DESPITE THE FACT THAT sexual behavior occurs within a dyad, most studies of sexual behavior and sexual risk are based on the self-reports of individuals about their own behavior and attitudes. Because of the sensitive nature of sexual behavior, concerns have been raised about potential biases in these self-reports. 1,2 Reference group norms, social desirability, and other factors could contribute to responses of study participants. Because sexual behavior is largely unobservable and no “gold standard” for the validity of such self-reports exists, determining the validity of these measures is usually not possible. 3
Previous research has sought to improve the validity and reliability of self-reported sexual behavior. One approach is to use biologic markers, such as sexually transmitted infection (STI) or pregnancy, to verify self-reports and thereby determine measurement validity. 4,5 Biologic markers are not always practical, however, as a result of the large numbers of subjects necessary to determine outcomes. 6–8 In addition, there is not a one-to-one correspondence between negative biologic outcomes and sexual behavior, resulting in bias even when biologic outcomes are used. 6,8
Another approach to evaluate the accuracy of self-reported sexual behavior is to collect data from both members of a sexual dyad and examine differences and similarities in reports about joint knowledge or behaviors. Comparison of the reports from both partners cannot absolutely determine the validity of self-report data, but it does allow assessment of reliability, a precursor of validity. If partners do not concur in their reports of a joint sexual behavior (ie, reliability is low), then the validity of the information is questionable.
In addition to evaluating the accuracy of self-reported sexual behavior, couple-level data can be used to assess the extent of agreement between members of a sexual dyad about factors that influence safer sexual behavior. For example, because of the dyadic nature of sexual behavior, the role of relationship power and sexual decision-making in HIV risk reduction has drawn increasing attention. 9–13 Findings from recent studies suggest that relationship power, decision-making dominance, and sexual assertiveness could be important contextual variables that influence women’s and men’s ability to engage their sexual partner(s) in safer sex behaviors. 12,14–17 These findings support the importance of obtaining accurate self-reports of relationship power and decision-making dominance. In addition, lack of agreement between partners about these relationship dynamics could increase HIV/STI risk in the relationship because beliefs about who is in control of decisions regarding condom use could lead one or both partners to refrain from initiating safer sex behavior. Despite the potential importance for the design of effective intervention programs, research on agreement in partners’ perceptions of power and decision-making in their relationships is generally lacking.
Another important topic to examine with couple-level data is the accuracy of individuals’ perceptions of their partners’ risky sexual behaviors and attitudes toward condom use. Partners might not be knowledgeable about one another’s behaviors and attitudes because they have inaccurate information or perceptions about their partner. Previous studies have indicated that individuals do not always have accurate assessments of their partners’ risky behaviors. 8,18,19 Unfortunately, an inaccurate perception of a sexual partner’s current or past risky behaviors could increase an individual’s risk of acquiring an STI. 20 This inaccurate assessment could explain how partners bring risk into their primary relationships that is not perceived by their partners. 20
In addition to accurate knowledge of a sexual partner’s current or past risky behaviors, attitudes toward safer sexual behavior could also influence safer sex practices. Attitudes toward condom use have been found to be an important predictor of condom-use behavior. 21–24 Moreover, the importance of perceived partner norms on condom-use behavior has been established in theoretical frameworks of behavior change. 18,25,26 It is likely, therefore, that incorrect perceptions about partner attitudes regarding the use of condoms could lead to increased risky sexual behavior. Only one study to our knowledge 8 has examined the extent of agreement between individuals’ perceptions about their main partners’ attitudes regarding condom use and their partners’ report of these attitudes.
Accordingly, the objectives of this study were to: 1) assess interpartner concordance on selected relationship characteristics, self-reported sexual behavior and condom use, and perceptions of relationship power and decision-making dominance regarding sexual and reproductive behavior; and 2) examine the extent of agreement between individuals’ perceptions about their main partners’ risky sexual behaviors and attitudes about condom use and their partners’ report of these behaviors and attitudes. Both objectives have implications for future research using data collected from individuals and both members of couples.
