This article was downloaded by: [UVA Universiteitsbibliotheek SZ] On: 26 January 2015, At: 02:42 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
AIDS Care: Psychological and Socio-medical Aspects of AIDS/HIV Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/caic20
Impacts of four communication programs on HIV testing behavior in South Africa a
b
b
Mai Do , D. Lawrence Kincaid & Maria Elena Figueroa a
Department of Global Health Systems and Development, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA b
Center for Communication Programs, Johns Hopkins University, Baltimore, MD, USA Published online: 07 Apr 2014.
Click for updates To cite this article: Mai Do, D. Lawrence Kincaid & Maria Elena Figueroa (2014) Impacts of four communication programs on HIV testing behavior in South Africa, AIDS Care: Psychological and Socio-medical Aspects of AIDS/HIV, 26:9, 1109-1117, DOI: 10.1080/09540121.2014.901487 To link to this article: http://dx.doi.org/10.1080/09540121.2014.901487
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
AIDS Care, 2014 Vol. 26, No. 9, 1109–1117, http://dx.doi.org/10.1080/09540121.2014.901487
Impacts of four communication programs on HIV testing behavior in South Africa Mai Doa*, D. Lawrence Kincaidb and Maria Elena Figueroab a Department of Global Health Systems and Development, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA; bCenter for Communication Programs, Johns Hopkins University, Baltimore, MD, USA
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
(Received 15 July 2013; accepted 3 March 2014) This paper aims to evaluate the impacts of four communication programs on promoting HIV testing behavior among sexually active individuals in South Africa. The four programs, implemented by Johns Hopkins Health and Education in South Africa, are aimed to promote HIV prevention behaviors, as well as gender-based violence prevention, tuberculosis screening and treatment, and reduction of alcohol consumption. Launched between 2009 and 2010, they all promoted HIV testing. Data came from the population-based Third National AIDS Communication Survey 2012; 6004 men and women who had sex within the last 12 months were included in the analysis. Multiple causal attribution analysis is used to justify causal reference and estimate the impact of communication programs. Findings indicate significant direct and indirect effects of the programs on HIV testing behavior. Indirect effects worked through increasing one’s likelihood of perceiving that their friends were tested and the probability of talking about HIV testing with sex partners and friends, which in turn increased the likelihood of HIV testing. Findings suggest multiple angles from which communication programs can promote HIV testing. The study also demonstrates the use of multiple statistical techniques for causal attribution in a post-only design, where randomization is not possible. Keywords: communication; HIV testing; multiple causal attribution; South Africa
Introduction
Theory of mass media effects
Since the beginning of the HIV epidemic, the mass media programs have proven important in increasing knowledge and promoting preventive practices. They have recently begun to address a full continuum of prevention, treatment, care, and support strategies (McKee, Bertrand, Becker-Benton, & Becker, 2004). Yet few studies have examined the impact of mass media programs on HIV testing behavior in low- and middleincome countries. Two recent reviews of communication programs in developing countries did not report any studies with HIV testing as an intended outcome of mass media communication (Bertrand, O’Reilly, Denison, Anhang, & Sweat, 2006; Noar, Palmgreen, Chabot, Dobransky, & Zimmerman, 2009); an exception is Pettifor, MacPhail, Bertozzi, and Rees (2007). This paper aims to evaluate the impact of communication programs in promoting HIV testing behavior in South Africa. Since 1999, the Johns Hopkins Health and Education in South Africa (JHHESA) has been providing technical support on strategic communication for HIV prevention, care, support, and treatment. Four recent programs (Intersexions, Brothers for Life, 4Play: Sex Tips for Girls, and Scrutinize), launched between 2009 and 2010, are included in this assessment – all are TV series and include messages that promote HIV testing (CADRE, 2011; http://www.brothersforlife.org; http://www.intersexions-tv.co.za/; http://www.jhuccp.org/).
Mass media programs initiate causal processes that occur in several stages. Information is expected to affect ideational factors such as knowledge, beliefs, and values (attitudes); self-efficacy; and interpersonal communication about it (Ajzen, 1991; Fishbein & Ajzen, 1975; Kincaid, 2009; Kincaid, Delate, Storey, & Figueroa, 2012). By observing the behavior of others in the media (mediated social modeling), one learns about new behaviors in a way that later guides action in similar situations (Bandura, 1986). The cognitive framework within which information is perceived determines how it is interpreted, understood, and acted upon (Kincaid et al., 2012). Framing creates a point of view that highlights certain facts to be interpreted in a particular manner. A television drama can reframe HIV-positive status from one of sexual promiscuity, irresponsibility, and immorality to that of a health condition that requires medical treatment and taking personal responsibility to protect oneself and others. Dialogs can lead to reframing the issues and making them more relevant. Discussion with others also produces a multistep process of mass media effects (Katz & Lazarsfeld, 2006) and accounts for the well-documented diffusion effect (Rogers, 2003). Discussion also allows for social comparison with one’s friends in order to reduce uncertainty about the issue and to obtain social
*Corresponding author. Email: [email protected] © 2014 Taylor & Francis
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
1110
M. Do et al.
and emotional support for change (Festinger, 1954; Suls, Martin, & Wheeler, 2002). Multistep effects are slow to occur, however, if potential initial adopters believe that the new position is not the social norm and are thus reluctant to talk about it with others for fears of being criticized, according to the spiral of silence theory (Noelle-Neumann, 1991, 1993). Mass media can be used to frame a new behavior as a new social norm, reducing social stigma, and thereby increasing the likelihood of talking about it with others. All the direct ideational changes that result from mass media exposure also result from this second stage of interpersonal communication. This paper examines this multistep process – whether communication programs could change existing norms and the likelihood of discussing HIV-related topics, thereby increasing HIV testing. Studies have shown that social norms related to testing and HIV-related communication were important determinants of the readiness for testing and testing behavior (Babalola, 2007; Baiden et al., 2005; Gage & Ali, 2005; Hendriksen et al., 2009; Kranzer et al., 2008; Nuwaha, Kabatesi, Muganwa, & Whalen, 2006; Rimal, Limaye, Roberts, Brown, & Mkandawire, 2013). Communications with sex partners about HIV testing was key for its uptake in Burkina Faso (Sarker, Sanou, Snow, Ganame, & Gondos, 2007). In South Africa, ever talking with parents about HIV/AIDS was also associated with HIV testing among young adults 15–24 (MacPhail, Pettifor, Moyo, & Rees, 2009).
Methods Data Data came from the 2012 National AIDS Communication Survey of 10,078 households across South Africa. Multistage cluster sampling was used to select households; one eligible respondent (a man or a woman 16–55 years old) was randomly selected from each household. A structured questionnaire was used to collect information on exposure to communication programs, as well as HIV-related knowledge, beliefs, norms, attitudes, and behaviors. This analysis is limited to 6004 men and women aged 16–55 who reported having sexual relations within 12 months before the survey. Ethical approval was obtained from the Ethics Committee at the University of Witwatersrand in Johannesburg. An Institutional Review Board/Independent Ethics Committee Authorization Agreement was signed between University of Witwatersrand and the Johns Hopkins University for the review and approval of this study.
