Читать книгу Bisexuality and the Challenge to Lesbian Politics - Paula C Rust - Страница 20

Chapters 4, 5, and 7 will present the findings of research in which I explored lesbian and bisexual women’s opinions on the topic of bisexuality. But before I present the findings it is important for you to know something about the women who participated in the study. The odds are that you and your friends were not among the women who participated, so you might be wondering how the findings could possibly be relevant to you, much less reflect your own opinions. This chapter will answer that question, and it will give you information that you need in order to draw your own conclusions about the findings. Readers who are familiar with scientific sampling methods might want to skip to the subheading “How Lesbian and Bisexual Women Were Recruited to Participate in the Study.”

It would have been impossible, for many practical reasons, to survey all lesbians and bisexual women in the U.S. Instead of surveying an entire population, social scientists usually select particular members of the population as representatives of the population and survey these people. In other words, we take a sample of the population and then we use the information these people give us to draw informed conclusions about the population as a whole. In this way, scientists use limited financial resources to focus on getting the most accurate information possible from the people they have selected, instead of obtaining poor quality information from a larger number of people.

It may seem risky to draw conclusions about a whole population based on information from just a sample, and it can be if it is not done properly. How do scientists know that the people who are sampled really represent everyone else fairly? We do, if we have drawn the sample using methods that guarantee that each member of a population has an equal chance¹ of being selected for inclusion in the sample. When these methods are used, we can be reasonably certain that various segments of the population, and their opinions, are represented in the sample in the same proportion in which they appear in the population as a whole. Therefore, the sample should provide an accurate micropicture of the whole population. There is always a possibility that this will not be the case, but even the magnitude of this possibility is known if scientific sampling methods have been used. Social scientists generally do not report findings from a sample unless they are at least 95% certain that the findings are an accurate reflection of the whole population. Readers who are unfamiliar with scientific sampling procedures can consult any textbook on social scientific research methods.

But lesbian and bisexual women cannot be sampled using representative sampling methods. In order to draw a representative sample a researcher must begin with a complete listing of all members of the population. It would be impossible to make a list of all lesbian and bisexual women because many of us are isolated or closeted. Therefore, lesbian and bisexual women have to be sampled by methods that are designed to maximize the diversity of the sample, rather than its representativeness. In other words, participants need to be recruited in ways that guarantee that women of different socioeconomic classes, racial/ethnic groups, educational levels, incomes, ages, political orientations, etc., are included in the sample. Their number in the sample will probably not be proportionate to their number in the whole population, but they will be represented.

When samples are recruited by methods that emphasize diversity rather than representativeness, findings have to be interpreted with special care. For example, if I had been able to draw a representative sample of lesbians and I found that 15% had children, then I would be able to say, with a known degree of certainty, that 15% of all lesbians have children. But, since the lesbians who participated in my study are not a representative sample, I do not know how accurate it would be to conclude that 15% of all lesbians have children. I can, however, look at differences between groups of study participants with some confidence. For example, later in this chapter I will report that 15% of lesbian-identified women in the study had children, whereas 25% of bisexual-identified women in the study had children. On the basis of these results, I have reason to believe that in the population at large bisexual-identified women are indeed more likely to have children than lesbian-identified women. Neither the 15% nor the 25% might be accurate; both of these figures might be inflated or deflated. But, unless I have some reason to believe that there were factors in my sampling methods that affected lesbian-identified mothers’ participation rate differently than bisexual-identified mothers’ participation rate, I can cautiously assume that both figures are equally inflated or deflated and that the difference between them reflects a real difference between the rates of motherhood among bisexual and lesbian-identified women in general.

In chapters 5 and 7, the findings will sometimes be presented in terms of differences between lesbian and bisexual women, or between women of different ages, racial/ethnic groups, classes, etc. Most findings will not be presented in terms of percentages or other numbers; most will be anecdotal or narrative descriptions of the different opinions that were expressed by women who participated in the study. However, when findings are expressed in terms of numbers, they will sometimes be accompanied by a “test of statistical significance.” There are many different kinds of statistical significance tests, and each is appropriate under particular circumstances, but all significance tests produce a “p-value.” Above, I mentioned that social scientists do not usually report findings from samples unless they are at least 95% certain that the finding is an accurate reflection of the whole population. The degree of uncertainty is measured by the p-value. The lower the p-value, the more certain the researcher can be that a finding in the sample is also true for the whole population. For example, if the p-value for a finding of difference between two groups in the sample is .05, then there is a 95% chance that these two groups really are different in the whole population. A p-value of .01 indicates 99% certainty, and so on. Significance tests were designed for use on representative samples, and they are accurate for these types of samples. Because the sample on which this study is based is not a representative sample, the significance tests are only guidelines to the certainty of the findings.