The data come from the formative phase of a project designed to develop, implement, and evaluate a couple-based intervention to reduce the risk of unintended pregnancy and STIs, including HIV among young (aged 18–25 years) women at increased risk and their male partners. The sample consists of 112 couples in four U.S. cities: Atlanta, Georgia (n = 30), Los Angeles, California (n = 39), Oklahoma City, Oklahoma (n = 21), and Portland, Oregon (n = 22). The women and their male partners were recruited from family planning and sexually transmitted disease clinics and other community locations in each city using both passive (eg, posters and fliers) and active (eg, recruiters approaching potential participants in the clinics) strategies.
We recruited the majority of couples (n = 106) through the women. Women were eligible if they were 18 to 25 years old, had a male sex partner aged 18 or older of any race/ethnicity, had sex without a condom within the previous 3 months, and met one or more of the following criteria: 1) more than one lifetime sex partner, 2) ever had an STI, 3) ever had sex with a man who she knew or thought was having sex with other men or women, 4) ever had sex with a man who she knew or thought was using intravenous drugs (using a needle to inject drugs into his body), or 5) ever had sex with a man who she knew or thought had an STI or HIV/AIDS. In addition, women had to self-identify as being black in Atlanta and Portland and as Hispanic in Los Angeles; women in Oklahoma could be of any race/ethnicity. Women also had to be willing to ask their partner to participate in the study. If women had more than one sexual partner, they decided which partner to ask to participate. The male partner that they identified had to be 18 years of age or older but did not have to be of the same race/ethnicity as the woman. Women who were pregnant, who intended to become pregnant within the year, or who self-identified as being HIV-positive were not eligible. Women were screened either over the telephone or in person at recruitment sites. Eligible women were asked to invite their partners to participate. Both members of a couple had to agree to participate for the couple to be enrolled in the study. In addition, we recruited 1 couple in Oklahoma City and 5 couples in Atlanta through the men; men had to meet similar criteria as those described previously for women and, additionally, they had to report that their female partners met those criteria.
The majority of women were black (55%) or Hispanic (35%) and 9% were non-Hispanic white (Table 1). Mean age for the sample was 22.9 years; men were slightly older than women (24.2 years and 21.5 years, respectively). Nearly two thirds of the participants had graduated from high school (69% of the men and 58% of the women). The majority of the men (86%) but only half of the women (49%) were employed. More women reported having had an STI other than HIV than did men (50% vs. 37%).
Interviews were conducted in Atlanta, Los Angeles, and Oklahoma City between July and November 1998, and in Portland between January and April 1999. Interviewers in Atlanta, Los Angeles, and Oklahoma City were trained together at a group meeting. Interviewers in Portland were trained at a separate group meeting. Procedures were standardized across all sites. Individual semistructured interviews lasting approximately 60 minutes were conducted in private. Members of the couple were interviewed individually but concurrently. Participants and interviewers were matched according to gender at all sites and according to race/ethnicity at all sites except Oklahoma City. In Los Angeles, participants could choose to have the interview conducted in Spanish or English. Interviewers referred to a standard interview guide for all interview questions. Interviews were audiotaped, and responses to close-ended questions were noted on hard copies of the guide. Each person received $30 for their participation and was compensated for travel and/or childcare costs. Institutional Review Board approval was obtained at each site, and written informed consent was obtained from each participant after the study was fully explained.
The interview guide was developed specifically for this study. It included open-ended as well as structured questions. A unique feature of our interview guide is the use of partner-specific measures rather than more general measures to assess correlates of heterosexual condom use. We made many of our measures partner-specific because we believe that risk perceptions, attitudes, and behaviors in the sexual domain will differ with different partners.
The interview guide contained questions concerning participants’ demographic characteristics, relationship characteristics, sexual behavior, and risk factors. We asked participants if there was anyone else they had sex with besides [PARTNER’S NAME] in the past 90 days and if they thought that [PARTNER’S NAME] had sex with other men/women in the past 90 days.