Measurement A measure of joint exposure to the communication programs was used. For each program, a series of questions were asked to assess respondent’s unaided recall of a program’s name and logo, recognition of still images of characters and animerts (i.e., an animated scene from Scrutinize), and correct interpretation of relevant health messages. Participants did not have to have watched any program to be included in the survey. Principle component analysis was used to construct a single measure of exposure (the standardized coefficient of internal reliability was .71) from yes/no and correct/incorrect responses. To have a reasonable range to estimate the programs’ dose-response effects, this measure was collapsed into 10 levels, beginning with no exposure at all (22% of the sample), followed by nine equal intervals. Discussion about HIV testing with partners and friends was measured as any positive response to three questions about the 12 months before the survey: “Did you talk with anyone you had sex with about getting tested for HIV?,” “Has anyone you had sex with ever asked you to get an HIV test?,” and “Have you asked anyone who you had sex with to get an HIV test?” Perceived prevalence of HIV testing among one’s friends was constructed from responses to the question “[Out of ten] About how many of your friends do you think have been tested for HIV?” High perceived testing prevalence was defined as having two or more (the median number) friends that had tested (52.8%). Lastly, HIV testing in the last 12 months came from the question “How long ago was your last HIV test?,” which was asked among those who had tested for HIV. Participants were grouped into those receiving a test during the 12 months before the survey and those who had a test more than 12 months ago or never had one (52.8%). We hypothesize that communication programs have both direct and indirect impacts on HIV testing by improving perceived norms of HIV testing and encouraging discussions about it (Kincaid, 2000a, 2000b; Kincaid et al., 2012). Following the spiral of silence theory (Noelle-Neumann, 1974, 1991), we hypothesize that perceived norm related to HIV testing influences the likelihood of discussing HIV testing with sex partners and friends. We controlled for a number of variables that may confound the relationships between communication exposure and the outcomes, including sociodemographic characteristics, media exposure, and factors related to HIV and individual sexual behavior. socio-economic status (SES) was constructed based on the ownership of seven household items, ranging from running water to vehicles (internal reliability coefficient = .82), and poverty was constructed based on the frequency that a
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care household was out of fuel, clean water, medicine, and food within the last 12 months (internal reliability coefficient = .80). High scores of SES and poverty indicate an individual being better off. Media exposure measures, proxies of exposure to other mass communication programs, include the frequency of media contact, and the use of social networks (Facebook and Twitter). HIV-related factors include community-level norms of HIV testing, whether an individual was a leader of a HIV-related community group, whether he/she attended a community meeting that discussed HIV, or heard HIV being discussed in church in the last 12 months, as well as perceived social capital for HIV in the community. Community-level norm of HIV testing is aggregated from individual responses to the aforementioned question on the perceived number of friends having tested for HIV, excluding the index respondent’s response. High norm of HIV testing in the community was defined as more than 50% of one’s friends were thought to have tested for HIV. Perceived social capital for HIV measures the perceived support for people with HIV in the community, which came from responses to five questions, ranging from whether HIV was taken seriously to whether people joined together to help those with HIV/ AIDS (Cronbach alpha = .88). Finally, we controlled for a set of factors that may be related to one’s sexual behavior, which may be associated with HIV preventive measures. These include whether an individual had slept away from home in the last 12 months, whether he/she had sex with someone they just met at a bar, the number of sex partners, and whether they had casual multiple partnerships. Individuals were also asked if any of their partners had other sex partners and whether they expected to have sex with the same partners again. Statistical analysis Multiple causal attribution (MCA) analysis (Babalola & Kincaid, 2009; Kincaid & Do, 2006) was used to justify causal reference and estimate the impact of communication programs on HIV testing. MCA is appropriate under the following conditions: a population-level intervention has been implemented that can be evaluated by a survey of the population after it has occurred; an appropriate theory of causality is assumed; the intervention is based on appropriate causal theories of change; and the statistical requirements for a causal inference have been met. In this case, the intervention (i.e., communication programs) already took place, followed by a populationbased survey; the relevant theories have been discussed above. SEM analysis statistically tests the directions and causal pathways between communication exposure and dependent variables. Multivariate probit regression tests the intervening causal pathway of the third variable
1111
between communication exposure and HIV testing. R-squared was used to indicate the goodness of fit of ordinary least square regressions; Hosmer–Lemeshow tests were used to examine the goodness of fit of logistic regressions. For the proper identification of potentially endogenous variables, a distinct set of variables was excluded from each equation on theoretical and empirical grounds. Frequencies of watching SABC1, ETV, and Top TV were hypothesized to be related only to communication exposure and excluded from the other three equations. Similarly, frequency of listening to the radio was hypothesized to be only related to perceived norms of HIV testing; exposure to SABC3 TV, employment, and whether one had a same sex partner were hypothesized to be only associated with talking about HIV testing. Finally, the community-level norm of HIV testing was hypothesized to be related only to HIV testing behavior. Exclusion tests (Hausman tests for the continuous measure of program exposure and log-likelihood ratio tests for the three binary outcomes) were used to assess if such exclusions significantly change the models. All analyses were performed with Stata 12 (StataCorp, 2011). Sample weights were applied to all descriptive statistics. Multiple regressions were not weighted because we already controlled for variables that were used to calculate sample weights (urban/rural and province) and to allow for statistical tests as part of the procedure. The final regressions were then adjusted for design effects using Stata’s survey (svy) commands.
Results Sample distribution The average exposure to the four communication programs was 4.16 on a scale from 0 to 9. Just over half of the sample believed that many of their friends were tested for HIV. Less than half (47%) had talked about HIV testing with friends and sex partners; the same proportion of respondents had tested for HIV in the last 12 months (Table 1). Nearly 60% of respondents were female; 70% lived in urban areas. Three-quarters of the sample were either married or in a steady relationship. Most people had completed primary or attained matric level of schooling. Mean SES was 4.23 on a 0–7 scale and mean poverty was 1.5 on a 1–4 scale. The sample was heavily concentrated in Gauteng and Kwazulu-Natal, while provinces like Northern Cape contributed only a small proportion of the sample. There are some variations in joint communication program exposure and the three HIV-related outcomes by socioeconomic characteristics. For example,
1112
M. Do et al.
Table 1. Sample distribution and differences in communication program exposure and outcomes by socioeconomic characteristics, South Africa, 2012.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Gender Male Female Race Black Colored White Indian Type of settlement Rural Urban Marital status Married/living together Widow/Divorced Steady Single Age groups 16–24 25–35 36–55 Education level Up to primary Primary to standard nine Matric Tertiary or above SES Poverty level Employment status Unemployed Employed Student Province Gauteng Western Cape Eastern Cape Free State Kwazulu-Natal Limpopo Mpumalanga North West Northern Cape Total
Joint communication exposure mean (SD)
High perceived no. of friends got tested (%)
Talking about testing (%)
HIV testing last 12 months (%)
40.52 59.48
3.82 (.12) 4.39 (.10)
54.91 51.30
41.88 51.02
38.02 53.77
80.99 13.59 1.68 1.73
4.42 3.14 2.36 1.68
(.10) (.17) (.38) (.41)
51.01 64.46 50.50 47.12
49.89 36.27 38.61 20.19
48.16 48.04 33.66 17.31
30.73 69.27
3.83 (.16) 4.31 (.13)
46.45 55.57
45.47 48.14
47.59 47.30
40.24 2.35 25.98 31.43
3.66 3.59 4.58 4.49
(.25) (.27) (.14) (.12)
53.92 60.28 54.21 52.44
53.92 43.97 44.36 51.47
50.49 41.84 47.85 48.27
29.76 40.87 29.37
4.66 (.12) 4.36 (.11) 3.38 (.11)
51.53 54.40 51.73
46.11 50.04 44.75
51.59 48.00 42.26
9.29 41.12 36.31 13.27 4.23 (2.13) 1.51 (.65)
1.85 3.80 4.80 5.16
(.11) (.11) (.13) (.15)
38.17 49.57 56.24 63.36
33.87 45.32 50.83 53.32
37.63 48.40 48.85 47.05
52.93 36.59 8.99
4.03 (.10) 4.11 (.11) 5.16 (.15)
50.38 56.90 50.93
46.73 49.25 43.52
48.36 45.15 50.37
21.97 14.36 11.23 6.80 20.45 8.93 8.06 6.38 1.83
4.34 3.26 6.27 4.79 3.24 3.64 4.93 3.89 3.38 4.16
59.51 57.86 55.88 49.58 55.46 33.77 48.76 57.70 58.18 52.76
45.82 28.93 67.40 55.12 46.25 37.50 51.86 49.35 36.36 47.32
51.39 42.14 59.80 45.94 44.54 45.90 63.22 32.38 40.91 47.39
Distribution mean (SD) or % (weighted)
(.25) (.25) (.22) (.25) (.18) (.27) (.17) (.32) (.46) (.10)
communication exposure seemed higher among blacks and more educated respondents compared to others. Perceived prevalence of HIV testing seemed higher among respondents in urban and with more education. Respondents who were female, black, married, and more educated were more likely to talk about HIV testing than others. Finally, HIV testing seemed more common
