopec? (meaning: I assume you know who this person is. Do you know about the recent salient item of news about him?)
--I don't even know who he is. (your presupposition that I know who he is, is false.
The reader should note that a simple "no" does not convey disagreement with the presupposition of familiarity with the subject named. A simple "no" has this effect:
--Have you heard about Clinton? (We have shared knowledge about who Clinton is. Do you know about the recent salient item of news is about him?)
--No. (I do not know the salient news item; but the fact that I say I don't cooperates with your presupposition that I know who Clinton is.)
--Oh that bill he signed; yes, it's awful!! (I know who he is, and I believe I know what item of news you're referring to.)
Ergo, asking "have you heard about" homosexuality implies that there is some salient item of news, which is a nearly nonsensical concept; who can imagine an item of news can be predicated of a sexual orientation common to vast segments of Earth's population, as in:
--Have you heard about heterosexuality?
--No, what. (I.e., I know what heterosexuality refers to, but I don't know what news item there is about it.)
--They've just proved it's caused by zucchini!! (or whatever)
So, no possible "yes"/"no" response (which is all the blanks allow) to that question "have you heard about homosexuality?" can express the idea "I don't know what that is" (much less "I'm not sure we mean the same thing by it"), because knowing what the x is in "have you heard about x" is presupposed, and isn't negated by a simple "yes" or "no".

This means that because of this imprecise wording on the part of the survey writers, the survey, in this question, will not accurately elicit the responses that the survey writers intended (i.e., "yes, I know what that refers to", or "no, I don't"). Thus, the statistics for this are not a reliable indicator of the true number of people who believe they know what the word "homosexuality" (&ct.) refers to.

This imprecision in wording of the questions, as well as the failure to consider the possible meanings of the answers (as with these "unanswerable" questions) is, in my estimation, the greatest source of uncontrolled variables in this survey, and this is illustrated perfectly in this first question.

I focus on this first question because this is the question that, if better worded, should be used to separate those who know what homosexuality and bisexuality are (or who report that they do, at least) from those who don't; that these groups should be analyzed separately is obvious-- people who've never heard of a given issue can't have opinions, only reactions.

By failing to separate (or throw out) subjects who are not familiar with the terms "homosexuality", "bisexuality", and "heterosexuality", the survey introduces a significant uncontrolled variable into the results; i.e., what the reactions are of people who do not have a clear conception of what sexual orientation is. Judging from Table One in Equal Opportunities[...], this group is at least 25% of the sample.

Now, in the case where the subject has a concept of all these sexual orientations, but where they differ from the survey writers' (e.g., in conflating transvestism with homosexuality, say), again a significant uncontrolled variable is then introduced, namely the variable of whether the subject is using own his definition of the word (which he has not been instructed to refrain from doing), or whether he's accommodating (i.e., using for the time being) the survey writers' definition which is given to him.

The only case in which question 1 does not introduce (or fail to screen) a uncontrolled variable is if the speaker has already has the same conception of sexual orientation it as the survey writers -- in which case question 1 is superfluous at least.

Further Linguistic Problems

Question 2, central to the mission of the evaluation paper, asks for the subject evaluation of these concepts: Question 6, which seeks to measure opinions on some very important issues, also contains a good number of ambiguities which are a source of uncontrolled variables.

Implicit Problems

The first of the two implicit problems in this survey is its incorrect working definitions of homosexuality and bisexuality.

The survey tells the subject that "heterosexuality" refers to the phenomenon of being attracted to people of the opposite sex, "homosexuality" to people of the same sex, and "bisexuality" to both. One can be attracted to the opposite sex without being heterosexual; the issue is whether one is primarily attracted to the opposite sex.

In fact, even just adding the "primarily" in doesn't resolve the problems with the definition of homosexuality/heterosexuality which the survey presupposes and, in fact, explains to the survey subjects (see p. 3 of the questionnaire).

The second of the implicit problems with this survey is the strong misconceived presupposition it gives the subjects that they do not already know a great number of gays, lesbians, and bisexuals. This is seen most strongly in survey questions 3, 4, and 5, which ask whether the subject "would mind or wouldn't mind" a number of activities in the company of homosexuals or bisexuals. The use of the conditional tense here gives a very strong "contrary-to-fact" reading to the actions listed, implicating that the subject does not already see movies with homosexuals and/or bisexuals, dine out with them, etc. This is the most widespread public misconception about homosexuals, and in some ways is primary to all others: it is the misconception that the average person does not know any gays, lesbians, or bisexuals.

The established distribution of gays, lesbians, and bisexuals through the populace ensures that almost everyone knows one. I believe it would be extremely illustrative to measure the extent of this misconception of "not knowing any of them": it would simply entail asking "Do you know [are friends with, are related to, do you work with, etc.] a gay, lesbian, or bisexual?" But this survey, oddly, does not do this.

It is a basic question whether survey subjects who harbor the misconception that they do not know any gays, lesbians, or bisexuals are fit to be answering a survey about them. This survey makes no attempt to weed out such subjects. Instead, it feeds just this misconception, via the wording of Question 3, 4, and 5 especially.


While I do not assert that this survey was intentionally biased, I do believe that linguistic bias and sloppy methodology has introduced a great number of significantly uncontrolled variables into the survey administration, which are serious enough to cast doubt on any and all of the findings of this consultation paper.
sburke@babel.ling.nwu.edu, Sat Mar 30 19:34:27 CST 1996