A Framework for Conceptualizing Reporting Bias and Its Antecedents in Interviews Assessing Human Sexuality.

As with research on other sensitive topics, research on human sexuality has its share of methodological problems (see Catania, Binson, Van der Straten, & Stone, 1995; Catania, Gibson, Chitwood, & Coates, 1990). One area of particular concern is the sexological interview. Although efforts have been made to improve interviewing procedures (Catania et al., 1996; Catania, Binson, Peterson, & Canchola, 1997) or reduce the interviewer's role (Turner, Danella, & Rogers, 1995; Turner, Miller, & Rogers, 1997), methodological studies in this area have not typically been guided by well-specified models. The present paper describes theoretical concepts and relationships that may be helpful for guiding research on the respondent-interviewer exchange. The proposed model focuses on self-presentation issues and their antecedents in the interview exchange, as opposed to memory concerns (Croyle & Loftus, 1993; Groves, 1989; Schuman & Presser, 1981; Tanur, 1992), or social exchanges that precede the interview (e.g., recruitment exchanges: see Groves, Cialdini, & Couper 1992). The present work is meant to be illustrative and integrative rather than comprehensive. More detailed reviews can be found elsewhere (see Bancroft, 1997; Catania et al., 1990, 1995; Groves et al., 1992; Ostrow & Kessler, 1993).

A synthesis of prior models yields two key variables that are hypothesized to directly impact people's tendencies to distort self-presentation in the interview: threat to self-esteem and emotional distress (or related emotions, e.g., guilt, embarrassment). Possible antecedents, hypothesized to directly or indirectly influence threat to self-esteem and emotional distress, are discussed in terms of four general factors: Respondent, Interviewer, Task, and Contextual.

THEORETICAL CONSIDERATIONS

Research on self-presentation or self-disclosure issues in the sexological interview has been loosely guided by three models and their derivatives: social desirability, threat, and self-disclosure (Catania et al., 1996; Meston, Heiman, Trapnell, & Paulhus, 1998; Sudman & Bradburn, 1982). By self-disclosure, investigators typically mean what the person says about him/herself. Self-presentation is how people present their personal information so as to create an image of themselves for others. Both concepts converge in assuming some type of true-self. My discussion, and the proposed model, assume that honest self-disclosure or accurate self-presentation overlap with what may be termed the true-self or "an honest presentation." The true-self is considered here to be a relative concept. That is, people may be doing their best to answer truthfully, but may still be in error because of recall difficulties, or because of psychological coping mechanisms and emotional experiences that inhibit full access to the past or unintentionally distort presentation. These types of distortions I would consider to be nonvolitional or occurring at a level of cognitive processing that is beyond easy and immediate control or access (e.g., self-deception: see DePaulo, 1992; Paulhus, 1984). The proposed model is relevant to volitional obfuscations of the truth. Unless stated otherwise, I use the terms honest self-disclosure and accurate self-presentation interchangeably in referring to volitional representations of the truth.

Social Desirability Model

Giving socially desirable responses is one form of self-presentation bias that influences response error (Galbraith, Strauss, Jordan-Viola, & Cross, 1974; Meston et al., 1998). The underlying assumption is that people distort their self-disclosures in order to maintain or enhance their social image. A two-component model of social desirability has been proposed (see Paulhus, 1984; Meston et al., 1998) composed of self-deception and impression management components. Self-deception is considered a personality construct that is a product of other tendencies, such as optimism, that cause a person to somewhat involuntarily shade their self-presentation. Impression management, which is susceptible to situational demands, concerns self-presentations that are deemed socially acceptable or make the person look good. Recently, Meston et al. (1998) examined this 2-factor social desirability scale. After controlling for other personality factors that might explain the association (e.g., conservatism), they found that only impression management was significantly associated with reports of sexual behavior. Further, impression management was significantly correlated with some intrapersonal sexual reports, but not interpersonal activities (reports made under anonymous conditions).

These findings are at variance with experimental studies showing that simple procedural changes in either interviewer-administered or self-administered surveys (Catania et al., 1996; Turner et al., 1997) can substantially modify self-reported interpersonal sexual behaviors in directions consistent with self-presentation model hypotheses. It is possible, however, that the underlying mechanism in these procedural experiments is not impression management but something else, such as avoiding an emotionally uncomfortable topic (i.e., the operating mechanism is, "I'd rather not talk about it," rather than, "I want to look good"). However, Agnew and Loving (1998) found that impression management scores correlated positively with condom attitudes, a variable that involves interpersonal sexual behavior. Further, they found that assurances of anonymity reduced the effects of impression management on condom attitude scores, particularly for males. These types of studies suggest a broader role for impression management across a wider array of sexual reports.

Current social desirability model formulations have limitations, however. For instance, although the social desirability model has an intuitive appeal, the direction of the bias predicted by the model is not specified. "Looking good" could mean both over- and underreporting of sexual behaviors. General social or procedural antecedents that enhance or reduce accurate self-presentations also have not been specified in great detail. Social norms about appropriate behavior and related social stigmatizations may stimulate self-presentation bias (see Catania et al., 1996). General social norms, however, may have little impact unless they coincide with reference group norms. Reference group norms may account for reasons why some people are ashamed of their sexual behavior, while others feel a need to brag about nonexistent activities. Under what conditions do reference group norms operate'? For people with multiple references groups, which set of norms predominates in the interview situation? In terms of interview procedures, very little work has been directed at specifying what conditions produce optimal accuracy in self-presentation. Reducing the social factor in the interview has been an obvious and more often researched antecedent (e.g. Agnew & Loving, 1998; Catania et al., 1990; Catania, McDermott, & Pollack, 1986; Meston et al., 1998; Millstein & Irwin, 1983; Turner et al., 1998). However, the effects on impression management tactics of other primary features of the sexological interview are not represented as well-articulated hypotheses in the social desirability formulations to date, such as how impression management is influenced by respondent role expectations, respondent's concerns about confidentiality, and interviewer strategies. For a recent exception see Agnew and Loving (1998).

Threat/Distress Model

The threat/distress model posits that some types of questions, when posed in a public context, stimulate emotional distress or threats to self-esteem in respondents. In an effort to avoid these negative circumstances, respondents will distort their self-disclosures (Catania et al., 1996; (Sudman & Bradburn, 1982). Some questions may elicit highly negative feelings such as embarrassment, guilt, or tear, that lead to distortions in response to avoid public disclosure. Studies using measures that assess discomfort with sexual questions have confirmed that people vary with respect to what kinds of sexual questions and situations make them most uncomfortable, and that the degree of discomfort is correlated with nonresponse and willingness to participate in face-to-face sexological interviews (Catania et al., 1986; Herold & Way, 1988; Snell, Belk, Papini, & Clark, 1989). Reducing threat and distress by making respondents feel more comfortable is hypothesized to facilitate honest or less distorted self-disclosures (Catania et al., 1996).

People may experience threats to their self-esteem or give negative emotional responses to sexual questions for a wide variety of reasons including, for example, learning experiences that teach that sex is bad and disgusting, coercive and abusive sexual experiences, and a simple lack of understanding that one is not unusual in one's sexual practices, beliefs, and problems. In this latter regard, Sudman and Bradburn (1982) have suggested that norm setting--questions that present respondents with the possibility that more extreme behavior may be more normative than they think--will reduce concerns about being unusual. Catania et al. (1996) tested this approach in an experimental field survey and found that norm setting increased reports of extramarital sex and sexual problems relative to more standard types of questions. Norm setting, however, is not always a workable solution, such as when asking respondents about antisocial behaviors. One alternative is to have the interviewer acknowledge to the respondent that the question may be difficult for some people to answer (e.g., questions regarding sexual abuse: see Catania et al., 1996). Understanding the conditions under which emotionally supportive types of questioning work or tail is needed (e.g., Catania et al., 1996, found that supportive wording increases reports of sexual abuse, but only for men). Research is needed on other interview procedural variations that might be useful in reducing threat and distress.