among female and black and colored respondents, as well as those with more than primary schooling. Table 2 (column 1) shows that several individual factors, including having attended a community meeting where HIV was discussed, having heard of HIV being discussed at church in the last 12 months, and increased perceived social capital for HIV in the community were
AIDS Care
1113
Table 2. Mulitple regression analysis of the joint exposure to four communication programs, talking about HIV testing with partners and friends, perceived social norms related to testing, and getting tested for HIV, South Africa, 2012.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Joint communication exposure High perceived no. of friends got tested Talking about HIV testing Gender Male Female Race Black Colored White Indian Type of settlement Rural Urban Marital status Married/living together Widow/divorced Steady Single Age groups 16–24 25–35 36–55 Education level Up to primary Primary to standard nine Matric Tertiary or above SES Poverty level Employment status Unemployed Employed Student Norm of HIV testing in community Leader of community group related to HIV Attended community meeting that discussed HIV Heard HIV discussed in church last 12 months Perceived social capital for HIV in community Ever slept away from home last 12 months Had sex with someone just met at bar Number of partners Had casual MCP relationship Any partner had other sex partners Any partner of same sex Expected to have sex again with a partner Frequency of watching SABC1 Frequency of watching SABC2 Frequency of watching SABC3
Joint communication exposure
High perceived no. of friends got tested
Talking about testing
HIV testing last 12 months
Coef. (SD) (1)
OR (SD) (2)
OR (SD) (3)
OR (SD) (4)
1.05 (.01)***
1.08 (.01)*** 1.50 (.10)***
1.03 (.01)* 1.25 (.09)** 3.47 (.21)***
1.00 .89 (.07)
1.00 1.36 (.09)***
1.00 1.56 (.10)***
1.00 1.17 (.26) .40 (.12)** .66 (.20)
1.00 .68 (.14) .87 (.32) .49 (.19)
1.00 1.25 (.27) .83 (.27) .63 (.24)
.10 (.09)
1.00 1.11 (.09)
1.00 .97 (.08)
1.00 1.10 (.09)
–.09 (.24) .28 (.09)** .35 (.10)**
1.00 1.38 (.25) 1.00 (.09) .98 (.09)
1.00 .90 (.17) 1.06 (.09) .88 (.08)
1.00 .82 (.16) .83 (.07)* .94 (.09)
.85 (.08) .98 (.07) 1.00
1.08 (.10) 1.12 (.10) 1.00
1.46 (.14)*** 1.16 (.10) 1.00
1.00 1.54 1.93 2.27 1.01 1.04
1.00 1.30 1.43 1.41 .99 1.22
1.00 1.21 1.19 1.20 .96 1.00
(.15) (.15) (.19) (.02)* (.06)
1.69 .87 .99 1.09 1.02 .91 .71 .90 .86 1.22
(.17)*** (.10) (.10) (.08) (.03) (.06) (.18) (.08) (.08) (.10)*
.48 (.09)*** –.90 (.27)** –1.94 (.40)*** –1.37 (.42)**
.59 (.10)*** .43 (.09)***
.77 1.18 1.29 .10 –.13
(.12)*** (.13)*** (.15)*** (.03)*** (.08)
(.17)*** (.23)*** (.32)*** (.02) (.05)
(.15)* (.18)** (.23)* (.02) (.07)***
1.00 1.16 (.08)* .84 (.09) –.15 .31 .18 .10 .16 –.16 .26 –.14 –.21
(.14) (.10)** (.09)* (.04)* (.08) (.28) (.11)* (.11) (.09)*
–.15 (.14) .40 (.04)*** .18 (.04)***
1.55 1.00 1.18 1.09 1.21 .99 .96 1.22 1.10
(.21)** (.08) (.09)* (.03)** (.08)** (.22) (.09) (.13) (.09)
1.29 (.13)* 1.00 (.02)
1.81 1.44 1.26 1.12 1.11 .80 1.25 .68 .86 .67 1.49
(.26)*** (.13)*** (.11)** (.03)** (.08) (.21) (.12)* (.06)*** (.07) (.13)* (.18)**
1.12 (.03)*** .92 (.03)**
.80 (.10) 1.03 (.02)
1114
M. Do et al.
Table 2 (Continued)
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Frequency of watching e TV Frequency of watching Top TV Frequency of listening to the radio Exclusion testa Chi-squared (df) Probability > χ2 Exogeneity tests between communication and outcomes Exogeneity tests between outcomes (2), (3) and (4) Rho Probability > χ2 Goodness-of-fit R2 Hosmer-Lemeshow χ2 (df) Probability > χ2
Joint communication exposure
High perceived no. of friends got tested
Talking about testing
HIV testing last 12 months
Coef. (SD) (1)
OR (SD) (2)
OR (SD) (3)
OR (SD) (4)
5.33 (5) .38 .93 (.05)
6.94 .54 1.00 (.05)
.28 (.04)*** .17 (.06)** .96 (.02)* 7.41 (53) 1.00
9.12 (8) .33 1.04 (.05) .02 .06
.07 .06
–.01 .06
.43 5.57 (8) .70
14.89 (8) .06
20.25 (8) .01
Note: All models controlled for language, frequency of watching DSTV, reading newspapers, magazines, using the internet, and ever use of Facebook and Twitter. a Hausman test was used for outcome (1) and log-likelihood tests were used for outcomes (2) through (4). *p < .05; **p < .01; ***p < .001.
related to increased exposure to programs, as was increased number of sex partners (p < .05). Very few factors were negatively associated with joint program exposure, including race and multiple partnerships of a respondent’s partner.
Effects of communication programs Table 2 also shows the effects of joint program exposure on two intermediate and the behavioral outcomes. Column 2 indicates a significant effect of program exposure on increased perception of friends getting tested (OR = 1.05; p < .001). As shown in Column 3, one’s perception of HIV testing prevalence, in turn, had a positive effect on the likelihood of talking about testing with friends and partners (OR = 1.50; p < .001). In addition, joint communication exposure had a direct, positive effect on one’s likelihood of talking about HIV testing (OR = 1.08; p < .001). Finally, Column 3 shows a direct, positive effect of joint program exposure on having tested in the last 12 months (OR = 1.03; p < .05); high perceived testing norm and talking about HIV testing both also had positive effects on HIV testing. Those who perceived a high level of testing among their friends were 25% more likely than others to have tested (p < .01), but talking about HIV testing was the strongest predictor of testing: those who had discussed HIV tests with friends and
partners were more than three times as likely as others to have tested in the last 12 months (p < .001). Figure 1 summarizes the direct and indirect effects of communication programs on HIV testing; 55% of the effect of communication on HIV testing was indirect through these mechanisms (results not shown). Goodness-of-fit tests (shown in Table 2) indicate fair to high levels of goodness for most equations. Variables excluded from each equation were already discussed above. Exclusion tests showed that their exclusions did not affect the results. Exogeneity tests (see Table 2) suggested that joint communication exposure was exogenous to all the three outcomes; there was also no evidence of endogeneity between perceived HIV testing norm, HIV testing discussion, and testing behavior. Finally, we estimated the dose-response effects of joint program exposure. For all outcomes, the adjusted probability of the outcomes increased with increasing levels of communication program exposure. For HIV testing behavior, the difference between no exposure and the highest level of exposure was 4.8 percentage points (results not shown). Discussion This paper examines the impact of four communication programs on HIV testing behavior among South African men and women aged 16–55. Less than half of the
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care
1115
Figure 1. Path model of the impact of four communication programs on HIV testing.
survey respondents had tested for HIV within 12 months before the survey. Findings indicate significant, positive effects of the programs on HIV testing, partly due to indirect effects through increasing an individual’s perception of HIV testing prevalence and the probability of talking about testing with partners and friends. Dose responses in the three outcomes were found with increasing program exposure. The findings are consistent with the earlier studies that found positive effects of communication programs on HIV-related norms, discussion with others, and testing in South Africa and elsewhere in sub-Saharan Africa (Goldstein, Usdin, Scheepers, & Japhet, 2005; Hutchinson, Mahlalela, & Yukich, 2007; Keating, Meekers, & Adewuyi, 2006; Pettifor et al., 2007). Although cross-sectional data were used, statistical tests provide evidence that communication exposure and the three outcomes were exogenous, hence, an SEM analysis could be used. It should be noted that although the Chi-square goodness-of-fit test is usually sensitive to large sample sizes (Hooper, Coughlan, & Mullen, 2008), it shows a statistical lack of fit for only one of the models, giving us more confidence that the models fit the data well. The results underline the importance of promoting discussion around HIV tests with friends and partners. Talking about it may contribute to reducing stigma and encourage greater acceptance of one’s HIV status. Communication programs should also target those less educated, who may be unlikely to be exposed to
such programs and to talk about HIV testing. In addition, community-based interventions such as community and religious meetings that discuss HIV may also be effective in promoting HIV testing discussion. Perceived norm of HIV testing among friends mediates the effects of communication programs on HIV testing not only by its direct effects on testing behavior but also by its indirect effects through increasing discussions about HIV testing. It suggests that interventions to promote HIV testing should address social barriers, in addition to individual-level barriers to HIV testing. The finding that community group leadership and hearing HIV discussions in church both were related to a higher perceived HIV testing prevalence underscores the importance of integrated interventions that rely on community activities and the mass media. Finally, the analysis revealed a lower level of exposure to the four communication programs among those who are male, married, or living with a partner, older, less educated, and have a lower socioeconomic status. Future programs should be tailored to address their particular motivations and barriers. On the other hand, the finding that those with more sex partners were more exposed to communication programs and more likely to talk about HIV testing is promising. It is possible that individuals who had multiple partners were somewhat aware of HIV risks and paying more attention to the TV series. The result suggests that these programs might have reached and had an impact on