We used several measures to assess condom use. We asked participants whether during the last 3 months they had used condoms with [PARTNER’S NAME] at least once. We also asked how often they used condoms with [PARTNER’S NAME] when they had vaginal sex. Response categories were “never,” “rarely,” “sometimes,” “most of the time,” and “always.” In addition, we asked whether they used a condom the last time they had sex with [PARTNER’S NAME]. A proportional measure of consistency of condom use was constructed by dividing the report of the number of times condoms were used for vaginal or anal sex in the past 90 days by the report of the number of times they had vaginal or anal sex. Finally, participants were asked to rate the importance of using condoms both to themselves and to their partners. Response categories were “not important,” “slightly important,” “moderately important,” “very important,” and “extremely important.”
As a general measure of relationship power, we asked participants “In your relationship with [PARTNER’S NAME], in general who has more power?” The response categories were “I do,” “we both do,” and “my partner does.” To examine dominance within specific decision-making domains, we asked participants who in their relationship decides 1) when to get pregnant, 2) whether to use something to keep from getting pregnant, 3) whether to use a condom, 4) whether to have sex, and 5) what kinds of things they do when they have sex. Response categories were “me,” “both of us,” and “my partner.”
We assessed couple concordance for nominal outcomes using raw agreement indices and the kappa statistic. The overall percent agreement was calculated as the total number of couples with concordant responses divided by the total number of couples in the sample. Although this value describes the raw level of concordance, it does not indicate the direction of agreement. The direction of agreement was assessed using positive and negative conditional probability indices. 27 These indices are interpreted as the probability of the response of one member of a couple given the response of the other member of the couple. 28,29 The positive conditional probability index (CP+) is the percent of couples in agreement that a statement is true or a behavior did occur. The negative conditional probability index (CP-) is the percent of couples who agreed that a statement is not true or that a behavior did not occur. When CP+ and CP− concurrently are high, concordance is high.
The kappa statistic is a widely used measure of reliability that corrects for chance agreement. 30,31 Although some uses of the kappa statistic have been criticized, it is appropriate for the assessment of reliability when the outcome is dichotomous. 32,33 For variables with 3 response categories, we created dichotomies and computed kappa statistics and conditional probability indices for each category compared with the other 2 response categories combined. The kappa statistic does not provide information about the direction of agreement and can be biased when marginal frequencies are very asymmetric. 29 Therefore, raw agreement, conditional probability, and the kappa statistic are all reported because they provide complementary information. 27 By convention, values of kappa ranging from 0 to 0.20 indicate slight agreement, 0.21 to 0.40 indicate fair agreement, 0.41 to 0.60 indicate moderate agreement, 0.61 to 0.80 indicate substantial agreement, and 0.81 and above indicate nearly perfect agreement. 28,34
Finally, for ordinal variables and interval variables, which were all highly skewed, we used the Spearman’s rank correlation coefficient (r) to estimate the degree of association between couples’ responses. Absolute values from 0.40 to 0.70 indicate moderate agreement and those above 0.70 indicate substantial agreement. 35
Concordance on Relationship Characteristics
Table 2 summarizes partners’ concordance on selected relationship characteristics. Over half of the couples in the sample reported that they live together, and the agreement of their reports was nearly perfect and significant (kappa = 0.836). Similarly, couples were highly concordant in their reports of marital status (kappa = 0.967), whether they have children together (kappa = 1.00), and the duration of their relationship (r = 0.932).
Concordance on Sexual Behavior and Condom Use
Couples also were concordant on their reports of sexual and condom-use behavior. We found no mean differences in the number of times partners reported having had vaginal and anal intercourse with one another in the past 90 days, and the correlations were moderately high for both measures (Table 3). Nearly two thirds of the sample reported condom use in the last 90 days, and the concordance of partners’ reports was substantial (kappa = 0.686). Agreement was also substantial (kappa = 0.646) for condom use during the most recent vaginal sex with slightly less agreement that condoms were used (CP+ 75.0) than that condoms were not used (CP− 89.6). Finally, partners’ reports of the frequency of condom use were moderately correlated when measured on a 5-point scale (r = 0.621) and as a proportion based on partners’ reports of the number of times condoms were used for vaginal or anal sex during the past 90 days (r = 0.692).