Self-Disclosure Model

Self-disclosure is at the heart of the interviewer-respondent exchange, and research in this area has a long and fruitful history. However, most self-disclosure research has been conducted with friends, couples, and agemate strangers in laboratory settings, or with patients in therapy (e.g., Cozby, 1973; Dindia & Allen 1992; Dindia, Fitzpatrick, & Kenny, 1997; Harris & Busby, 1998; Shaffer, Pegalis, & Cornell, 1988). Although several studies of self-disclosure in the sexual context have been published, these studies have not typically concerned methodological issues in the interview situation (Catania et al., 1986; Herold & Way, 1988; Snell et al., 1989). One problem is that self-disclosure research has typically been guided by equity or social penetration theory considerations. These perspectives concern reciprocity in self-disclosure, and expectations about or the actual effects of future interactions with another person on self-disclosure (Dindia et al., 1997; Tolstedt & Stokes, 1984). However, unlike most social exchanges the interviewing situation is asymmetrical with respect to exchange of information. Thus, many of the standard rules for social discourse are controlled for or do not apply. For instance, self-disclosure research suggests that people try to maintain equity in social exchanges through reciprocity with respect to type and depth of material disclosed. However, most researchers prohibit interviewers from self-disclosing to respondents, for fear that their disclosures will shape or bias the respondents' answers. Furthermore, expectations about future interactions with the interviewer are typically minimal. In short, very little of the current body of self-disclosure research speaks directly to the survey interview context.

The primary focus, to date, has been on the assumption that self-disclosure will increase when respondents are emotionally comfortable (e.g., Catania et al., 1996, 1997). A variety of studies suggest that emotional comfort does elicit more self-disclosing (Catania et al., 1986; Dindia & Allen, 1992; Gebhard, 1972; Miller, Berg, & Archer, 1983); however, self-disclosure can also be emotionally comforting (Pennebaker, 1995). Emotional comfort may be influenced by perceived similarities between the respondent and the interviewer, the nature of the question, and interpersonal control issues (Catania et al., 1996, 1997). However, past studies have not directly tested the linkage between interview procedures and the comfort that presumably influences disclosure. Other antecedents of self-disclosure might be considered from the literature, but, as noted, very little of the self-disclosure research has actually concerned the interview situation.

Control-Disclosure Model

A variant on the self-disclosure model is the control-disclosure model (Catania et al., 1996). Personal and interpersonal control issues may be particularly relevant to the interview-disclosure process. Historically, respondents have not had much control over the interview situation, other than to refuse to participate. Control, however, may be very important when respondents feel threatened or distressed. A general tenet of control theory is that in ambiguous situations that are potentially harmful, things that enhance a person's perceptions of being in control of the situation will reduce distress (Archer, 1979; Lefcourt, 1982). In particular, situations perceived to have unpredictable negative outcomes have a high probability of increasing distress. An interview situation has ambiguous and threatening facets that may elicit uncertainty about what will happen or what questions will be asked, as well as anxiety about the credibility of the interviewer and the confidentiality of one's responses. Some respondents' concerns probably will be high when the survey topic covers sensitive issues like sexuality. Although it is possible for respondents to control the situation by terminating the interview after they have started, this does not frequently occur. For people who stay in an interview even after they have become distressed, the Control-Disclosure Model predicts that procedures that increase a respondent's perceptions of control will enhance comfort and, consequently, facilitate honest self-disclosure. For instance, giving respondents the opportunity to select the gender of their interviewer (Catania et al., 1996; 1997) or to control the interview process through self-administered computer systems (e.g., Tourangeau & Smith, 1996; Turner, Ku, Sonenstein, & Pleck 1996; Turner et al., 1997) have all been found to improve responding (i.e., increased reports of presumably stigmatized behaviors such as same gender sex), and the underlying mechanism may well involve enhancing respondent control over the interview situation. Past work has not directly examined this mechanism; thus, the model needs to be more fully tested.

Model Overlap and Synthesis

These four models have significant points of overlap at the social-cognitive and emotional levels. At the social-cognitive level, social desirability models concern avoiding looking bad, threat models involve avoiding or guarding against revealing information that will result in socially uncomfortable situations or negative responses from the other person, and the self-disclosure models consider comfort with revealing your true-self to others. Thus, all the models concern self-disclosure about revealing one's true self to some degree, which, under the appropriate conditions, may be conceptualized in terms of threat to self-esteem. That is, avoiding looking bad and avoiding self-disclosures that leave one vulnerable (to social criticism, scorn, censure) are self-esteem issues. Self-esteem conceptually crosscuts the social-cognitive-emotional elements of all three models. Furthermore, all the models contain similar emotional elements: (a) the need to avoid looking bad to someone else, which implies that looking bad might feel bad to the person; (b) the need to avoid or guard against negative emotions (threat/distress); and (c) feeling uncomfortable because being too self-revealing may make one vulnerable to criticism or scorn. Thus, avoiding or guarding against negative emotions is another element common to all four models.

A self-esteem derivative. The role of self-esteem in behavior (including self-disclosure) is guided by the principle that activities or behaviors implying a high sense of personal competence are valued because they enhance or maintain self-esteem (e.g., Bandura, 1986; Fisher, Nadler, & Whitcher-Alagna, 1982; Stotland & Canon, 1972; Tessler & Schwartz, 1972; White, 1959). Thus, in the interview situation people will manage their disclosures in a fashion that they expect will enhance or maintain their self-esteem. In essence, this is social impression management guided by the dictates of self-esteem (Jones & Pittman, 1982). Self-esteem hypotheses are testable in the context of interview procedure experiments: for example, by examining the effects of threat-reducing item wording on responses to sexual questions by low and high self-esteem respondents.

Self-esteem and emotion. Preserving one's self-image may not, however, always be equivalent to avoiding emotional distress. Childhood sexual abuse, for instance, may result in painful memories and a loss of self-worth (Finkelhor, 1987). Thus, distress and self-esteem may co-occur and, therefore, may be correlated because they have a common set of antecedents, although they are not causally related. Correspondingly, I believe that efforts to model the self-disclosure process in the interview situation require consideration of both self- and emotion-based concepts. By following this cognitive-emotion pathway, the door is opened for inclusion of other conceptual frameworks that hopefully will result in a more eloquent model. For instance, some memories of negative sexual experiences may be extremely painful emotionally and require particular types of coping skills in order to report in an interview situation. Respondent's coping styles may play a key role in determining the form of the bias (e.g., refusing to answer, denial of the behavior vs. distortion of answers through minimizing occurrences). These theoretical conditions go beyond the scope of the present paper, but are presented as suggestions for further empirical and theoretical work.

A PROPOSED MODEL: OVERVIEW

In general, the focus of the proposed model is to articulate conditions that influence honest self-disclosures by their direct or indirect effects on threat to self-esteem and/or emotional distress. These conditions include task, respondent, interviewer, and contextual factors that either increase or reduce threats to self or distress. These factors are assumed to interact, but the extent of this interaction is a matter of further empirical work. The model also assumes that the interviewer and the respondent have a dynamic, asymmetric, but nonetheless bidirectional relationship of limited duration. However brief, this relationship is contained within broader contextual frameworks, including the larger research context (e.g., recruitment procedures, informed consent, location of the research), a social structural context (reflected in gender, ethnicity, and social class differences), and historical and developmental contexts. Contextual factors, in turn, may modify or moderate the impact of task variables (e.g., the topic, question order, and wording), and interviewer and respondent variables (e.g., motivations, emotions, and social skills) on emotional reactions and threat to self-esteem and, ultimately, on the quality (depth and detail, accuracy and honesty) of the respondent's self-disclosure and self-presentation.

Historical and Developmental Context

Although prior methodological research has examined proximal influences on the interview process, there are larger societal and historical contexts that may influence the more general research context and influence the conduct of respondents and interviewers in the interview. For instance, in the area of survey research, people in today's society are inundated with increasingly larger numbers of surveys both commercial and scientific (see Groves et al., 1992). This period effect may produce reluctant participants, which may make for difficult interviews even under the best of conditions. Age-cohort differences also effect research experiences. For instance, older adults are typically less likely than younger adults to have participated in research interviews. Understanding how broad differences in experiences with research interviews influence people's motives for participation and for answering honestly is an interesting research question. It is important to develop methods that rectify the negative experiences that some people have had in participating in research interviews. Increased intolerance of standard research practices may account for declines in survey cooperation rates over the last 40 years (Groves, 1989). The success of our work in the future depends on having a population of people who feel good about scientific research and about contributing to the general welfare of human beings.