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
1116
M. Do et al.
those who may be at increased risk of HIV infection. In general, male and older individuals should also be targeted because they were less likely to get tested, due in part to their lower likelihood of using health care services in general. This study is limited to individuals who had sex 12 months before the survey. Although our analysis shows no differences between those included in the study and those who were not in terms of SES, poverty, race, or urban residence, several differences exist between the two groups. For example, those included in the sample were more likely than those who were not to be male, in a relationship, more educated, and employed; they were also likely to be older than those excluded (results not shown). Therefore, results of this study are only applicable to South African men and women who were sexually active within the last year and fit the characteristics of the sample. Another limitation is potential recall bias related to the program exposure measure. However, we believe it is not a major problem as our measure of communication exposure was constructed based on responses to several questions, using multiple visual aids to minimize social desirability and acquiescence response set. Talking about HIV testing was measured by a dichotomous indicator; the frequency and content of such conversations are unknown. It is not clear if individuals talked about HIV as a general or a personal issue, or if they talked about it in the context of the communication programs. For example, conversations about HIV among friends may center on rumors about who might be infected rather than care and prevention related topics (Smith, Lucas, & Latkin, 1999). Some of the friends and partners may also be opinion leaders, who may be more influential than others in encouraging one to get tested (Sivaram et al., 2005). It is not possible to know in this assessment which aspects of the conversations may be more effective than others in promoting HIV testing. Conclusions Notwithstanding the limitations, this paper suggests that communication programs can be powerful in promoting HIV testing. Investment in behavior change communication programs may have long-term effects through changes in norms and perceptions related to the desired behavior. References Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. doi:10.1016/0749-5978(91)90020-T Babalola, S. (2007). Readiness for HIV testing among young people in northern Nigeria: The roles of social norm and perceived stigma. AIDS and Behavior, 11, 759–769. doi:10.1007/s10461-006-9189-0
Babalola, S., & Kincaid, D. L. (2009). New methods for estimating the impact of health communication programs. Communication Methods and Measures, 3(1–2), 61–83. doi:10.1080/19312450902809706 Baiden, F., Remes, P., Baiden, R., Williams, J., Hodgson, A., Boelaert, M., & Buve, A. (2005). Voluntary counseling and HIV testing for pregnant women in the KassenaNankana district of northern Ghana: Is couple counseling the way forward? AIDS Care, 17, 648–657. doi:10.1080/ 09540120412331319688 Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: PrenticeHall. Bertrand, J. T., O’Reilly, K., Denison, J., Anhang, R., & Sweat, M. (2006). Systematic review of the effectiveness of mass communication programs to change HIV/AIDS-related behaviors in developing countries. Health Education Research, 21, 567–597. doi:10.1093/her/cyl036 CADRE. (2011). 4Play: DVD discussion guide. Talking about sex, love and relationships. Johannesburg: JHHESA. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117–140. doi:10.1177/001872675 400700202 Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. Gage, A. J., & Ali, D. (2005). Factors associated with selfreported HIV testing among men in Uganda. AIDS Care, 17, 153–165. doi:10.1080/09540120512331325635 Goldstein, S., Usdin, S., Scheepers, E., & Japhet, G. (2005). Communicating HIV and AIDS, what works? A report on the impact evaluation of Soul City’s fourth series. Journal of Health Communication, 10, 465–483. doi:10.1080/ 10810730591009853 Hendriksen, E. S., Hlubinka, D., Chariyalertsak, S., Chingono, A., Gray, G., Mbwambo, J., … Coates, T. J. (2009). Keep talking about it: HIV/AIDS-related communication and prior HIV testing in Tanzania, Zimbabwe, South Africa, and Thailand. AIDS and Behavior, 13, 1213–1221. doi:10.1007/s10461-009-9608-0 Hooper, D., Coughlan, J., & Mullen, M. (2008). Structural equation modelling: Guidelines for determining model fit. Articles, p. 2. Hutchinson, P., Mahlalela, X., & Yukich, J. (2007). Mass media, stigma, and disclosure of HIV test results: Multilevel analysis in the Eastern Cape, South Africa. AIDS Education & Prevention, 19, 489–510. doi:10.1521/ aeap.2007.19.6.489 Katz, E., & Lazarsfeld, P. F. (2006). Personal influence: The part played by people in the flow of mass communications. New Brunswick, NJ: Transaction. Keating, J., Meekers, D., & Adewuyi, A. (2006). Assessing effects of a media campaign on HIV/AIDS awareness and prevention in Nigeria: Results from the VISION Project. BMC Public Health, 6(1), 123. doi:10.1186/1471-2458-6-123 Kincaid, D. L. (2000a). Mass media, ideation, and contraceptive behavior. Communication Research, 27, 723–763. doi:10.1177/009365000027006003 Kincaid, D. L. (2000b). Social networks, ideation, and contraceptive behavior in Bangladesh: A longitudinal analysis. Social Science and Medicine, 50, 215–231. doi:10.1016/ S0277-9536(99)00276-2 Kincaid, D. L. (2009). Convergence theory. In S. W. Littlejohn & K. A. Foss (Eds.), Encyclopedia of communication theory (pp. 188–191). Thousand Oaks, CA: Sage.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care Kincaid, D. L., Delate, R., Storey, D., & Figueroa, M. E. (2012). Closing the gaps in practice and in theory. Public Communication Campaigns, 305, 305–319. Kincaid, D. L., & Do, M. P. (2006). Multivariate causal attribution and cost-effectiveness of a national mass media campaign in the Philippines. Journal of Health Communication, 11(S2), 69–90. doi:10.1080/1081073 0600974522 Kranzer, K., McGrath, N., Saul, J., Crampin, A. C., Jahn, A., Malema, S., … Glynn, J. R. (2008). Individual, household and community factors associated with HIV test refusal in rural Malawi. Tropical Medicine & International Health, 13, 1341–1350. doi:10.1111/j.1365-3156.2008.02148.x MacPhail, C., Pettifor, A., Moyo, W., & Rees, H. (2009). Factors associated with HIV testing among sexually active South African youth aged 15–24 years. AIDS Care, 21, 456–467. doi:10.1080/09540120802282586 McKee, N., Bertrand, J. T., Becker-Benton, A., & Becker, B. L. (2004). Strategic communication in the HIV/AIDS epidemic. New Delhi, Thousand Oaks, CA and London: Sage. Noar, S. M., Palmgreen, P., Chabot, M., Dobransky, N., & Zimmerman, R. S. (2009). A 10-year systematic review of HIV/AIDS mass communication campaigns: Have we made progress? Journal of Health Communication, 14 (1), 15–42. doi:10.1080/10810730802592239 Noelle-Neumann, E. (1991). The theory of public opinion: The concept of the spiral of silence. In J. A. Anderson (Ed.), Communication yearbook (Vol. 14, pp. 188–191). Thousand Oaks, CA: Sage. Noelle-Neumann, E. (1993). The spiral of silence: Public opinion – Our social skin. Chicago: University of Chicago Press. Noelle-Neumann, E. (1974). The spiral of silence a theory of public opinion. Journal of Communication, 24(2), 43–51. doi:10.1111/j.1460-2466.1974.tb00367.x
1117
Nuwaha, F., Kabatesi, D., Muganwa, M., & Whalen, C. (2006). Factors influencing acceptability of voluntary counseling and testing for HIV in Bushenyi district of Uganda. East African Medical Journal, 79, 626–632. Pettifor, A. E., MacPhail, C., Bertozzi, S., & Rees, H. V. (2007). Challenge of evaluating a national HIV prevention programme: the case of lovelife, South Africa. Sexually Transmitted Infections, 83, i70–i74. doi:10.1136/sti.2006. 023689 Rimal, R. N., Limaye, R. J., Roberts, P., Brown, J., & Mkandawire, G. (2013). The role of interpersonal communication in reducing structural disparities and psychosocial deficiencies: Experience from the Malawi BRIDGE project. Journal of Communication, 63(1), 51–71. Rogers, E. (2003). Diffusion of innovations. New York, NY: Free Press. Sarker, M., Sanou, A., Snow, R., Ganame, J., & Gondos, A. (2007). Determinants of HIV counselling and testing participation in a prevention of mother to child transmission programme in rural Burkina Faso. Tropical Medicine & International Health, 12, 1475–1483. doi:10.1111/ j.1365-3156.2007.01956.x Sivaram, S., Johnson, S., Bentley, M. E., Go, V. F., Latkin, C., Srikrishnan, A., … Solomon, S. (2005). Sexual health promotion in Chennai, India: Key role of communication among social networks. Health Promotion International, 20, 327–333. doi:10.1093/heapro/dai012 Smith, L. C., Lucas, K. J., & Latkin, C. (1999). Rumor and gossip: Social discourse on HIV and AIDS. Anthropology & Medicine, 6(1), 121–131. doi:10.1080/13648470.1999.9964577 StataCorp. (2011). Stata statistical software: Release 12. College Station, TX: StataCorp LP. Suls, J., Martin, R., & Wheeler, L. (2002). Social comparison: Why, with whom, and with what effect? Current Directions in Psychological Science, 11, 159–163. doi:10.1111/ 1467-8721.00191
AIDS Care: Psychological and Socio-medical Aspects of AIDS/HIV Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/caic20
Impacts of four communication programs on HIV testing behavior in South Africa a
b
b
Mai Do , D. Lawrence Kincaid & Maria Elena Figueroa a
Department of Global Health Systems and Development, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA b
Center for Communication Programs, Johns Hopkins University, Baltimore, MD, USA Published online: 07 Apr 2014.