Concordance on Relationship Power and Decision-Making Dominance
In contrast, couples did not agree on who they perceived has more power in their relationship and dominance within specific decision-making domains (Table 4). Based on kappa statistics, partners reported only slight or fair agreement on all of the responses for the 6 items. The highest kappas were for deciding whether to have sex (kappa = .257) and for deciding when to get pregnant (kappa = .253). Additional information from raw agreement and conditional probability indices provided a more detailed understanding of the direction and level of agreement.
For example, with regard to relationship power, we found that given one partner’s report that the woman had more power, the probability that the other partner agreed was only 0.357 (CP+ 35.7). Similarly, given one partner’s report that the male partner had more power, the probability that the other partner agreed was only 0.379 (CP+ 37.9). The highest positive conditional probability for this item was for both partners having power; given one partner’s response that both partners have power, the probability that the other agreed was 0.589 (CP+ 58.9). However, the conditional probability was highest (CP− 82.4) for the belief that the woman does not have more power. We also found fair agreement for this category according to the kappa statistic (kappa = .205).
In summary, across all of the power and decision-making items, a similar pattern emerged. When either partner reported that the man or woman had the power or made the decision, the other partner was unlikely to agree. Accordingly, the highest conditional probabilities for each of the 6 items were the following: women not having more power (CP− 82.4), men not deciding when to get pregnant (CP− 93.1), men not deciding whether to use something to keep from getting pregnant (CP− 90.4), women not deciding whether to use a condom (CP− 85.4), men not deciding whether to have sex (CP− 91.1), and women not deciding what kinds of things to do during sex (CP− 94.2). The raw percent agreement was highest for decisions about when to get pregnant (60.9%) and what kinds of things they do when they have sex (59.5%). For all decision-making items, however, at least 40% of the couples gave discordant responses. When asked who in the couple has more power, over 50% of couples gave discordant responses.
Concordance Between Participants’ Perceptions and Their Partners’ Reports
This section reports findings from the analysis that examined the accuracy of individuals’ perceptions of their partners’ attitudes and behaviors. Nearly one fourth of the men (22%) and 14% of women reported having had sex with someone other than their main partners in the past 3 months. We tested the accuracy of individuals’ perceptions regarding whether their partners had had sex with someone else. The 8 participants who indicated that they “did not know” whether their partners had sex with someone else were excluded from these analyses. We found only fair agreement between women’s perceptions and their partners’ reports (kappa = 0.298, P <0.01). In contrast, we found substantial agreement between men’s perceptions and their partners’ report of sex with someone else (kappa = 0.619, P <0.001). Male partners had a very high conditional probability (CP− 95.2) of correctly identifying that their female partners had not had sex with someone else and a fairly high probability (CP+ 66.7) of correctly identifying that she had. However, given a male partner’s self-report that he did have sex with someone else, the conditional probability that the female partner knew or suspected that he had was fairly low (CP+ 45.8).
In additional analyses, we examined the extent of agreement between individuals’ perceptions about their main partners’ attitudes regarding the importance of condom use and their partners’ report of these attitudes. Although women’s perceptions of their male partners’ attitudes toward condom use were significantly correlated with their partners’ reports, the level was slightly below moderate (r = 0.370, P <0.001). In contrast, men’s perceptions were not significantly related to their female partners’ attitudes about condom use (r = 0.099, not significant).
Both women’s and men’s perceptions of their partners’ attitudes toward condom use were moderately correlated with their own condom-use attitudes (r = 0.660 and r = 0.682, respectively, P <0.001). In addition, based on paired t tests, individuals’ perceptions of their partners’ ratings of the importance of condom use were significantly lower than their partners’ actual ratings (data not shown). Thus, both women and men significantly underestimated the importance that their partners placed on condom use.