Developmental issues are also important considerations in how people react to sexological interviews. Cohort differences, as well as age-graded phenomena (e.g., stage of sexual development), may influence the degree of threat to self and distress that respondents experience in the sexological interview in complex ways. Stage of sexual development may be relevant to response bias in regard to the respondent's developmental stage or the developmental time frame of the question. In this latter context, recent studies have found that response biases sometimes differ depending on whether one is asking adult respondents about adult sexual behavior or adolescent sexual behavior. Catania et al. (1996, 1997) conducted two national level probability survey experiments (general population and African American samples; age 18-49 yrs.). They found, with respect to age of coital onset during adolescence, that supportive wording manipulations were associated with an increase in the percentage of men who reported being a virgin during adolescence and, conversely, a significant decrease in the percentage of women who reported being a virgin during adolescence. This pattern of findings was not observed for reports of numbers of partners in adulthood. Catania et al. (1996) found that more supportively worded items were associated with decreased reports of numbers of adult sexual partners for both men and women. Consequently, particular types of survey topics may elicit different forms of response bias, depending on the developmental importance of the behaviors and gender of the respondent. Apparently, self image concerns in adolescence carry forward into adulthood, so that even in the interviewer-respondent relationship the respondent still tries to maintain a semblance of the image that he/she wanted to project as an adolescent (i.e., in terms of males being sexual or females being nonsexual: see Meston et al., 1998). In general, sexual developmental processes may help structure the personal meaning that the question has for the respondent and, therefore, the degree of potential threat/distress the item evokes.

Socio-Structural Context

Respondents and interviewers exist in a broad social structure stratified by age, ethnicity, class, gender, and religion. This social structure provides another context for the interview, and may modify other contextual factors. Social structural variables have been examined in survey methods research (Groves, 1989) with respect to (a) variations between respondent and interviewer, (b) respondent's demographic characteristics relative to variations in task variables, and (c) historical factors (e.g., age cohort differences in response bias). In this area of research, demographics typically represent cues to the respondent as to what the interviewer is like as a person. Demographics may also represent proxies for underlying social and psychological conditions of the respondent that are true causes of the self-presentation variation. As proxies for other causative factors, the demographic variables considered in this section may be decomposed into other elements by future investigations, but for the moment the proxy measures are better represented in the literature. These proxy studies, however, are important in providing sign posts for future research into underlying social psychological mechanisms, and upon further investigation may be subsumed under other conceptual categories, such as developmental context.

Age. Respondents' age may be important as a mediator of self-presentation bias, either because individuals herald from different historical contexts (age-cohort phenomena) in terms of societal comfort with sexuality or because of developmental experiences (age-graded phenomena) that influence comfort with sexual topics. Surprisingly, there has been little research on age-related biases in sexological surveys, cohort or graded. Peterson and Catania (1997) provide recent data that suggest that older respondents, in their 60s and 70s, have less experience than younger respondents with both survey participation and in discussing sexual matters (see also Ostrow & Kessler, 1993) and, therefore, may be more threatened by sexual questions. That is, there may be an age-cohort effect evidenced in survey and interpersonal sexual experiences. Older respondents would be expected to come to the interview situation with a different set of motives, expectations, and emotional comfort than would respondents in their 20s. Whether the motives and expectations are qualitatively different is a matter for research, but a conservative expectation would be that the differences are a matter of degree rather than a matter of the type.

Gender. Gender also impacts self-presentation, as men and women have been found to have opposite responses to the same methodological stimuli (Catania et al., 1995, 1996; see also Abramson & Handschumacher, 1978). As indicated previously, supportive wording manipulations are associated with an increase in the percentage of men who report being a virgin during adolescence, but a decrease in the percentage of women who report being a virgin during adolescence. The results conform to expectations about response bias patterns that might arise for men wishing to demonstrate their virility in sexual experiences and for women tending to downplay their sexual experiences so as not to seem sexually promiscuous. This particular pattern of findings may reflect the influence of a sexual double standard in the area of response bias. That is, with standard, less supportive questions the form of the bias for reports of adolescent sexual activity conforms to the view that sex is okay for men but not for women. Questions that are more supportive appear to break down the need to adhere to a sexual double standard.

The issue of gender differences in response to methodological procedure was explored further in a series of analyses conducted using the recently completed National Survey of Sexual Health (unpublished report to the National Center for Health Statistics; available from the author). In general, men and women did not differ significantly in terms of dropout rates or overall nonresponse to survey questions. However, upon closer inspection distinct gender differences in patterns of item nonresponse emerged within particular families of items. Among questions on attitudes towards condoms and first sexual experiences, women were significantly more likely than men to not respond. The condom attitude and early sexual experience items have overlapping themes relating in a direct way to genitalia or things that come in contact with the genitalia. Thus, it is possible that women are more threatened or distressed than men are by questions that concern direct experiences with genitalia and specific sexual activities. In some regards, this fits the more general observation that men, more so than women, tend to be genital focused in their sexuality while women are more relationship focused (Meston et al., 1998). Interview items with a genital focus may elicit higher levels of nonresponse for female compared to male respondents. Historical context may be a modifier of the gender role-response bias relationship. I would expect that the gender role requirements for males and females have changed over the years and are quantitatively, if not qualitatively, different for men and women in their 20s today than they were for men and women in their 20s sixty years ago. These speculations underscore the need to examine the interplay between contextual factors.

Race, religion, and social class. Ethnic, religious, and social class differences in self-presentation have been speculated on (Catania et al., 1990, 1995), but systematic investigations into the importance of these variables have been scant. Recently, Catania et al. (1997) examined differences between Whites and African Americans in how their question responses varied as a function of randomized experimental variations in interview procedures and item wording using phone interviews. The findings suggest that gender is a much more important issue than ethnicity, although ethnic differences are evident in several circumstances. Item wording that was supportive of more deviant responses (vs. standard wording) produces significantly more reports of not using condoms and teen sexual activity by African American men relative to White men. The effects of interviewer gender were not differentially evident by ethnicity, and effects were similar for White women and African American women. Investigations are underway to examine social class as a modifier of wording and interviewer gender.

Research Context

Sexological research interviews typically occur in research settings on college campuses or in the person's home (community-based studies). Despite obvious differences, both settings involve a recruitment process, an interview setting, and an interview mode (usually face-to-face in University studies, but not always). Community-based studies are probably more complex. For community-based household studies, the research context begins with the initial household contact (who may or may not be with the actual respondent). This initial contact involves explaining the purpose of the study, providing information that testifies to the credibility of the interviewer and importance of the study (e.g., interviewer identification, supporting letters), enumerating the household to determine eligibility, and making a random selection of a respondent from among all eligible persons in the household. Respondents may be offered incentives to participate, and those who initially refuse may be recontacted at a later date to determine if they may have changed their mind. This entire process may take place with others present (just as the interview might) or in privacy, and a variety of modes may be used to collect the data (e.g., telephone, computer, paper and pencil). At present, the effects of these research contextual parameters on sexological interviews have not been extensively studied with respect to self-presentation in the interview. However, these are important stepping stones for building trust and liking of the interviewer, and confidence in the belief that what is disclosed will be kept confidential. Uncertainties at this juncture, even among those who become participants, may expand when the respondent is faced with questions that they perceive to be highly sensitive. For campus-based studies, the recruitment process differs and the general academic institutional context surrounding the interview (norms around pursuit of intellectual activity, pursuit of new experiences) may offer additional and different contextual factors to consider than the community-based study.

Recruitment. Groves et al. (1992) proposed a model for studying participation bias, which is relevant to much of sex research. As past work indicates, the bulk of nonparticipants in general population surveys fail to participate long before they know the study is about human sexuality, which suggests that most nonparticipation in survey studies is due to general proclivities not to participate in research rather than sex surveys per se (Catania et al., 1995). This type of nonparticipation is clearly different from that experienced in studies obtaining psychophysical measures (Catania et al., 1990), and models specific to those contexts need to be developed.