Click for updates To cite this article: Mai Do, D. Lawrence Kincaid & Maria Elena Figueroa (2014) Impacts of four communication programs on HIV testing behavior in South Africa, AIDS Care: Psychological and Socio-medical Aspects of AIDS/HIV, 26:9, 1109-1117, DOI: 10.1080/09540121.2014.901487 To link to this article: http://dx.doi.org/10.1080/09540121.2014.901487
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
AIDS Care, 2014 Vol. 26, No. 9, 1109–1117, http://dx.doi.org/10.1080/09540121.2014.901487
Impacts of four communication programs on HIV testing behavior in South Africa Mai Doa*, D. Lawrence Kincaidb and Maria Elena Figueroab a Department of Global Health Systems and Development, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA; bCenter for Communication Programs, Johns Hopkins University, Baltimore, MD, USA
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
(Received 15 July 2013; accepted 3 March 2014) This paper aims to evaluate the impacts of four communication programs on promoting HIV testing behavior among sexually active individuals in South Africa. The four programs, implemented by Johns Hopkins Health and Education in South Africa, are aimed to promote HIV prevention behaviors, as well as gender-based violence prevention, tuberculosis screening and treatment, and reduction of alcohol consumption. Launched between 2009 and 2010, they all promoted HIV testing. Data came from the population-based Third National AIDS Communication Survey 2012; 6004 men and women who had sex within the last 12 months were included in the analysis. Multiple causal attribution analysis is used to justify causal reference and estimate the impact of communication programs. Findings indicate significant direct and indirect effects of the programs on HIV testing behavior. Indirect effects worked through increasing one’s likelihood of perceiving that their friends were tested and the probability of talking about HIV testing with sex partners and friends, which in turn increased the likelihood of HIV testing. Findings suggest multiple angles from which communication programs can promote HIV testing. The study also demonstrates the use of multiple statistical techniques for causal attribution in a post-only design, where randomization is not possible. Keywords: communication; HIV testing; multiple causal attribution; South Africa
Introduction
Theory of mass media effects
Since the beginning of the HIV epidemic, the mass media programs have proven important in increasing knowledge and promoting preventive practices. They have recently begun to address a full continuum of prevention, treatment, care, and support strategies (McKee, Bertrand, Becker-Benton, & Becker, 2004). Yet few studies have examined the impact of mass media programs on HIV testing behavior in low- and middleincome countries. Two recent reviews of communication programs in developing countries did not report any studies with HIV testing as an intended outcome of mass media communication (Bertrand, O’Reilly, Denison, Anhang, & Sweat, 2006; Noar, Palmgreen, Chabot, Dobransky, & Zimmerman, 2009); an exception is Pettifor, MacPhail, Bertozzi, and Rees (2007). This paper aims to evaluate the impact of communication programs in promoting HIV testing behavior in South Africa. Since 1999, the Johns Hopkins Health and Education in South Africa (JHHESA) has been providing technical support on strategic communication for HIV prevention, care, support, and treatment. Four recent programs (Intersexions, Brothers for Life, 4Play: Sex Tips for Girls, and Scrutinize), launched between 2009 and 2010, are included in this assessment – all are TV series and include messages that promote HIV testing (CADRE, 2011; http://www.brothersforlife.org; http://www.intersexions-tv.co.za/; http://www.jhuccp.org/).
Mass media programs initiate causal processes that occur in several stages. Information is expected to affect ideational factors such as knowledge, beliefs, and values (attitudes); self-efficacy; and interpersonal communication about it (Ajzen, 1991; Fishbein & Ajzen, 1975; Kincaid, 2009; Kincaid, Delate, Storey, & Figueroa, 2012). By observing the behavior of others in the media (mediated social modeling), one learns about new behaviors in a way that later guides action in similar situations (Bandura, 1986). The cognitive framework within which information is perceived determines how it is interpreted, understood, and acted upon (Kincaid et al., 2012). Framing creates a point of view that highlights certain facts to be interpreted in a particular manner. A television drama can reframe HIV-positive status from one of sexual promiscuity, irresponsibility, and immorality to that of a health condition that requires medical treatment and taking personal responsibility to protect oneself and others. Dialogs can lead to reframing the issues and making them more relevant. Discussion with others also produces a multistep process of mass media effects (Katz & Lazarsfeld, 2006) and accounts for the well-documented diffusion effect (Rogers, 2003). Discussion also allows for social comparison with one’s friends in order to reduce uncertainty about the issue and to obtain social
*Corresponding author. Email: [email protected] © 2014 Taylor & Francis
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
1110
M. Do et al.
and emotional support for change (Festinger, 1954; Suls, Martin, & Wheeler, 2002). Multistep effects are slow to occur, however, if potential initial adopters believe that the new position is not the social norm and are thus reluctant to talk about it with others for fears of being criticized, according to the spiral of silence theory (Noelle-Neumann, 1991, 1993). Mass media can be used to frame a new behavior as a new social norm, reducing social stigma, and thereby increasing the likelihood of talking about it with others. All the direct ideational changes that result from mass media exposure also result from this second stage of interpersonal communication. This paper examines this multistep process – whether communication programs could change existing norms and the likelihood of discussing HIV-related topics, thereby increasing HIV testing. Studies have shown that social norms related to testing and HIV-related communication were important determinants of the readiness for testing and testing behavior (Babalola, 2007; Baiden et al., 2005; Gage & Ali, 2005; Hendriksen et al., 2009; Kranzer et al., 2008; Nuwaha, Kabatesi, Muganwa, & Whalen, 2006; Rimal, Limaye, Roberts, Brown, & Mkandawire, 2013). Communications with sex partners about HIV testing was key for its uptake in Burkina Faso (Sarker, Sanou, Snow, Ganame, & Gondos, 2007). In South Africa, ever talking with parents about HIV/AIDS was also associated with HIV testing among young adults 15–24 (MacPhail, Pettifor, Moyo, & Rees, 2009).
Methods Data Data came from the 2012 National AIDS Communication Survey of 10,078 households across South Africa. Multistage cluster sampling was used to select households; one eligible respondent (a man or a woman 16–55 years old) was randomly selected from each household. A structured questionnaire was used to collect information on exposure to communication programs, as well as HIV-related knowledge, beliefs, norms, attitudes, and behaviors. This analysis is limited to 6004 men and women aged 16–55 who reported having sexual relations within 12 months before the survey. Ethical approval was obtained from the Ethics Committee at the University of Witwatersrand in Johannesburg. An Institutional Review Board/Independent Ethics Committee Authorization Agreement was signed between University of Witwatersrand and the Johns Hopkins University for the review and approval of this study.
