Choosing the Sample

The questions asked by sexologists often apply to populations that are too large to study in their entirety. For example, if you wanted to obtain information about the sexual practices of American married couples in their later years, your population would include all married couples in the United States over a given age, say, 65. Clearly, it would be impossible

Sex Research: Methods and problems

to question everyone in this group. Sex researchers resolve this problem by obtain­ing data from a relatively small sample of the target population. The confidence with which conclusions about the larger population can be drawn depends on the technique used to select this sample.

Typically, researchers strive to select a representative sample (sometimes called a probability sample)—that is, a sample in which various subgroups are represented proportionately to their incidence in the target population. Target populations can be subdivided into smaller subgroups by such criteria as age, economic status, geographic locale, and religious affiliation. In a representative sample, every individual in the larger target population has a chance of being included.

What procedures would you use to select a representative sample that could be sur­veyed to assess the sexual practices of older American married couples? How would you ensure the representativeness of your selected sample? A good beginning would be to obtain U. S. Census Bureau statistics on the number of married couples whose partners are age 65 and older who reside in major geographic regions of the United States (East, South, and so on). Next, you would select subgroups of your sample according to the actual distribution of the larger population. Thus, if 25% of older married couples live in the East, 25% of your sample would be drawn from this region. Similarly, if 15% of older married couples in the East fall into an upper socioeconomic status category, 15% of those subjects selected from the East would be drawn from this group.

Once you had systematically compiled your lists of potential subjects, your final step would be to select your actual subjects from these lists. To ensure that all members of each subgroup had an equal chance of being included, you might use a table of ran­dom numbers to generate random selections from your lists. If these procedures were correctly applied, and if your final sample was large enough, you could be reasonably confident that your findings could be generalized to all married American couples age 65 or older.

Another kind of sample, the random sample, is selected from a larger population, using randomization procedures. A random sample may or may not be the same as a representative sample. For example, assume that you are a social scientist on the faculty of a rural university in the Midwest whose students are inclined to hold relatively con­servative social views. You wish to conduct a survey to assess American university stu­dents’ experience with "hook-ups" (short-term or one-time loveless sexual encounters between strangers or casual acquaintances). Since it is convenient to draw your subjects from the student population enrolled at your university, you randomly select your sur­vey sample from a roster of all enrolled students.

A substantial majority of your sample respond to your well-designed, anonymously administered questionnaire. Can you now be relatively confident that your results reflect the propensity of students to engage in hook-ups—if not in the greater United States, at least in your geographic region? Unfortunately, you cannot, because you have selected subjects from a sample that is not necessarily representative of the broader commu­nity of university students. Students at your university tend to have conservative social views, a trait that may both influence their likelihood of engaging in hook-ups and ren­der them atypical of large segments of American university students, especially those enrolled in urban universities on the East and West Coasts, whose student populations tend to hold more-liberal social views.

Thus, even though randomization is often a valid selection tool, a study sample can­not be truly representative unless it reflects all the important subgroups in the target population. All things considered, representative samples generally allow for more — accurate generalizations to the entire target population than do random samples. How­ever, random samples are often quite adequate and thus are used widely.

Updated: 02.11.2015 — 15:23