Volunteerism and refusal. Respondents' motives for volunteering to participate may have a substantial, but as yet unspecified, effect on perceived threat and distress and, subsequently, self-presentation in the interview. Respondents may come to the interview with a variety of motives. Motives for volunteering may reflect altruism, curiosity, help-seeking, financial need, or a need for catharsis (Groves et al., 1992). A respondent who may be seeking help with a sexual problem, for instance, may be substantially more likely to reveal details of sexual problems than respondents who are motivated to participate solely out of curiosity. Although motives to participate may carry over into the interview situation, the motives may also change over the course of the interview as respondents become more fully aware of the topics of the questions. For instance, a respondent may begin the interview out of curiosity, but shift to a help-seeking motive if the interview delves into a topic area for which the respondent would like some help. Motivation also has an effect on the respondent's willingness to stay with the interview and provide detailed answers. How different types of motives for participation effect the degree of self-disclosure is not known. Motives for refusal are also pertinent to consider. Some respondents may initially refuse to participate and, upon recontact, decide to participate. Do the initial motives for refusing impact subsequent disclosure? This remains another unexplored area for research.

Incentives. Equity theory suggests that social exchanges are guided in part by expectations for reciprocity. As noted earlier, researchers avoid procedures that would allow interviewers to reciprocate in kind. Instead, incentives are sometimes used to create a balanced exchange. Whether incentives actually produce an increase in honest self-disclosures has not been established. Respondents may feel a need to disclose information when they receive incentives, but there is no guarantee that these will be truthful reports.

Mode. The interview mode also provides a general contextual factor that has been thought to influence self-disclosure of sensitive information (for reviews see Catania et al., 1990, 1995). The evidence to date suggests that modes that increase privacy also increase honest self-disclosures. Self-administered formats (paper and pencil, computer driven), interviewer formats (telephone, face-to-face), and their variants have not been sufficiently studied with respect to specific sexual topics to have a clear picture as to when mode is important and when it is not. Further, other mechanisms may be operating to influence the type of mode effect. For example, bragging behavior by teenagers may be reduced somewhat by modes that have an observer component: that is, when the respondents do not believe they will be caught, they can exaggerate with abandon. In addition, different modes provide respondents with varying amounts of control over the interview process, and increased control may impact self-presentation bias in sexological interviews (Catania et al., 1996). Thus, further work may be needed to specify the underlying mechanisms that explain the effects of interview mode on honest self-disclosure.

Interview Variables: Task Conditions

Task variables include the various elements of the interview proper, including the topic(s) to be studied, item wording, and related questions or total interview contextual matters that impact threat to self-esteem and emotional distress, either by affecting how hard the respondent has to work or by the emotional impact of the question(s) on respondents. Respondents who become distressed because they do not understand questions, because they are having difficulty providing answers, or because they are experiencing negative emotional reactions to questions may begin to alter their self-presentation in ways that reduce threat or distress (e.g., refusing to answer items, falsifying answers to avoid a long series of questions, providing more "don't know" responses or "zero" responses).

Topic features. Interview topics can be thought of as both isolated units as well as a complex interplay between a succession of topics that form a broader interview-topic context. Research to date has focused primarily on respondents' reactions to specific topics rather than the cumulative effects produced by multiple topics or topic order (for review see Catania et al., 1990, 1995). In general, the interview topic may concern a broad array of sexual topics, or it may concern a more specific line of inquiry which determine the goals of the interview exchange. The interview topic may play a key role in determining how important interviewer variables are to the respondent's decision to disclose honestly. Within the interview, topic shifts may cause respondents' emotional states, disclosure tendencies, and motives for continuing with the interview to change. The degree of topic change over the course of an interview and its influence on self-presentation is an important area of inquiry.

Topic sensitivity. Question threat or sensitivity is a central stimulus feature of sexological interviews (Catania et al., 1986; Sudman & Bradburn, 1982). Although question sensitivity is a matter of individual perception, some topic areas are generally perceived as threatening and may be described as socially stigmatized behaviors or events (e.g., abortion, sexual dysfunctions). Whether the degree of perceived sensitivity is determined by personal standards or societal norms regulating deviant behavior, topic sensitivity varies both within and across individuals. Indeed, if all survey topic areas are arrayed along a sensitivity dimension, sexuality questions, like mental health questions, fall near the extreme end of the sensitivity distribution. However, it is important to note that questions on income often elicit the highest levels of nonresponse. Anywhere from 10 to 20 percent of respondents may fail to complete income questions. Only condom attitude questions, particularly those related to hands-on experience with condoms, and questions on masturbation attain correspondingly high levels of nonreporting (see Catania et al., 1990; Peterson & Catania, 1997).

Experimental studies of the issue have provided evidence that sexual attitudes and behavior do vary in the degree of discomfort they stimulate in respondents (Catania, 1997; Catania et al., 1996, 1997; Peterson & Catania, 1997). In general, questions on miscarriage, masturbation, abortion, sexual dysfunction, extramarital sex, and rape are questions that elicit moderate to high levels of discomfort among respondents (Catania et al., 1996, 1997). Questions on same gender sex and anal intercourse, however, elicit extremely high levels of discomfort (Catania, 1997). For any given topic, sensitivity has also been found to vary as a function of the assessment period (more recent activities elicit more threat than distant events), characteristics of the respondent (gender, age, ethnicity), interviewer characteristics (i.e., gender), item wording, and opportunities for the respondent to control the interview process (Catania et al., 1990, 1996; Peterson & Catania, 1997). In brief, the sensitivity of the topic may vary as a function of the social stigma attached to the behavior or belief being assessed, and these topics form a central element in the survey context, which defines the nature of the interviewer-respondent exchange. The degree of impact of these stigmatized topics on respondents' answers may vary as a function of respondent, interviewer, and other contextual factors.

Item wording and emotional reactions. How the question topic is perceived may be most immediately influenced by the surrounding linguistic context. The words that are used to ask questions and make up the response categories influence the respondent and the interviewer. They stimulate emotional reactions in both individuals, reactions that may be communicated to each other through voice or nonverbal behavior. Recent studies suggest that this is the case (Catania et al., 1996), but direct assessments of respondent emotional reactions are lacking. Further, aside from simple perceived sensitivity ratings, investigators know little of the range of emotional reactions that questions stimulate. Understanding the different kinds of emotions questions elicit may aid in determining wording or procedures to help respondents better cope with those reactions.

Item comprehension. Another fundamental task factor to consider is whether respondents understand and comprehend the words and concepts that make up the question. Respondents who have difficulty understanding what is being asked of them are more likely to become distressed with the interview and may begin to slant answers to bring an earlier end to the process. Respondents may also be threatened by their lack of understanding, and may provide bogus answers as an attempt to shield their comprehension problem from the interviewer. Thus, comprehension problems can lead to self-presentation biases motivated by efforts to cover up the lack of comprehension. Comprehension issues in interviews on sexual behavior have not been widely examined. However, numerous small scale qualitative approaches have been used to understand whether people comprehend terms that describe different kinds of sexual behaviors (Catania et al., 1995; Sudman, Bradburn, & Schwarz, 1996). This work has shown that most people understand terms such as vaginal or anal intercourse. In a national probability sample of adults, Binson and Catania (1998), however, have shown that a sizable segment of individuals of low socioeconomic status (19-25%) report having difficulty understanding the term anal intercourse. Thus, small scale, qualitative comprehension studies (e.g., focus groups) are useful, but may understate the magnitude of the problem for some population segments. In addition to language issues, difficulties for respondents also may arise when investigators examine more complex topics that employ concepts that are beyond the respondents' observational powers or cognitive schema.

Interview Variables: The Interviewer

The interviewer may impact perceived threat and emotional distress by the respondent in a variety of ways. Interviewer effects may occur because of (a) how the exchange is influenced by cues associated with social structural or demographic characteristics of the interviewer, (b) how interviewers administer the interview, and (c) how interviewers help respondents with difficult items. Groves (1989) found that methodological research has generally provided little empirical documentation on the affects of interviewer characteristics, interviewer training guidelines, supervision of the interviewing process, and evaluation of procedures for interviewers on sell-presentation. The situation as pertains to sexological interviews is not substantially different today (Catania et al., 1995).