Measurement A measure of joint exposure to the communication programs was used. For each program, a series of questions were asked to assess respondent’s unaided recall of a program’s name and logo, recognition of still images of characters and animerts (i.e., an animated scene from Scrutinize), and correct interpretation of relevant health messages. Participants did not have to have watched any program to be included in the survey. Principle component analysis was used to construct a single measure of exposure (the standardized coefficient of internal reliability was .71) from yes/no and correct/incorrect responses. To have a reasonable range to estimate the programs’ dose-response effects, this measure was collapsed into 10 levels, beginning with no exposure at all (22% of the sample), followed by nine equal intervals. Discussion about HIV testing with partners and friends was measured as any positive response to three questions about the 12 months before the survey: “Did you talk with anyone you had sex with about getting tested for HIV?,” “Has anyone you had sex with ever asked you to get an HIV test?,” and “Have you asked anyone who you had sex with to get an HIV test?” Perceived prevalence of HIV testing among one’s friends was constructed from responses to the question “[Out of ten] About how many of your friends do you think have been tested for HIV?” High perceived testing prevalence was defined as having two or more (the median number) friends that had tested (52.8%). Lastly, HIV testing in the last 12 months came from the question “How long ago was your last HIV test?,” which was asked among those who had tested for HIV. Participants were grouped into those receiving a test during the 12 months before the survey and those who had a test more than 12 months ago or never had one (52.8%). We hypothesize that communication programs have both direct and indirect impacts on HIV testing by improving perceived norms of HIV testing and encouraging discussions about it (Kincaid, 2000a, 2000b; Kincaid et al., 2012). Following the spiral of silence theory (Noelle-Neumann, 1974, 1991), we hypothesize that perceived norm related to HIV testing influences the likelihood of discussing HIV testing with sex partners and friends. We controlled for a number of variables that may confound the relationships between communication exposure and the outcomes, including sociodemographic characteristics, media exposure, and factors related to HIV and individual sexual behavior. socio-economic status (SES) was constructed based on the ownership of seven household items, ranging from running water to vehicles (internal reliability coefficient = .82), and poverty was constructed based on the frequency that a
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care household was out of fuel, clean water, medicine, and food within the last 12 months (internal reliability coefficient = .80). High scores of SES and poverty indicate an individual being better off. Media exposure measures, proxies of exposure to other mass communication programs, include the frequency of media contact, and the use of social networks (Facebook and Twitter). HIV-related factors include community-level norms of HIV testing, whether an individual was a leader of a HIV-related community group, whether he/she attended a community meeting that discussed HIV, or heard HIV being discussed in church in the last 12 months, as well as perceived social capital for HIV in the community. Community-level norm of HIV testing is aggregated from individual responses to the aforementioned question on the perceived number of friends having tested for HIV, excluding the index respondent’s response. High norm of HIV testing in the community was defined as more than 50% of one’s friends were thought to have tested for HIV. Perceived social capital for HIV measures the perceived support for people with HIV in the community, which came from responses to five questions, ranging from whether HIV was taken seriously to whether people joined together to help those with HIV/ AIDS (Cronbach alpha = .88). Finally, we controlled for a set of factors that may be related to one’s sexual behavior, which may be associated with HIV preventive measures. These include whether an individual had slept away from home in the last 12 months, whether he/she had sex with someone they just met at a bar, the number of sex partners, and whether they had casual multiple partnerships. Individuals were also asked if any of their partners had other sex partners and whether they expected to have sex with the same partners again. Statistical analysis Multiple causal attribution (MCA) analysis (Babalola & Kincaid, 2009; Kincaid & Do, 2006) was used to justify causal reference and estimate the impact of communication programs on HIV testing. MCA is appropriate under the following conditions: a population-level intervention has been implemented that can be evaluated by a survey of the population after it has occurred; an appropriate theory of causality is assumed; the intervention is based on appropriate causal theories of change; and the statistical requirements for a causal inference have been met. In this case, the intervention (i.e., communication programs) already took place, followed by a populationbased survey; the relevant theories have been discussed above. SEM analysis statistically tests the directions and causal pathways between communication exposure and dependent variables. Multivariate probit regression tests the intervening causal pathway of the third variable
1111
between communication exposure and HIV testing. R-squared was used to indicate the goodness of fit of ordinary least square regressions; Hosmer–Lemeshow tests were used to examine the goodness of fit of logistic regressions. For the proper identification of potentially endogenous variables, a distinct set of variables was excluded from each equation on theoretical and empirical grounds. Frequencies of watching SABC1, ETV, and Top TV were hypothesized to be related only to communication exposure and excluded from the other three equations. Similarly, frequency of listening to the radio was hypothesized to be only related to perceived norms of HIV testing; exposure to SABC3 TV, employment, and whether one had a same sex partner were hypothesized to be only associated with talking about HIV testing. Finally, the community-level norm of HIV testing was hypothesized to be related only to HIV testing behavior. Exclusion tests (Hausman tests for the continuous measure of program exposure and log-likelihood ratio tests for the three binary outcomes) were used to assess if such exclusions significantly change the models. All analyses were performed with Stata 12 (StataCorp, 2011). Sample weights were applied to all descriptive statistics. Multiple regressions were not weighted because we already controlled for variables that were used to calculate sample weights (urban/rural and province) and to allow for statistical tests as part of the procedure. The final regressions were then adjusted for design effects using Stata’s survey (svy) commands.
Results Sample distribution The average exposure to the four communication programs was 4.16 on a scale from 0 to 9. Just over half of the sample believed that many of their friends were tested for HIV. Less than half (47%) had talked about HIV testing with friends and sex partners; the same proportion of respondents had tested for HIV in the last 12 months (Table 1). Nearly 60% of respondents were female; 70% lived in urban areas. Three-quarters of the sample were either married or in a steady relationship. Most people had completed primary or attained matric level of schooling. Mean SES was 4.23 on a 0–7 scale and mean poverty was 1.5 on a 1–4 scale. The sample was heavily concentrated in Gauteng and Kwazulu-Natal, while provinces like Northern Cape contributed only a small proportion of the sample. There are some variations in joint communication program exposure and the three HIV-related outcomes by socioeconomic characteristics. For example,
1112
M. Do et al.
Table 1. Sample distribution and differences in communication program exposure and outcomes by socioeconomic characteristics, South Africa, 2012.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Gender Male Female Race Black Colored White Indian Type of settlement Rural Urban Marital status Married/living together Widow/Divorced Steady Single Age groups 16–24 25–35 36–55 Education level Up to primary Primary to standard nine Matric Tertiary or above SES Poverty level Employment status Unemployed Employed Student Province Gauteng Western Cape Eastern Cape Free State Kwazulu-Natal Limpopo Mpumalanga North West Northern Cape Total
Joint communication exposure mean (SD)
High perceived no. of friends got tested (%)
Talking about testing (%)
HIV testing last 12 months (%)
40.52 59.48
3.82 (.12) 4.39 (.10)
54.91 51.30
41.88 51.02
38.02 53.77
80.99 13.59 1.68 1.73
4.42 3.14 2.36 1.68
(.10) (.17) (.38) (.41)
51.01 64.46 50.50 47.12
49.89 36.27 38.61 20.19
48.16 48.04 33.66 17.31
30.73 69.27
3.83 (.16) 4.31 (.13)
46.45 55.57
45.47 48.14
47.59 47.30
40.24 2.35 25.98 31.43
3.66 3.59 4.58 4.49
(.25) (.27) (.14) (.12)
53.92 60.28 54.21 52.44
53.92 43.97 44.36 51.47
50.49 41.84 47.85 48.27
29.76 40.87 29.37
4.66 (.12) 4.36 (.11) 3.38 (.11)
51.53 54.40 51.73
46.11 50.04 44.75
51.59 48.00 42.26
9.29 41.12 36.31 13.27 4.23 (2.13) 1.51 (.65)
1.85 3.80 4.80 5.16
(.11) (.11) (.13) (.15)
38.17 49.57 56.24 63.36
33.87 45.32 50.83 53.32
37.63 48.40 48.85 47.05
52.93 36.59 8.99
4.03 (.10) 4.11 (.11) 5.16 (.15)
50.38 56.90 50.93
46.73 49.25 43.52
48.36 45.15 50.37
21.97 14.36 11.23 6.80 20.45 8.93 8.06 6.38 1.83
4.34 3.26 6.27 4.79 3.24 3.64 4.93 3.89 3.38 4.16
59.51 57.86 55.88 49.58 55.46 33.77 48.76 57.70 58.18 52.76
45.82 28.93 67.40 55.12 46.25 37.50 51.86 49.35 36.36 47.32
51.39 42.14 59.80 45.94 44.54 45.90 63.22 32.38 40.91 47.39
Distribution mean (SD) or % (weighted)
(.25) (.25) (.22) (.25) (.18) (.27) (.17) (.32) (.46) (.10)
communication exposure seemed higher among blacks and more educated respondents compared to others. Perceived prevalence of HIV testing seemed higher among respondents in urban and with more education. Respondents who were female, black, married, and more educated were more likely to talk about HIV testing than others. Finally, HIV testing seemed more common
