Page 46-72, 26 pages Page 1 of7
Chapter 3: Sex Research
• Sex research varies in term of: 1) relying on self-reports or direct observation; 2) large studies by
surveys or small studies in the lab, qualitative research, or case studies; 3) studies conducted in the lab
or the field; 4) behaviour studied naturally or manipulated experimentally
ISSUES IN SEX RESEARCH
• Population refers to the group of people a researcher wants to study and make inferences about – e.g.
Canadian adults, adolescents in Saskatchewan, etc. Generally, a sample, or part, or a population is
• Sampling affects generalizability – study of those only with sexual pathologies is not representative
• To get a sample which is representative of the population at large, we can use probability sampling
where each member of the population has a known probability of being included in the sample. The
simplest form is random sampling, e.g. randomly select 1 out of every 50 teenagers in Saskatchewan.
• Stratified random sampling divides the population into groups, and then random sampling occurs in
• 3 phases of sampling: Population identified, method of sampling adopted, people in sample contacted
and asked to participate. The problem of refusal or non-response when people refuse to participate,
making random sampling difficult.
• There may be a volunteer bias when those in the sample may differ than those who refused to
• Self-Selection Bias: Personality characteristics may predispose some individuals to participate, and
others to avoid participation. It is likely that those who volunteer hold more permissive attitudes about
sexuality and are more sexually experienced.
• Convenience sample: A sample chosen in a haphazard manner relative to the population of interest,
not a random or probability sample.
Accuracy of Self-Reports of Sexual Behaviour
• Purposeful distortion is purposely giving false information in a survey. People may either exaggerate
their sexual activity (enlargement) or minimize or hide it (concealment), in the direction they believe will
be seen as more socially desirable or acceptable.
• Distortion can be minimized by stressing scientific purpose and confidential nature of the survey.
• Some questions require respondents to recall sexual behaviour from many years prior. Studying
childhood behaviour presents this problem, but alternative of getting data from children currently raises
ethical and practical problems.
• Diary method can be used to record behaviour every 1-2 days
Difficulties with Estimates
• It is difficult to estimate time, especially when involved in an absorbing activity such as precoital
• Men report duration as 13.4 minutes, significantly longer than female estimation of 11.3 minutes.
Interpreting the Question
• Subjects may not agree in definitions of seuxla terms such as sexual partner, having sex, and
abstinence – does sex include oral sex, any time an orgasm was had, etc.? Page 46-72, 26 pages Page 2 of 7
Evidence on Reliability of Self-Reports
• Test-Retest Reliability identifies whether self-reports are reliable or accurate, where participants are
tested a second time some time later to determine similarity of answers. The correlation measures
reliability, with 1.0 meaning perfect reliability, i.e. identical answers.
• Another method is to obtain independent reports from two people who share sexual activity, i.e.
Interviews vs. Questionnaires
• With interviews (face-to-face or phone), the interviewer can establish rapport and honesty, and has
flexibility to vary the sequence of questions depending on responses. Interviews can be administered to
persons who cannot read or write. This comes at the cost of feelings of anonymity.
• Sensitive information like risky sexual behaviour is more likely to be revealed in a questionnaire, while
and traumatic experiences such as sexual assault are more likely to be revealed in a face-to-face
• Computer-Assisted Self-Interview Method (CASI): Method of data collection where respondent fills
out questionnaires on a computer combined with audio component to accommodate children or poor
readers. The computer can also be programmed to follow varying sequences of questions depending on
• Web-based surveys can recruit much larger and broader samples – recruit across geographical
• Advantage for studying special populations, particularly if they are defined by taboo sexual behaviour
and do not actively participate in special interest organizations, e.g. LGBTQ individuals.
• Web-based surveys can eliminate extraneous influences such as gender of interviewer.