Interviewer's role. The interviewer has, historically, been the active agent in the research interview process, and the respondent's behavior is partly a reaction to the interviewer. Interviewers are typically trained to be assertive and persistent in a polite and considerate manner. The interviewer's role helps balance these somewhat disparate characteristics. This role is defined by the investigator within the ethical guidelines formulated by various governmental organizations. How the general public perceives this role is not well understood, but the common expectation is that the ambiguity of the situation for the respondent will allow the interviewer to define appropriate behavioral expectations for the respondent. The interviewer may need to overcome considerable resistance in some types of studies because the interviewer violates polite customs regarding behavior by a stranger in seeking entry to the respondent's home, wanting an hour or more of time, and asking questions that, in other contexts, would be considered highly personal or even rude, particularly when the questions are about sex. In this general context, the interviewer's role behavior is geared toward helping to guide the respondent towards evolving the respondent role, putting the respondent at ease, and being nonjudgmental without shaping the respondent's answers. These crucial influences on threat, distress, and, therefore, self-presentation, have been little studied.

Interviewer training. In general, methodological studies have shown that training, supervision, and feedback in the form of audio/video recordings create better interviewers (see Groves, 1989). These general observations are probably no less applicable to sexological interviews, but other significant training issues have not been studied. For instance, although most sex researchers would agree that interviewers need to be comfortable with the language used in sex surveys, there are no established outcome measures to assess the degree to which interviewers are comfortable with sexual terms. Such measures would be useful for comparing procedures designed to reduce interviewer discomfort with sexual terms. Interviewers, even when well trained, come to the interview with a variety of skills and predispositions about the topics they'll be asking about, as well as expectations about answers that may be given. A common example in the area of human sexuality interviews is the emphasis placed on training interviewers to quiet their nonverbal feedback to respondents when using sexual words not typically used in day-to-day interpersonal communication. If the interviewers blush or show embarrassment in some other way, this may cue a respondent to the idea that the interviewer actually thinks something is wrong with these behaviors. Such an interpretation could lead the respondent to inhibit their answers. Calm, comfortable, nonjudgmental cues enhance topic acceptance and calmness on the part of the respondent.

Administering the interview. Interviewers perform somewhere between two extremes: (a) administering the interview under strict, nonvarying conditions that require all questions to be asked in a standard manner by all interviewers; and (b) administering the interview under conditions that allow for probing, rephrasing, and adapting questions to individuals. Research using standardized measures views adherence to the script as essential. Extensive rephrasing and rewording is discouraged. Recent studies question this long-held assertion, particularly in its extreme form. Schober and Conrad (1997), for instance, conducted an experiment in which respondents received either a standardized interview or a more conversational style interview. The investigators report a substantial decrease in reporting error with the conversational style interview, particularly when the respondent's experience does not map precisely onto the question/response categories. Conversational style may also help put some respondents at ease and reduce self-presentation bias, although this particular issue has not been studied.

Groves et al. (1992) identified two other general interviewer issues that seem to be useful in eliciting high levels of response from respondents. The first is interviewer tailoring, which involves the use of different dress styles, physical behaviors, words, and persuasion strategies for different types of respondents. The second issue involves strategies that reduce the likelihood of the respondent prematurely terminating the discussion. These two general strategies are important when getting respondents to participate in research, and are helpful in augmenting the degree of perceived similarity and liking the respondents have for the interviewers, thereby facilitating reduction in distress or threat. At present, it is unclear if there are tailoring or maintaining procedures that are particularly salient for reducing threat or distress in sexological interviews.

Emotion management. In addition to experiences and training that enhance skills and shape appropriate expectations, interviewers need to learn to manage the emotional content of the interview. The respondent may become embarrassed or get upset, and the interviewer needs to know how to respond to these emotional reactions and how to cope with their own feelings so as not to bias the respondent's answers. Studies have been conducted on interviewing skills and persuasion strategies used in a variety of interview contexts (e.g., Groves et al., 1992), but very little research has been conducted on these issues in the context of sex interviews. The issue of emotion management skills is, to my knowledge, completely untapped, and is an area of inquiry that deserves considerable attention. To aid the interviewer, investigators sometimes construct the wording of a question so that when the words are spoken by an interviewer the words and voice reflect emotional comfort with the topic. Examples are given by the work of Blair, Sudman, Bradburn, and Stocking (1977) and Catania et al. (1996), which show that asking questions on sexual behavior in more emotionally supportive ways increases reports of less socially acceptable events, situations, or conditions. Interviewers also need to be skilled at providing information that justifies the line of inquiry that is being followed. At the same time, they need to be skilled at authenticating the interview. Respondents will be more comfortable and, therefore, more disclosing, when they believe that the sexuality survey is being conducted for a good purpose and that the data are being collected by ethical people who will honor the confidentiality of the interview. Little study has been done on the skills and information needed to provide respondents with adequate justification and authentication of the interview. In some instances, too much emphasis on confidentiality may actually make respondents more suspicious (Singer, Von Thurn, & Miller, 1995).

Demographic cues. Social structural variables may be considered in terms of proxies for underlying factors or as cues to the respondent as to what the interviewer is like as a person and, consequently, how the interviewer may relate to the respondent. Two conceptual formulations may help organize research in this area. First, Sudman has observed that interviewer demographic characteristics increase in their relevance to the interviewer-respondent relationship as a function of the research topic (Sudman & Bradburn, 1982). Consequently, topics that have a large interpersonal component, such as sexuality, would be expected to evoke a different and highly salient set of interpersonal issues than other potentially less emotionally laden topics (e.g., consumer habits). The respondent's interpretation of interviewer demographic characteristics may then be filtered through this set of interpersonal issues underlying the interview topic. Gender and perceived sexual orientation of the interviewer relative to the respondent, for example, might then be a highly salient feature in interviews concerning sexuality (Catania et al., 1996). Thus, female interviewers interviewing a heterosexual male respondent about sexuality may evoke a different type of self-presentation than when the interviewer is a gay male. Although this perspective provides insight into the interplay between topic and interview-respondent characteristics, it does not include consideration of how the respondent's demographic characteristics might determine or modify the impact of interviewer cues. For instance, how does the degree of similarity between respondents and interviewers influence self-disclosure? Studies in social psychology touch on this issue.

Similarity. Results suggesting that matching interviewers and respondents on gender elicit higher levels of reporting on many sensitive sexual activity questions (Catania et al., 1996), and results from studies on factors affecting liking when strangers meet (Byrne, 1961; Newcomb, 1961) support the proposition that perceived similarity may produce more liking. Thus, greater similarity between interviewer and respondents in terms of age, social class, gender, and beliefs is hypothesized to produce more liking and, consequently, more self-disclosure (Jourard, 1971). Past studies provide mixed support for this perspective. In these studies, male and female respondents tend to report more sexual information or make more sexual statements in general to female interviewers (Abramson & Handschumacher 1978; Catania et al., 1990; Darrow et al., 1986; DeLamater, 1974; DeLamater & MacCorquodale, 1975; Hansen & Schuldt, 1982; Johnson & DeLamater, 1976; Singer, Frankel, & Glassman, 1983). Catania et al. (1996) recently investigated the interviewer-gender issue in a study where respondents were randomized to male or female interviewer conditions, and found that same gender interviewers significantly decreased dropouts in the study. Catania et al. (1996) also found that same gender interviewers significantly increased responses of condom nonuse, extramarital sex, sexual problems, and sexual violence, but had no impact on reports for numbers of sexual partners or age of first coitus. These observations suggest that interviewer-respondent similarity is important for interviewing on some but not all sexual topics. Similarity principles may also influence the effects of interview topic. For example, when the topic is sexuality and the respondent is heterosexual, opposite gender interviewers may stimulate sexual motives for the respondent that distort the self-disclosure process. However, as important as similarity is, not all respondents are comfortable with a person who is like themselves.

Complimentarity. Complimentary characteristics between people may be important to the interview situation, just as they are to relationship formation in general. Some men (40-50%), for instance, are much more willing to select female interviewers than male interviewers when given a choice (Catania et al., 1996), but women overwhelmingly select female interviewers when provided a choice of interviewer gender. Thus, there are gender differences in the importance of similarity and complimentary conditions. One clue to why some men prefer opposite gender interviewers may come from consideration of early developmental experiences. One study, though methodologically flawed, suggests that men with a history of sexual abuse by males may, in fact, be more comfortable with female interviewers than male interviewers (Kaplan, Becker, & Tenke, 1991). In general, past studies do not clearly indicate when similarities or differences in respondent-interviewer characteristics are important to the degree of threat or distress respondents experience and respondents' subsequent response distortions. Many of the studies that looked specifically at the topic of sexuality and considered characteristics of the respondent and the interviewer were not based on systematic studies of these demographic characteristics where respondents are randomized to different types of interviewers. Most of these past studies are nonrandomized group comparisons, where an investigator at the end of the study does a comparison between the male and female interviewers (for review see Catania et al., 1990, 1995). These studies are interesting, but do not provide the necessary support lot a particular characteristic. Randomized experiments examining ethnic, age, and social class characteristics need to be conducted. More subtle characteristics may also come into play; for instance, Catania et al. (1992) reported the results of qualitative studies that suggest matching the accents of Hispanic interviewers and respondents is an important consideration.