among female and black and colored respondents, as well as those with more than primary schooling. Table 2 (column 1) shows that several individual factors, including having attended a community meeting where HIV was discussed, having heard of HIV being discussed at church in the last 12 months, and increased perceived social capital for HIV in the community were
AIDS Care
1113
Table 2. Mulitple regression analysis of the joint exposure to four communication programs, talking about HIV testing with partners and friends, perceived social norms related to testing, and getting tested for HIV, South Africa, 2012.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Joint communication exposure High perceived no. of friends got tested Talking about HIV testing Gender Male Female Race Black Colored White Indian Type of settlement Rural Urban Marital status Married/living together Widow/divorced Steady Single Age groups 16–24 25–35 36–55 Education level Up to primary Primary to standard nine Matric Tertiary or above SES Poverty level Employment status Unemployed Employed Student Norm of HIV testing in community Leader of community group related to HIV Attended community meeting that discussed HIV Heard HIV discussed in church last 12 months Perceived social capital for HIV in community Ever slept away from home last 12 months Had sex with someone just met at bar Number of partners Had casual MCP relationship Any partner had other sex partners Any partner of same sex Expected to have sex again with a partner Frequency of watching SABC1 Frequency of watching SABC2 Frequency of watching SABC3
Joint communication exposure
High perceived no. of friends got tested
Talking about testing
HIV testing last 12 months
Coef. (SD) (1)
OR (SD) (2)
OR (SD) (3)
OR (SD) (4)
1.05 (.01)***
1.08 (.01)*** 1.50 (.10)***
1.03 (.01)* 1.25 (.09)** 3.47 (.21)***
1.00 .89 (.07)
1.00 1.36 (.09)***
1.00 1.56 (.10)***
1.00 1.17 (.26) .40 (.12)** .66 (.20)
1.00 .68 (.14) .87 (.32) .49 (.19)
1.00 1.25 (.27) .83 (.27) .63 (.24)
.10 (.09)
1.00 1.11 (.09)
1.00 .97 (.08)
1.00 1.10 (.09)
–.09 (.24) .28 (.09)** .35 (.10)**
1.00 1.38 (.25) 1.00 (.09) .98 (.09)
1.00 .90 (.17) 1.06 (.09) .88 (.08)
1.00 .82 (.16) .83 (.07)* .94 (.09)
.85 (.08) .98 (.07) 1.00
1.08 (.10) 1.12 (.10) 1.00
1.46 (.14)*** 1.16 (.10) 1.00
1.00 1.54 1.93 2.27 1.01 1.04
1.00 1.30 1.43 1.41 .99 1.22
1.00 1.21 1.19 1.20 .96 1.00
(.15) (.15) (.19) (.02)* (.06)
1.69 .87 .99 1.09 1.02 .91 .71 .90 .86 1.22
(.17)*** (.10) (.10) (.08) (.03) (.06) (.18) (.08) (.08) (.10)*
.48 (.09)*** –.90 (.27)** –1.94 (.40)*** –1.37 (.42)**
.59 (.10)*** .43 (.09)***
.77 1.18 1.29 .10 –.13
(.12)*** (.13)*** (.15)*** (.03)*** (.08)
(.17)*** (.23)*** (.32)*** (.02) (.05)
(.15)* (.18)** (.23)* (.02) (.07)***
1.00 1.16 (.08)* .84 (.09) –.15 .31 .18 .10 .16 –.16 .26 –.14 –.21
(.14) (.10)** (.09)* (.04)* (.08) (.28) (.11)* (.11) (.09)*
–.15 (.14) .40 (.04)*** .18 (.04)***
1.55 1.00 1.18 1.09 1.21 .99 .96 1.22 1.10
(.21)** (.08) (.09)* (.03)** (.08)** (.22) (.09) (.13) (.09)
1.29 (.13)* 1.00 (.02)
1.81 1.44 1.26 1.12 1.11 .80 1.25 .68 .86 .67 1.49
(.26)*** (.13)*** (.11)** (.03)** (.08) (.21) (.12)* (.06)*** (.07) (.13)* (.18)**
1.12 (.03)*** .92 (.03)**
.80 (.10) 1.03 (.02)
1114
M. Do et al.
Table 2 (Continued)
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
Characteristics Frequency of watching e TV Frequency of watching Top TV Frequency of listening to the radio Exclusion testa Chi-squared (df) Probability > χ2 Exogeneity tests between communication and outcomes Exogeneity tests between outcomes (2), (3) and (4) Rho Probability > χ2 Goodness-of-fit R2 Hosmer-Lemeshow χ2 (df) Probability > χ2
Joint communication exposure
High perceived no. of friends got tested
Talking about testing
HIV testing last 12 months
Coef. (SD) (1)
OR (SD) (2)
OR (SD) (3)
OR (SD) (4)
5.33 (5) .38 .93 (.05)
6.94 .54 1.00 (.05)
.28 (.04)*** .17 (.06)** .96 (.02)* 7.41 (53) 1.00
9.12 (8) .33 1.04 (.05) .02 .06
.07 .06
–.01 .06
.43 5.57 (8) .70
14.89 (8) .06
20.25 (8) .01
Note: All models controlled for language, frequency of watching DSTV, reading newspapers, magazines, using the internet, and ever use of Facebook and Twitter. a Hausman test was used for outcome (1) and log-likelihood tests were used for outcomes (2) through (4). *p < .05; **p < .01; ***p < .001.
related to increased exposure to programs, as was increased number of sex partners (p < .05). Very few factors were negatively associated with joint program exposure, including race and multiple partnerships of a respondent’s partner.
Effects of communication programs Table 2 also shows the effects of joint program exposure on two intermediate and the behavioral outcomes. Column 2 indicates a significant effect of program exposure on increased perception of friends getting tested (OR = 1.05; p < .001). As shown in Column 3, one’s perception of HIV testing prevalence, in turn, had a positive effect on the likelihood of talking about testing with friends and partners (OR = 1.50; p < .001). In addition, joint communication exposure had a direct, positive effect on one’s likelihood of talking about HIV testing (OR = 1.08; p < .001). Finally, Column 3 shows a direct, positive effect of joint program exposure on having tested in the last 12 months (OR = 1.03; p < .05); high perceived testing norm and talking about HIV testing both also had positive effects on HIV testing. Those who perceived a high level of testing among their friends were 25% more likely than others to have tested (p < .01), but talking about HIV testing was the strongest predictor of testing: those who had discussed HIV tests with friends and
partners were more than three times as likely as others to have tested in the last 12 months (p < .001). Figure 1 summarizes the direct and indirect effects of communication programs on HIV testing; 55% of the effect of communication on HIV testing was indirect through these mechanisms (results not shown). Goodness-of-fit tests (shown in Table 2) indicate fair to high levels of goodness for most equations. Variables excluded from each equation were already discussed above. Exclusion tests showed that their exclusions did not affect the results. Exogeneity tests (see Table 2) suggested that joint communication exposure was exogenous to all the three outcomes; there was also no evidence of endogeneity between perceived HIV testing norm, HIV testing discussion, and testing behavior. Finally, we estimated the dose-response effects of joint program exposure. For all outcomes, the adjusted probability of the outcomes increased with increasing levels of communication program exposure. For HIV testing behavior, the difference between no exposure and the highest level of exposure was 4.8 percentage points (results not shown). Discussion This paper examines the impact of four communication programs on HIV testing behavior among South African men and women aged 16–55. Less than half of the
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care
1115
Figure 1. Path model of the impact of four communication programs on HIV testing.
survey respondents had tested for HIV within 12 months before the survey. Findings indicate significant, positive effects of the programs on HIV testing, partly due to indirect effects through increasing an individual’s perception of HIV testing prevalence and the probability of talking about testing with partners and friends. Dose responses in the three outcomes were found with increasing program exposure. The findings are consistent with the earlier studies that found positive effects of communication programs on HIV-related norms, discussion with others, and testing in South Africa and elsewhere in sub-Saharan Africa (Goldstein, Usdin, Scheepers, & Japhet, 2005; Hutchinson, Mahlalela, & Yukich, 2007; Keating, Meekers, & Adewuyi, 2006; Pettifor et al., 2007). Although cross-sectional data were used, statistical tests provide evidence that communication exposure and the three outcomes were exogenous, hence, an SEM analysis could be used. It should be noted that although the Chi-square goodness-of-fit test is usually sensitive to large sample sizes (Hooper, Coughlan, & Mullen, 2008), it shows a statistical lack of fit for only one of the models, giving us more confidence that the models fit the data well. The results underline the importance of promoting discussion around HIV tests with friends and partners. Talking about it may contribute to reducing stigma and encourage greater acceptance of one’s HIV status. Communication programs should also target those less educated, who may be unlikely to be exposed to
such programs and to talk about HIV testing. In addition, community-based interventions such as community and religious meetings that discuss HIV may also be effective in promoting HIV testing discussion. Perceived norm of HIV testing among friends mediates the effects of communication programs on HIV testing not only by its direct effects on testing behavior but also by its indirect effects through increasing discussions about HIV testing. It suggests that interventions to promote HIV testing should address social barriers, in addition to individual-level barriers to HIV testing. The finding that community group leadership and hearing HIV discussions in church both were related to a higher perceived HIV testing prevalence underscores the importance of integrated interventions that rely on community activities and the mass media. Finally, the analysis revealed a lower level of exposure to the four communication programs among those who are male, married, or living with a partner, older, less educated, and have a lower socioeconomic status. Future programs should be tailored to address their particular motivations and barriers. On the other hand, the finding that those with more sex partners were more exposed to communication programs and more likely to talk about HIV testing is promising. It is possible that individuals who had multiple partners were somewhat aware of HIV risks and paying more attention to the TV series. The result suggests that these programs might have reached and had an impact on