• Bias from relying on self-reports, Internet users typically having higher income, and lack of control in
Self-Reports vs. Direct Observations
• Direct observations tend to be more accurate; however, they are expensive and time-consuming,
leading to small samples, and tend to introduce even more volunteer bias
• Sexual behaviour in the laboratory is likely not the same as in the privacy of one’s relationship –
• Robert Latou Dickinson, an American gynecologist, make significant contributions to understanding
women’s reproductive physiology by using a glass observation tube in the vagina and cervix
• Factors like gender, race, and age of the interviewer, and wording of questions all influence answers
• Questions about sensitive issues such as extramarital sex may be worded in the standard or supporting
• Cardinal principle is respect for human dignity
Free & Informed Consent
• Free & Informed Consent: Participants have a right to be informed, before participating, of the
purpose of the research and what they will be asked to do. When children are too young to give consent,
parents are consulted
Protection from Harm Page 46-72, 26 pages Page 3 of 7
• Researchers must minimize the amount of physical and psychological stress participants undergo; this
includes the principle of respect for privacy and confidentiality to ensure no future repercussions
• Vulnerable persons particularly need to protected
• The justice principle holds that the risks of participation and the benefits of research results should be
distributed fairly across groups in society, e.g. birth control pills tested across social class and geography
Balancing Harms & Benefits
• Harms-benefits analysis analyzes the ethics of a study by weighing harms such as stress to subjects,
against benefits such as knowledge about human sexuality. This is particularly true for sex research, as
people consider sexuality very private and may find it distressing to disclose their experiences,
behaviour, or attitudes/
• “Tuskegee Study of Untreated Syphilis in the Negro Male” (1932-1972): 399 poor, uneducated African
American men did not give informed consent, did not get medical treatment or were warned against
taking penicillin, and were given coercive incentives to continue in the experiment. The harm to the
participants far outweighed knowledge from syphilis learned.
SOME STATISTICAL CONCEPTS
• Mean is the average of respondents’ scores, calculated by adding the scores and dividing by the
number of people
• Median: The median score, the score that splits the sample in half with half of people scoring below the
number and half scoring above it.
• Mode: The most frequent score, the score with the greatest number of responses.
• Variability tells us the range of responses from all the respondents – great difference when average
number of intercourse/week is 3 and range is 2 to 4, vs. range is 0 to 15.
Average vs. Normal
• Average =/= normal; there is so much variability in sexual behaviour that any within a wide range is
Incidence vs. Frequency
• Incidence is the percentage of people giving a particular response. Frequency is how often people do
• E.g. Incidence of masturbation among males may be 92%, while frequency is about once a week.
• Cumulative incidence refers to the percentage of people who have engaged in a behaviour before a
certain age, e.g. of masturbation is 10% by age 11, 80% by age 15, etc.
• Correlation measures the relationship between two numbers, ranging from +1.0 to -1.0
• Test-retest reliability correlations typically range between +0.6 to +0.9
THE MAJOR SEX SURVEYS
• In these major sex surveys, data were collected from a large sample of people by questionnaires or
interviews. Page 46-72, 26 pages Page 4 of7
The Kinsey Report
• Kinsey and colleagues interviewed 5300 men and 5940 women between 1938 and 1949 in the United
• Kinsey’s goal was to collect sex histories from as wide a variety of people as possible, and was not
concerned with sampling issues, for fear of problems of non-response
• Overrepresented populations included: university students, young people, well-educated people,
Protestants, city-dwellers, and people living in Indiana and the Northeast
• Underrepresented populations included: manual labourers, the less educated, older people, Roman
Catholics, Jews, racial minorities, and rural-dwellers
• Interviewers established rapport and treated all reports matter-of-factly. They phrased questions in
language that was easily understood and which encouraged responses, e.g. “At what age did you begin
masturbating” over “Have you ever masturbated?”; this minimized purposeful distortion
• Strict precautions to ensure anonymity, with data stored on IBM cards with code that only a few people
directly involved had memorized. Researchers even made contingency plans for destroying the data if
police attempted to access it for prosecution.
How Accurate Were the Kinsey Statistics?
• Criticisms focus on failure to use probability sampling, causing systematic errors of unknown magnitude
• Level of sexual activity, particularly high incidence of same-sex activity, may have been particularly
subject to selective sampling errors. Single most doubtful statistical figure is likely incidence of
Sexual Behaviour in t