Physical attractiveness. It is also important to consider how sexually attractive the interviewer is to the respondent and whether this factor comes into play in reducing threat or distress associated with questions on sexual behavior. One might expect that increased attraction to an interviewer could produce more liking and comfort and, consequently, more self-disclosure (Jourard, 1971: Walster, Aronson, Abrahams, & Rottman, 1966). However, it could also produce embarrassment and related self-presentation bias. A recent experiment (Harris & Busby, 1998) found that participants were more likely to disclose a sexual problem to attractive therapists than to unattractive therapists. Although not directly generalizable to the interview context, these findings suggest that attractiveness is an important consideration and studies more directly relevant to the interview situation need to be conducted.

Motives and emotions. In addition to the interviewer's skills, demographic characteristics, and physical attributes, there are other emotional and motivational factors that may be pertinent to the interviewing situation. The interviewer's motivations and expectations are relevant areas of inquiry. For example, when interviewers believe the interview questions are sensitive they tend to achieve higher missing data rates (Singer & Kohnke-Aguirre, 1979). In addition, the added skills of being able to manage the emotional content of the interview, manage one's own emotion, and manage the emotional issues that come up for the respondent, are all necessary and important issues to consider in interviewer training and research on interviewer-respondent relationships. Further, the general emotional state of the interviewer and the respondent at the time of administering the interview may also be important. An interviewer who is feeling depressed may communicate a very different tone to a respondent than an interviewer who seems alert and interested.

Personality variables. More general personality constructs may also be important to consider. For instance, Shaffer, Ruammake, & Pegalis (1990) examined a personality variable, termed opener, that reflects the ability to facilitate self-disclosures in others, and found that high opener interviewers were more successful at eliciting sell-disclosures from respondents, particularly if the respondent was a low opener: Thus, some general personality variables may be relevant for future study with respect to both interviewers and respondents. Our understanding of how relevant characteristics of the person influence the interview process is an unexplored area.

Respondent Factors

Respondents come to the interview situation with a complex assortment of motivations and interview experience that may influence their levels of perceived threat and distress, and, consequently, the honesty of their self-disclosures. In addition, respondents may have predispositions with regard to self-disclosure to strangers, particularly when the topic is sexuality. In the interview, respondents must cope with a potentially ambiguous situation that plies them with incentives, assurances of confidentiality, and statements on the scientific importance of the research. During the interview, respondents must confront a variety of emotional and time/effort barriers that ultimately influence self-presentation bias and their general willingness to continue. Like the interviewer, they also come to the interview with a complex set of demographic, personality, emotional, cognitive, and social skill characteristics that may influence their perceived threat to self, their emotional reactions, and, subsequently, their self-presentation. In many ways, research on these issues is less developed than work investigating interviewer variables.

Respondent role. Turner and Martin (1984) noted in an earlier, comprehensive review of the literature that there had been relatively little systematic research on how respondents interpret the research interview and their role in it. This situation remains substantially unchanged. Respondents often do not know what is expected of them: They do not understand the purpose of research or how they are to answer some types of questions. When the interview involves sexual questions, the respondent's anxiety about credibility of the interviewer, uses of the answers, and confidentiality may all be substantially heightened. Lacking clear guidelines or roles to guide appropriate behavior, respondents must cope with a situation of uncertainty. They may look to prior interview experiences, such as those related to employment, schooling, or even therapy, as guides for appropriate conduct, and they often may look to the interviewer for guidance. How these prior experiences and interviewer expectations shape self-presentation bias in sexological interviews is unknown. In addition, how do respondents perceive the interview context? Do respondents believe that the research is worthwhile? Do they believe that the information will be used for a good purpose and that their confidentiality will not be violated? Do respondents see the interviewer as comfortable with the topic? These positive perceptions may reduce role ambiguity, alleviate concerns, and subsequently decrease threat or emotional discomfort, thus eventuating in greater disclosure honesty.

Motives for participation. Respondents come to the interview with a variety of motives. Motives for participation may reflect altruism, curiosity, help-seeking, financial need, or a need for catharsis (Groves et al., 1992). It is probably reasonable to assume that some motives to participate carry over into the interview situation and influence perceived threat and distress. However, although they carry over, they may also change over the course of the interview as the respondent becomes more fully aware of the topics being discussed. Respondent motives may also need to be balanced against the time and energy necessary to complete the interview. Thus, altruism may turn into regret, reluctance, and refusal to continue if the task burden becomes too high. On a positive note, the asymmetric relationship between respondent and interviewer may become acceptable to the respondent when respondents have motivations (e.g., help-seeking, altruism) that may override the normal social rules governing appropriate behavior when two strangers meet. Do such motives also override threats to self-esteem and distress associated with sexual questions, and permit detailed disclosure in this atypical social exchange? What are the optimal conditions that give the respondent permission for breaking the normative rules for social conduct with a stranger in a fashion that reduces threat and distress while meeting the goals of the study?

Burden. The degree to which respondents perceive the interview as a burden is a consideration that may offset respondent motives that may otherwise keep the individual engaged, emotionally comfortable, and forthcoming throughout the interview process. Length of the interview, comprehension problems, complex questions, and assessment of biological measures may all add to the respondent's burden, that, in turn, accentuates perceived threat and distress, and leads to dropping out, increased "don't know" responses, and denial of behavior. In addition, the degree of perceived difficulty with the interview may be influenced by the respondent's health status, their mistrust of authority, and the perceived reciprocity in the interviewer-respondent relationship.

Emotional considerations. There are a variety of emotional considerations in sexological interviews that may impact the respondent's self-disclosures. There are emotional factors that may effect particular motives for participation, and emotional reactions that develop over the course of the interview related to question topic, perceptions of the interviewer, task burden, and so on. Thus, it may be relevant to consider how the mood state of the respondent changes over the course of the interview. For instance, some topics may elicit little emotional reaction from the respondent, while other topics may elicit strong feelings that impact subsequent topics and, thereby, make it more difficult for the respondent to stay focused or to switch to another area of inquiry. For instance, a person may be asked a series of questions about their current sexual behavior which appear on the surface to be relatively innocuous. They then may be asked about instances of sexual violence, abuse, or rape. The respondent who has had such experiences, but has not had an opportunity to deal with their emotional soliloquy, could find such questioning to be extremely disturbing. Responses to later questions might then be overshadowed or colored by the emotional associations of past sexual trauma. In addition, respondents may have affective disorders that color their self-disclosures over the entire interview in a pervasive manner. In an area that has focused almost exclusively on cognitive memory issues for respondents, investigation into emotional elements of the interview are somewhat overdue.

Impression management and perceptions of the interviewer. Impression management, as noted in earlier sections, is a central concept in understanding self-report bias in sex research. People may, for instance, want to please the interviewer, provide answers that will not be viewed as socially deviant, or avoid embarrassment over problems they are having with the interview. How such impression management tactics interact with other respondent or contextual factors (e.g., trust, reciprocity rules, need for catharsis) to elicit particular biases have not been investigated. Impression management tactics may also change over the course of the interview. For example, as the respondent becomes more comfortable with and trusting of the interviewer, their need to give socially desirable answers should decrease. The process of warming up the respondent so that they feel good about the interviewer typifies procedures that attempt to reduce impression management tactics that might bias answers. The type of warm-up and its content, length, and mode of delivery are all pertinent areas of inquiry in the context of sexological interviews. The process by which respondents balance conflicting motives and their emotional responses to questions, and at the same time gain a sense of the interviewer and their acceptance of the respondent's answers, might also be considered in terms of interpersonal impression management. Studies are needed on how the multiple motivational and emotional demands on the respondent help shape the sexual persona the respondent projects to the interviewer.