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
1116
M. Do et al.
those who may be at increased risk of HIV infection. In general, male and older individuals should also be targeted because they were less likely to get tested, due in part to their lower likelihood of using health care services in general. This study is limited to individuals who had sex 12 months before the survey. Although our analysis shows no differences between those included in the study and those who were not in terms of SES, poverty, race, or urban residence, several differences exist between the two groups. For example, those included in the sample were more likely than those who were not to be male, in a relationship, more educated, and employed; they were also likely to be older than those excluded (results not shown). Therefore, results of this study are only applicable to South African men and women who were sexually active within the last year and fit the characteristics of the sample. Another limitation is potential recall bias related to the program exposure measure. However, we believe it is not a major problem as our measure of communication exposure was constructed based on responses to several questions, using multiple visual aids to minimize social desirability and acquiescence response set. Talking about HIV testing was measured by a dichotomous indicator; the frequency and content of such conversations are unknown. It is not clear if individuals talked about HIV as a general or a personal issue, or if they talked about it in the context of the communication programs. For example, conversations about HIV among friends may center on rumors about who might be infected rather than care and prevention related topics (Smith, Lucas, & Latkin, 1999). Some of the friends and partners may also be opinion leaders, who may be more influential than others in encouraging one to get tested (Sivaram et al., 2005). It is not possible to know in this assessment which aspects of the conversations may be more effective than others in promoting HIV testing. Conclusions Notwithstanding the limitations, this paper suggests that communication programs can be powerful in promoting HIV testing. Investment in behavior change communication programs may have long-term effects through changes in norms and perceptions related to the desired behavior. References Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. doi:10.1016/0749-5978(91)90020-T Babalola, S. (2007). Readiness for HIV testing among young people in northern Nigeria: The roles of social norm and perceived stigma. AIDS and Behavior, 11, 759–769. doi:10.1007/s10461-006-9189-0
Babalola, S., & Kincaid, D. L. (2009). New methods for estimating the impact of health communication programs. Communication Methods and Measures, 3(1–2), 61–83. doi:10.1080/19312450902809706 Baiden, F., Remes, P., Baiden, R., Williams, J., Hodgson, A., Boelaert, M., & Buve, A. (2005). Voluntary counseling and HIV testing for pregnant women in the KassenaNankana district of northern Ghana: Is couple counseling the way forward? AIDS Care, 17, 648–657. doi:10.1080/ 09540120412331319688 Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: PrenticeHall. Bertrand, J. T., O’Reilly, K., Denison, J., Anhang, R., & Sweat, M. (2006). Systematic review of the effectiveness of mass communication programs to change HIV/AIDS-related behaviors in developing countries. Health Education Research, 21, 567–597. doi:10.1093/her/cyl036 CADRE. (2011). 4Play: DVD discussion guide. Talking about sex, love and relationships. Johannesburg: JHHESA. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117–140. doi:10.1177/001872675 400700202 Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. Gage, A. J., & Ali, D. (2005). Factors associated with selfreported HIV testing among men in Uganda. AIDS Care, 17, 153–165. doi:10.1080/09540120512331325635 Goldstein, S., Usdin, S., Scheepers, E., & Japhet, G. (2005). Communicating HIV and AIDS, what works? A report on the impact evaluation of Soul City’s fourth series. Journal of Health Communication, 10, 465–483. doi:10.1080/ 10810730591009853 Hendriksen, E. S., Hlubinka, D., Chariyalertsak, S., Chingono, A., Gray, G., Mbwambo, J., … Coates, T. J. (2009). Keep talking about it: HIV/AIDS-related communication and prior HIV testing in Tanzania, Zimbabwe, South Africa, and Thailand. AIDS and Behavior, 13, 1213–1221. doi:10.1007/s10461-009-9608-0 Hooper, D., Coughlan, J., & Mullen, M. (2008). Structural equation modelling: Guidelines for determining model fit. Articles, p. 2. Hutchinson, P., Mahlalela, X., & Yukich, J. (2007). Mass media, stigma, and disclosure of HIV test results: Multilevel analysis in the Eastern Cape, South Africa. AIDS Education & Prevention, 19, 489–510. doi:10.1521/ aeap.2007.19.6.489 Katz, E., & Lazarsfeld, P. F. (2006). Personal influence: The part played by people in the flow of mass communications. New Brunswick, NJ: Transaction. Keating, J., Meekers, D., & Adewuyi, A. (2006). Assessing effects of a media campaign on HIV/AIDS awareness and prevention in Nigeria: Results from the VISION Project. BMC Public Health, 6(1), 123. doi:10.1186/1471-2458-6-123 Kincaid, D. L. (2000a). Mass media, ideation, and contraceptive behavior. Communication Research, 27, 723–763. doi:10.1177/009365000027006003 Kincaid, D. L. (2000b). Social networks, ideation, and contraceptive behavior in Bangladesh: A longitudinal analysis. Social Science and Medicine, 50, 215–231. doi:10.1016/ S0277-9536(99)00276-2 Kincaid, D. L. (2009). Convergence theory. In S. W. Littlejohn & K. A. Foss (Eds.), Encyclopedia of communication theory (pp. 188–191). Thousand Oaks, CA: Sage.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 02:42 26 January 2015
AIDS Care Kincaid, D. L., Delate, R., Storey, D., & Figueroa, M. E. (2012). Closing the gaps in practice and in theory. Public Communication Campaigns, 305, 305–319. Kincaid, D. L., & Do, M. P. (2006). Multivariate causal attribution and cost-effectiveness of a national mass media campaign in the Philippines. Journal of Health Communication, 11(S2), 69–90. doi:10.1080/1081073 0600974522 Kranzer, K., McGrath, N., Saul, J., Crampin, A. C., Jahn, A., Malema, S., … Glynn, J. R. (2008). Individual, household and community factors associated with HIV test refusal in rural Malawi. Tropical Medicine & International Health, 13, 1341–1350. doi:10.1111/j.1365-3156.2008.02148.x MacPhail, C., Pettifor, A., Moyo, W., & Rees, H. (2009). Factors associated with HIV testing among sexually active South African youth aged 15–24 years. AIDS Care, 21, 456–467. doi:10.1080/09540120802282586 McKee, N., Bertrand, J. T., Becker-Benton, A., & Becker, B. L. (2004). Strategic communication in the HIV/AIDS epidemic. New Delhi, Thousand Oaks, CA and London: Sage. Noar, S. M., Palmgreen, P., Chabot, M., Dobransky, N., & Zimmerman, R. S. (2009). A 10-year systematic review of HIV/AIDS mass communication campaigns: Have we made progress? Journal of Health Communication, 14 (1), 15–42. doi:10.1080/10810730802592239 Noelle-Neumann, E. (1991). The theory of public opinion: The concept of the spiral of silence. In J. A. Anderson (Ed.), Communication yearbook (Vol. 14, pp. 188–191). Thousand Oaks, CA: Sage. Noelle-Neumann, E. (1993). The spiral of silence: Public opinion – Our social skin. Chicago: University of Chicago Press. Noelle-Neumann, E. (1974). The spiral of silence a theory of public opinion. Journal of Communication, 24(2), 43–51. doi:10.1111/j.1460-2466.1974.tb00367.x
1117
Nuwaha, F., Kabatesi, D., Muganwa, M., & Whalen, C. (2006). Factors influencing acceptability of voluntary counseling and testing for HIV in Bushenyi district of Uganda. East African Medical Journal, 79, 626–632. Pettifor, A. E., MacPhail, C., Bertozzi, S., & Rees, H. V. (2007). Challenge of evaluating a national HIV prevention programme: the case of lovelife, South Africa. Sexually Transmitted Infections, 83, i70–i74. doi:10.1136/sti.2006. 023689 Rimal, R. N., Limaye, R. J., Roberts, P., Brown, J., & Mkandawire, G. (2013). The role of interpersonal communication in reducing structural disparities and psychosocial deficiencies: Experience from the Malawi BRIDGE project. Journal of Communication, 63(1), 51–71. Rogers, E. (2003). Diffusion of innovations. New York, NY: Free Press. Sarker, M., Sanou, A., Snow, R., Ganame, J., & Gondos, A. (2007). Determinants of HIV counselling and testing participation in a prevention of mother to child transmission programme in rural Burkina Faso. Tropical Medicine & International Health, 12, 1475–1483. doi:10.1111/ j.1365-3156.2007.01956.x Sivaram, S., Johnson, S., Bentley, M. E., Go, V. F., Latkin, C., Srikrishnan, A., … Solomon, S. (2005). Sexual health promotion in Chennai, India: Key role of communication among social networks. Health Promotion International, 20, 327–333. doi:10.1093/heapro/dai012 Smith, L. C., Lucas, K. J., & Latkin, C. (1999). Rumor and gossip: Social discourse on HIV and AIDS. Anthropology & Medicine, 6(1), 121–131. doi:10.1080/13648470.1999.9964577 StataCorp. (2011). Stata statistical software: Release 12. College Station, TX: StataCorp LP. Suls, J., Martin, R., & Wheeler, L. (2002). Social comparison: Why, with whom, and with what effect? Current Directions in Psychological Science, 11, 159–163. doi:10.1111/ 1467-8721.00191