Interpartner concordance varies depending on the information being reported. It is noteworthy that we observed substantial concordance on reports of sexual intercourse and condom use for both dichotomous and continuous measures and for different recall periods (ie, last sex, prior 3 months). In general, these results are consistent with most previous research, 3,8,27,36 but differ from the findings of Ellish and colleagues in which recall time was explicitly evaluated. 37 Differences in findings could in part reflect differences in samples (eg, age range, recruitment source). Overall, however, this growing body of work suggests that discrepancies between partners’ reports of sexual behavior could be small. The implications of these findings are important. An emergent issue involved in analyzing data collected from couples is that individuals in a dyad do not always agree on outcome variables, and researchers are faced with the dilemma of how best to measure the behavior of couples. When concordance is substantial, measurement of couple behavior poses fewer problems. In addition, studies based on reports from only one couple member might not be seriously flawed. One partner’s report could provide a reasonably accurate reflection of sexual behavior and condom use in the relationship.
A unique contribution of this study is the examination of interpartner concordance on reports of relationship power and sexual decision-making dominance. Studies have begun to examine the association between relationship power and condom use. 12,14,15 Research on agreement in partners’ perceptions of power and decision-making in their relationships is, however, generally lacking. In a study of 39 couples of Mexican origin living in Los Angeles County, large percentages of couples disagreed about who makes decisions regarding sexual activity and contraceptive use. 38 Similarly, in the current study, we found that interpartner concordance was lowest for subjective reports regarding power and decision-making. Couples tended to disagree on who has more power in their relationship as well as dominance within specific decision-making domains. The discordance could be the result of different reporting tendencies (ie, men might tend to say that they have more power in the relationship even if this is not actually the case as a result of social desirability) or different subjective perceptions of the relationship (ie, partners actually differ in who they think makes decisions).
Whereas couples’ reports of sexual behavior are accounts of their joint behavior (ie, what they did together), their perceptions of power and decision-making are more subjective and based on their own opinion. In this sense, it is not surprising that there is less agreement on these topics. Discrepancies in partners’ beliefs about their relationship (eg, who makes decisions) could contribute to misunderstandings about what the other partner thinks or wants, suggesting that increasing couples’ communication skills could be important. Based on the overall percent agreement, couples were least likely to agree on who has more power in the relationship. The response category with the greatest agreement for this item was that women did not have more power. In addition, half of the couples disagreed about who decided whether to use a condom. Again, the response category with the highest agreement for this item was that the woman did not decide. Because of the significance of these relationship dynamics for STI prevention, 12,14–17 further research is needed to better understand interpartner concordance regarding relationship power and condom-use decision-making and the importance of concordance on safer sex behaviors.
We also examined the accuracy of participants’ perceptions of their partners’ attitude regarding the importance of condom use. Like Seal, 8 we found that couples did not appear to be very knowledgeable about their partners’ attitudes toward condoms. Women’s reports were more accurate than the men’s. One reason for this could be related to the fact that condoms are male-controlled. If women want to use condoms, they must seek their male partners’ cooperation. As a result, women could be more likely to elicit their partners’ attitudes toward condoms and could be more aware of their partners’ thinking about condoms. In contrast, men might have less need to know their partners’ attitudes regarding condom use. It is important to note that both men and women significantly underestimated the importance their partners placed on using condoms, suggesting that both men and women view condoms more favorably than their partners realize. Because attitudes toward using condoms are an important predictor of condom-use behavior, 21–24 interventions directed at HIV risk reduction might need to enhance couples’ communication skills to increase knowledge about partners’ motivation and attitudes.
We also found that men’s reports of whether their partners recently had sex with someone else were more accurate than the women’s reports. These results could indicate that, compared with men, women are less aware of their partners’ sexual behavior outside of their relationship. Alternatively, women could be less willing to acknowledge that their partner is having sex with someone else. These findings suggest that interventions directed at HIV risk reduction need to address women’s misperceptions about the risk that their partners bring to their relationships.
The current study has several limitations as a result of selection. Like previous research on this topic, the current study did not use probability sampling techniques. Rather, our sample consisted of young couples at increased risk of HIV/STIs who were recruited from family planning and sexually transmitted disease clinics and other community locations in 4 cities. By design, the study did not include women who reported using condoms during all acts of sexual intercourse in the prior 3 months. Agreement on condom-use items in our sample could be lower than would be found in a population containing respondents with 100% condom use. Furthermore, to be eligible for the study, women (and, in a few cases, men) had to be willing to ask their partners to participate in the study and their partners had to agree. For these reasons, the couples who participated in our study could differ from other couples, which could affect the external validity of our results. Yet, a strength of the current study is that the sample was ethnically and geographically diverse and consisted of couples at increased risk for STIs, including HIV.