How the interviewer influences the respondent's impression management tactics is also an unexplored arena. From the person perception framework, the interviewer's behaviors carry social meanings for the respondent which affect how the respondent evaluates and categorizes the interviewer and interview situation. These evaluations and categorizations of the interviewer may then influence how the respondent behaves towards the interviewer and how the respondent views him/herself. Thus, person perception considerations may provide an important link to other factors. That is, particular judgments by the respondent of the interviewer and the interview may well stimulate biases that conform to the respondent's self-presentational style in the interview situation. For instance, a heterosexual Don Juan being interviewed by a beautiful female interviewer may perceive the act of asking sexual questions as hinting at a sexual invitation that, in turn, stimulates responses that are tailored to be seductive rather than truthful. Impression management, although understood to bias responses, has not been studied in the interview context in terms of specific tactics and influences on choice of tactics by the respondent. Investigators have not mapped impression management tactics in interview situations across age, ethnic, and social class groups, although evidence exists that such variation may indeed exist (see Catania et al., 1990, 1986; Franco, Malloy, & Gonzalez, 1984).

Self-disclosure tendencies and other personality characteristics. As a function of more general personality characteristics, respondents may have broader tendencies to behave in particular ways in the interview situation. For instance, self-disclosure is a central situational outcome in the proposed model, but the situational disclosures that may typify the interview context are presumably influenced by pre-interview self-disclosure experiences. Catania et al. (1986), for example, reported a significant, but nonredundant, association between general self-disclosure tendencies and a measure specific to disclosure of sexual information. Patterns of general disclosure may vary across individuals. Some individuals may have histories that represent a consistent pattern of disclosure tendencies across social situations, while others may show more situational specificity in what and in how much they reveal about themselves. Situationalism may be the hallmark of sell-disclosure histories. An obvious example is the difference that is often observed between what an adolescent may be willing to tell a teenage friend versus what they may be willing to tell their parents. How general disclosure tendencies are influenced by prior interviewing experiences is unknown, and is likely a salient factor for self-disclosure tendencies in future interview situations. Other personality characteristics may also be relevant to the interview situation. Self-monitoring, openers, Machiavellianism, and androgyny have all been found to influence self-disclosure behavior, albeit none of this research is specific to the sexual interview (e.g., Dingler-Duhon & Brown, 1987; Shaffer et al., 1990; Shaffer, Ogden, & Wu, 1987).

CONCLUSION

I have proposed a broad-based model of self-presentation/disclosure in the sexological interview. The model hypothesizes that contextual, task, interviewer, and various respondent variables influence honesty in sexological interviews via a set of final common pathways represented by self-esteem and distress/emotional considerations. By its nature and by virtue of a limited data base, the proposed model is open ended and preliminary. The focus of the proposed model is to articulate conditions that influence honest self-disclosures by their direct or indirect effects on threat to self and/or emotional distress. I have briefly summarized research on these antecedent conditions and suggested areas for further research.

In many regards, the breadth of research needed in this particular area is quite large. Nevertheless, methodological research on sexological interviews continues at a pace that is unprecedented historically. Research on mode effects, ways of combining different modes of collecting data, different ways of asking questions, interviewer variables, and respondent variables continues and has led to some very interesting and consistent results regarding the direction of biases (see Catania et al., 1995; Turner et al., 1997, 1998). Despite progress, there is a note of caution. As a pragmatic research enterprise, methodological studies are sometimes taken at face value, and at times the rush to implement new techniques may be premature. A recent example is an editorial by Bloom (1998), that proclaims all past research on sensitive topics invalid on the basis of a mode experiment conducted by Turner et al. (1998) which, at best, may be only generalizable to adolescent males. I believe that there should be a general conservative impulse in moving from methodological findings to substantive research. Before investigators begin to consider changing their time-honored methods, they should have on hand multiple replication studies conducted across different population segments, and research which rules out alternative explanations and documents underlying social and psychological mechanisms hypothesized to facilitate honest self-disclosures. To date, few, if any, methodological studies on the sexological research interview accomplish these goals.

REFERENCES

Abramson, P., & Handschumacher, I. (1978). Experimenter effects on responses to double-entendre words. Journal of Personality Assessment, 42, 592-596.

Agnew, C. R., & Loving, T. J. (1998). The role of social desirability in self-reported condom use attitudes and intentions. AIDS and Behavior, 2, 229-239.

Archer, R. (1979). Relationships between locus of control and anxiety. Journal of Personality Assessment, 43, 617-626.

Bancroft, J. (Ed.). (1997). Researching sexual behavior: Methodological issues. Bloomington, IN: Indiana University Press.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Binson, D., & Catania, J. A. (1998). Respondents' understanding of the words used in sexual behavior questions. Public Opinion Quarterly, 62, 190-208.

Blair, E., Sudman, S., Bradburn, N., & Stocking, C. (1977). How to ask questions about drinking and sex: Response effects in measuring consumer behavior. Journal of Marketing Research, 14, 316-321.

Bloom, D. (1998). Technology, experimentation, and the quality of survey data. Science, 280, 847-848.

Byrne, D. (1961). Interpersonal attraction and attitude similarity. Journal of Abnormal and Social Psychology, 62, 713-715.

Catania, J. A. (1997). A model for investigating respondent-interviewer relationships in sexual surveys. In J. Bancroft (Ed.), Researching sexual behavior (pp. 417-435). Bloomington, IN: Indiana University Press.

Catania, J. A., Binson, D., Canchola, J., Pollack, L. M., Hauck, W., & Coates, T. (1996). Effects of interviewer gender, interviewer choice, and item context on responses to questions concerning sexual behavior. Public Opinion Quarterly, 60, 345-375.

Catania, J. A., Binson, D., Peterson, J., & Canchola, J. (1997). The effects of question wording, interviewer gender, and control on item response by African American respondents. In J. Bancroft (Ed.), Researching sexual behavior (pp. 110-113). Bloomington, IN: Indiana University Press.

Catania, J. A., Binson, D., Van der Straten, A., & Stone, V. (1995). Methodological research on sexual behavior in the AIDS era. Annual Review of Sex Research, 6, 77-125.

Catania, J. A., Coates, T. J., Kegeles, S., Thompson-Fullilove, M., Peterson, J., Marin, B., Siegel, D., & Hulley, S. (1992). Condom use in multi-ethnic neighborhoods of San Francisco: The population-based AMEN (AIDS in Multi-Ethnic Neighborhoods) study. American Journal of Public Health, 82, 284-287.

Catania, J. A., Gibson, D., Chitwood, D., & Coates, T. J. (1990). Methodological problems in AIDS behavioral research: Influences on measurement error and participation bias in studies of sexual behavior. Psychological Bulletin, 108, 339-362.

Catania, J. A., McDermott, L., & Pollack, L. (1986). Questionnaire response bias and face-to-face interview sample bias in sexuality research. The Journal of Sex Research, 22, 52-72.

Cozby, P. C. (1973). Self-disclosure: A literature review. Psychological Bulletin, 79, 73-91.

Croyle, R., & Loftus, E. (1993). Recollection in the kingdom of AIDS. In D. Ostrow & R. Kessler (Eds.), Methodological issues in AIDS behavioral research (pp. 163-180). New York: Plenum Press.

Darrow, W., Jaffe, H., Thomas, P., Haverkos, H., Rogers, M., Guinan, M., Auerbach, D., Spira, T., & Curran, J. (1986). Sex of interviewer, place of interview, and responses of homosexual men to sensitive questions. Archives of Sexual Behavior, 15, 79-88.

DeLamater, J. (1974). Methodological issues in the study of premarital sexuality. Sociological Methods and Research, 3, 30-61.

DeLamater, J., & MacCorquodale, P. (1975). The effects of interview schedule variations on reported sexual behavior. Sociological Methods and Research, 4, 215-236.

Depaulo, B. M. (1992). Nonverbal behavior and self-presentation. Psychological Bulletin, 111, 203-243.

Dindia, K., & Allen, M. (1992). Sex differences in self-disclosure: A metaanalysis. Psychological Bulletin, 112, 106-124.

Dindia, K., Fitzpatrick, M., & Kenny, D. (1997). Self-disclosure in spouse and stranger interaction: A social relations analysis. Human Communications Research, 23, 388-412.

Dingler-Duhon, M., & Brown, B. (1987). Self-disclosure as an influence strategy: Effects of Machiavellianism, androgyny, and sex. Sex Roles: A Journal Of Research, 16, 109-123.

Finkelhor, D. (1987). The sexual abuse of children: Current research reviewed. Psychiatric Annals, 17, 233-241.

Fisher, J., Nadler, A., & Whitcher-Alagna, S. (1982). Recipient reactions to aid. Psychological Bulletin, 91, 27-54.

Franco, J., Malloy, T., & Gonzalez, R. (1984). Ethnic and acculturation differences in self-disclosure. The Journal of Social Psychology, 122, 21-32.

Galbraith, G., Strauss, M., Jordan-Viola, E., & Cross, H. (1974). Social desirability ratings from males and females: A sexual item pool. Journal of Consulting and Clinical Psychology, 42, 909-910.

Gebhard, P. (1972). Incidence of overt homosexuality in the United States and Western Europe. In L. M. Livingwood (Ed.), National Institute of Mental Health task force on homosexuality. Washington, DC: National Institute of Mental Health.

Groves, R. M. (1989). Survey errors and survey costs. New York: Wiley.

Groves, R. M., Cialdini, R. B., & Couper, M. P. (1992). Understanding the decision to participate in a survey. Public Opinion Quarterly, 56, 475-495.

Hansen, J. E., & Schuldt, W. J. (1982). Physical distance, sex, and intimacy in self-disclosure. Psychological Reports, 51, 3-6.

Harris, S., & Busby, D. (1998). Therapist physical attractiveness: An unexplored influence on client disclosure. Journal of Marital and Family Therapy, 24, 251-257.

Herold, E., & Way, L. (1988). Sexual self-disclosure among university women. The Journal of Sex Research, 24, 1-14.

Johnson, W., & DeLamater, J. (1976). Response effects in sex surveys. Public Opinion Quarterly, 40, 165-181.

Jones, E. E., & Pittman, T. S. (1982). Toward a general theory of strategic self-presentation. In J. Suls (Ed.), Psychological perspectives on the self: Vol. 1 (pp. 231-262). Hillsdale, NJ: Lawrence Erlbaum Associates.

Jourard, S. M. (1971). Self disclosure: An experimental analysis of the transparent self New York: Wiley-Interscience.

Kaplan, M., Becker, J., & Tenke, C. (1991). Influence of abuse history on male adolescent self-reported comfort with interviewer gender. Journal of Interpersonal Violence, 6, 3-11.

Lefcourt, H. (1982). Locus of Control: Current trends in theory and research. Hillsdale, NJ: Lawrence Erlbaum Associates.

Meston, C., Heiman, J., Trapnell, P., & Paulhus, D. (1998). Socially desirable responding and sexuality self-reports. The Journal of Sex Research, 35, 148-157.

Miller, C. M., Berg, J. H., & Archer, R. (1983). Openers: Individuals who elicit intimate self-disclosure. Journal of Personality and Social Psychology, 44, 1234-1244.

Millstein, S., & Irwin, C. (1983). Acceptability of computer-acquired sexual histories in adolescent girls. The Journal of Pediatrics, 103, 815-819.

Newcomb, T. M. (1961). The acquaintance process. New York: Holt.

Ostrow, D. G., & Kessler, R. C. (Eds.). (1993). Methodological issues in AIDS behavioral research. New York: Plenum Press.

Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 45, 598-609.

Pennebaker, J. W. E. (1995). Emotion, disclosure, & health. Washington, DC: American Psychological Association.

Peterson, J., & Catania, J. A. (1997). Item nonresponse in the National AIDS Behavioral Surveys among African American and White respondents. In J. Bancroft (Ed.), Researching Sexual Behavior (pp. 106-109). Bloomington, IN: Indiana University Press.

Schober, M., & Conrad, F. (1997). Does conversational interviewing reduce survey measurement error? Public Opinion Quarter(v, 61, 576-602.

Schuman, H., & Presser, S. (1981). Questions and answers in attitude surveys: Experiment in question form, wording, and context. Orlando, FL: Academic Press.

Shaffer, D., Ogden, J. K., & Wu, C. (1987). Effects of self-monitoring and prospect of future interaction on self-disclosure reciprocity during the acquaintance process. Journal of Personality, 55, 75-96.

Shaffer, D., Pegalis, L., & Cornell, D. (1988). Gender and self-disclosure revisited: Personal and contextual variations in self-disclosure to same-sex acquaintances. The Journal of Social Psychology, 132, 307-315.

Shaffer, D., Ruammake, C., & Pegalis, L. (1990). The "Opener": Highly skilled as interviewer or interviewee. Personality and Social Psychology Bulletin, 16, 511-520.

Singer, E., Frankel, M., & Glassman, M. (1983). The effect of interviewer characteristics and expectations on response. Public Opinion Quarterly, 47, 68-83.

Singer, E., & Kohnke-Aguirre, L. (1979). Interviewer expectation effects: A replication and extension. Public Opinion Quarterly, 43, 245-260.

Singer, E., Von Thurn, D., & Miller, E. (1995). Confidentiality assurances and response. Public Opinion Quarterly, 59, 66-77.

Snell, Jr., W., Belk, S., Papini, D., & Clark, S. (1989). Development and validation of the Sexual Self-Disclosure Scale. Annals of Sex Research, 2, 307-334.

Stotland, E., & Canon, L. (1972) Social psychology: A cognitive approach. Philadelphia: W. R. Saunders.

Sudman, S., & Bradburn, N. (1982). Asking questions: A practical guide to questionnaire design. San Francisco: Jossey-Bass.

Sudman, S., Bradburn, N., & Schwarz, N. (1996). Thinking about answers: The application of cognitive processes to survey methodology. San Francisco: Jossey-Bass.

Tanur, J. (Ed.). (1992). Questions about questions: Inquiries into the cognitive bases of surveys. New York: Russell Sage.

Tessler, R., & Schwartz, S. (1972). Help seeking, self-esteem, and achievement motivation: An attributional analysis. Journal of Personality and Social Psychology, 21, 318-326.

Tolstedt, B., & Stokes, J. (1984). Self-disclosure, intimacy, and the depenetration process. Journal of Personality and Social Psychology, 46, 84-90.

Tourangeau, R., & Smith, T. (1996). Asking sensitive questions: The impact of data collection mode, question format, and question context. Public Opinion Quarterly, 60, 275-304.

Turner, C. F., Danella, R. D., & Rogers, S. M. (1995). Sexual behavior in the United States, 1930-1990: Trends and methodological problems. Sexually Transmitted Diseases, 22, 173-190.

Turner, C. F., Ku, L., Rogers, S. M., Lindberg, L. D., Pleck, J. H., & Sonenstein, F. L. (1998). Adolescent sexual behavior, drug use, and violence: Increased reporting with computer survey technology. Science, 280, 867-873.

Turner, C. F., Ku, L., Sonenstein, F. L., & Pleck, J. H. (1996). Impact of Audio-CASI on bias in reporting of male-male sexual contacts. In R. Warnecke (Ed.), NCHS Proceedings of the 6th Conference on Health Survey Research Methods. Hyattsville, MD: National Center for Health Statistics.

Turner, C. F., & Martin, E. (Eds.). (1984). Surveying subjective phenomena (Vol. 1). New York: Russell Sage.

Turner, C. F., Miller, H. G., & Rogers, S. M. (1997). Survey measurement of sexual behavior: Problems and progress. In J. Bancroft (Ed.), Researching Sexual Behavior (pp. 37-60). Bloomington, IN: Indiana University Press.

Walster, E., Aronson, V., Abrahams, D., & Rottman, L. (1966). Importance of physical attractiveness in dating behavior. Journal of Personality and Social Psychology, 4, 508-516.

White, R. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66, 297-333.

Manuscript accepted October 15, 1998

This work is funded in part by NIMH/NIA grants MH52022, MH/AG51523, and MH54320. An earlier version of this paper appeared in Bancroft (1997).

Address correspondence to Joseph A. Catania, Center for AIDS Prevention Studies, University of California, San Francisco, 74 New Montgomery, Suite 600, San Francisco, CA 94105.