As a result of our relatively small sample size, we could not analyze the data by site. Further research should assess whether concordance is especially high or especially low among some subgroups. For example, couples who have been together longer could be more concordant on some topics than couples in new relationships (eg, couple members could become more aware of their partners’ attitudes as their relationship develops). In addition, concordance could be lower among men and women who are younger and less sexually experienced. Studies that identify the individuals and relationships for which concordance is likely to be lowest and highest could be especially helpful in the design of interventions to reduce HIV risk. For example, couples with low concordance could need different types of interventions (eg, ones with more of a focus on communication skills) than those with high concordance.
In conclusion, our findings suggest that interpartner concordance is higher for reports of relationship characteristics and sexual and condom-use behavior as compared with perceptions of decision-making regarding reproductive health matters. Whereas differences in the reporting of events and behaviors could indicate reporting errors, differences in the report of perceptions and opinions are expected. Further exploration of interpartner discordance is needed. Lack of agreement could indicate: 1) differing interpretations of the meaning of questions or terms (eg, meaning of power) or the scales used to measure responses; 2) differing opinions, perceptions, or subjective judgments about an interaction (eg, one partner could believe he or she alone makes the decision about an issue, but the other partner believes he or she has a say); 3) not knowing or forgetting over time the correct or “true” answer; and 4) random or systematic distortion in recollection. Most likely, lack of concordance is influenced by a combination of these error sources. Disentangling their relative effects can lead to clearer questions, more valid and reliable measures, and more precise data collection procedures.
1. Catania JA, Gibson DR, Chitwood DD, Coates TJ. Methodologic problems in AIDS behavioral research: Influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull 1990; 108: 339–362.
2. James NJ, Bignell CJ, Gillies PA. The reliability of self-reported sexual behavior. AIDS 1991; 5: 333–336.
3. Padian NS, Aral S, Vranizan K, Bolan G. Reliability of sexual histories in heterosexual couples. Sex Transm Dis 1995; 22: 169–172.
4. Catania JA, Binson D, Van Der Straten A, Stone V. Methodological research on sexual behavior in the AIDS era. Annu Rev Sex Res 1995; 6: 77–125.
5. Peterman TA, Lin LS, Newman DR, et al. Does measured behavior reflect STD risk? An analysis of data from a randomized controlled behavioral intervention study. Sex Transm Dis 2002; 29: 530–532.
6. Fishbein M, Pequegnat W. Evaluating AIDS prevention interventions using behavioral and biological outcome measures. Sex Transm Dis 2000; 27: 101–110.
7. Pequegnat W, Fishbein M, Celentano D, et al. STD NIMH/APPC workgroup on behavioral and biological outcomes in HIV/STD prevention studies: A position statement. Sex Transm Dis 2000; 27: 127–132.
8. Seal DW. Interpartner concordance of self-reported sexual behavior among college dating couples. J Sex Res 1997; 34: 39–55.
9. Amaro H. Love, sex, and power: Considering women’s realities in HIV prevention. Am Psychol 1995; 50: 437–443.
10. Fullilove MT, Fullilove RE, Haynes K, Gross S. Black women and AIDS prevention: A view towards understanding the gender rules. J Sex Res 1990; 27: 47–64.
11. Gómez CA, Marín BV. Gender, culture, and power: Barriers to HIV-prevention strategies for women. J Sex Res 1996; 33: 355–362.
12. Pulerwitz J, Amaro H, De Jong W, et al. Relationship power, condom use and HIV risk among women in the USA. AIDS Care 2002; 14: 789–800.
13. Wingood GM, DiClemente RJ. Application of the theory of gender and power to examine HIV-related exposures, risk factors, and effective interventions for women. Health Educ Behav 2000; 27: 539–565.
14. Harvey SM, Bird ST, DeRosa CJ, Montgomery SB, Rohrbach LA. Sexual decision-making and safer sex behavior among young female injection drug users and female partners of IDUs. J Sex Res 2003; 40: 50–60.
15. Harvey SM, Bird ST, Galavotti C, Duncan EA, Greenberg D. Relationship power, sexual decision-making and condom use among women at risk for HIV/STDs. Women Health 2002; 36: 69–84.
16. Noar SM, Morokoff PJ, Redding CA. Sexual assertiveness in heterosexually active men: A test of three samples. AIDS Educ Prev 2002; 14: 330–342.
17. Wingood GM, DiClemente RJ. Partner influences and gender-related factors associated with noncondom use among young adult African American women. Am J Community Psychol 1998; 26: 29–51.
18. Ellen JM, Vittinghoff E, Bolan G, Boyer CB, Padian NS. Individuals’ perceptions about their sex partners’ risk behaviors. J Sex Res 1998; 35: 328–332.
19. Niccolai LM, Farley TA, Ayoub MA, Magnus Y, Kissinger PJ. HIV-infected persons’ knowledge of their sexual partners’ HIV status. AIDS Educ Prev 2002; 14: 183–189.
20. Misovich SJ, Fisher JD, Fisher WA. Close relationships and elevated HIV risk behavior: Evidence and possible underlying psychological processes. Rev Gen Psychol 1997; 1: 72–107.
21. Bryan AD, Aiken LS, West SG. Young women’s condom use: The influence of acceptance of sexuality, control over the sexual encounter and perceived susceptibility to common STDs. Health Psychol 1997; 16: 468–479.
22. Campbell SM, Peplau LA, DeBro SC. Women, men and condoms: Attitudes and experiences of heterosexual college students. Psychol Women Quart 1992; 16: 273–288.
23. Helweg-Larsen M, Collins BE. The UCLA multidimensional condom use attitudes scale: Documenting the complex determinants of condom use in college students. Health Psychol 1994; 13: 224–237.
24. Sacco WP, Levine B, Reed LL, Thompson K. Attitudes about condom use as an AIDS-relevant behavior: Their factor structure and relation to condom use. Psychol Assess 1991; 3: 264–272.
25. Fishbein M. The role of theory in HIV prevention. AIDS Care 2000; 12: 273–278.
26. Fisher JD, Fisher WA. Changing AIDS-risk behavior. Psychol Bull 1992; 111: 455–474.
27. Ochs EP, Binik YM. The use of couple data to determine the reliability of self-reported sexual behavior. J Sex Res 1999; 36: 374–384.
28. Fleiss JL. Statistical Methods for Rates and Proportions, 2nd ed. New York: Wiley, 1981.
29. Cicchetti DV, Feinstein AR. High agreement but low kappa: II. Resolving the paradoxes. J Clin Epidemiol 1990; 43: 551–558.
30. Cohen JA. Coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 20: 346.
31. Kraemer HC. Kappa coefficient. In: Kotz S, Johnson NL, eds. Encyclopedia of Statistical Sciences. New York: Wiley, 1982.
32. Kraemer HC, Bloch DA. Kappa coefficients in epidemiology: An appraisal of a reappraisal. J Clin Epidemiol 1988; 41: 959–968.
33. Kraemer HC, Periyakoil VS, Noda A. Kappa coefficients in medical research. Stat Med 2002; 21: 2109–2129.
34. Landis JR, Koch GG. The measure of observer agreement for categorical data. Biometrics 1977; 33: 159–174.
35. Williams F. Reasoning With Statistics, 4th ed. Fort Worth: Harcourt Brace Jovanovich College Publishers, 1992.
36. Upchurch DM, Weisman CS, Shepherd M, et al. Interpartner reliability of reporting of recent sexual behaviors. Am J Epidemiol 1991; 134: 1159–1165.
37. Ellish NJ, Weisman CS, Celentano D, Zenilman JM. Reliability of partner reports of sexual history in a heterosexual population at a sexually transmitted diseases clinic. Sex Transm Dis 1996; 23: 446–452.
38. Harvey SM, Beckman LJ, Doty M. Couple dynamics in sexual and reproductive decision-making among Mexican immigrants. Adv Popul 1999; 3: 251–279.