Recruitment procedure
We used maps from the transgender organization in Jakarta (Yayasan Srikandi Sejati) to select the sites of the study. The waria in this study were recruited using a cluster sampling procedure with the five municipalities in Jakarta as groups (Central, North, South, East and West), representing all residential locations of waria. In total, we recruited 210 participants from the Center of Jakarta (n = 23), East Jakarta (n = 75), West Jakarta (n = 45), South Jakarta (n = 45), and North Jakarta (n = 22) between September and October 2007. To gather information, interviews were conducted during the day, starting about 12 pm (in the afternoon) to 6–7 pm (in the evening). This seemed the best time for conducting interviews in waria, as they were still sleeping in the morning and they were starting to dress up and go out to the park, railway station and other places to meet the clients. The inclusion criteria were the following: Being identified as a transgender by mamis, living in one of the five selected districts in Jakarta, be able to speak and understand Bahasa Indonesia, and participate voluntarily in the study. One participant from East Jakarta was excluded from this study due to a serious language barrier. Therefore 209 waria participated, comprising almost 16 % of the last total estimated number of waria in Jakarta [19].
Waria usually migrate from rural areas to big cities like Jakarta. Upon arrival they will join and link with the waria community, that is coordinated by a senior ‘mami’ who provides guidance and protection. A mami usually provides support to the 5–15 waria in her group, including violence protection and promotion of condom use, uptake of HIV testing and regular STI check-ups [3]. ‘Mami’, as a senior person also will ensure that the newcomer will be socially accepted in the network. This support and protection are needed because waria are a stigmatized and discriminated group, who are easily recognized by their looks. Waria communities may provide waria with a “safe haven” of support and recognition.
Since mamis’ role may be rather instrumental, some of them are recruited and placed as field coordinators in the waria community organization. Therefore, a field coordinator is usually a key person in a district. By recruiting and cooperating with the mamis, the waria organization is able to provide the list of waria’s living areas and all mamis in the five districts were asked to contact participants for the assessments. In the selected areas, mami listed the waria who were present and available in sampled living areas at the time of data collection. All these waria were approached and included in the study.
The cooperation between mamis in five districts with the waria organization brings mutual benefits: The waria organization usually has easier access to the health services and social welfare programs which are managed by the district health and social welfare office, while mamis have ability to enter waria’s social networks and mobilize them.
Waria who live in the areas that were selected for this study, by design, have been exposed to HIV/AIDS infection prevention intervention and have access to HIV-related health services. The services were designed to cover the whole districts and all waria communities in Jakarta. Accordingly, we assume that all waria in Jakarta have been exposed to the intervention in some way. Therefore, to the extent that the present sample of waria is representative of the whole waria population in Jakarta, the results can be generalized to the Jakarta population. However, because the Jakarta population has been exposed by designed preventive interventions, care must be taken to generalize the present results to waria populations in different health care environments.
Interview procedure
The study presented in this article was part of a broader study that covered condom use and HIV-related health-seeking behaviors. The ethical approval has been provided by the Ethical Committee Psychology University of Groningen in the Netherlands, and the IRB approval from the local institution was provided by the Ethical Committee of Psychology University of Indonesia. This article only provides data on health-seeking behaviors, including adherence to STI treatment. The study of condom use-related behaviors is presented in another paper [20].
The interviews were structured and conducted face-to-face. The interview first, assessed demographic variables, sexual history and sexual practices. Secondly, all the questions related to health-seeking behavior were asked: visiting STI services regularly, adherence to STI treatment, taking for HIV test and picking up the test result. The interviews took 45–60 min.
Due to waria’s limited understanding of the meaning of questions related to sexual and HIV-related health-seeking behaviors, this study selected the interview as a method to gather the data. The interviews had been conducted by using Bahasa Indonesia combined with local and slang language. Another consideration is that the illiteracy level among waria still is very high. For that reasons, the four interviewers were trained in advance by the first author to cope with limited literacy. Five interviewers, including the first author, were involved in the data collection.
Informed consent was prepared on two levels. First, permission was granted by the “mamis” as the coordinators and leaders of waria in each district. The permission was given after the first author conducted a meeting with all “mamis” from the five districts, to explain the purpose of the study and the requirement to have an interview only on voluntary basis. Secondly, the interviewers introduced themselves to the respondents and provided them with information about the purpose of the study. The interviewers explained the respondents that the participation was on a voluntary basis, meaning that they could withdraw at any time without having to state a reason. Subsequently, the respondents were asked whether they had understood the information and were willing to participate. The actual interview started after the individual had given a verbal consent to participate. When the respondents were under 18 years old, a verbal consent was given by the “mami” as the responsible person in each group. Unwritten informed consent was chosen for this study to assure the anonymity and confidentiality of the respondents.
The interviews were conducted at different venues. Waria usually live in groups, in the middle of a kampong (Spelled kampung in Malay and Indonesian). In Malaysia, Brunei, Indonesia and Singapore, the term kampong (village) applies to traditional villages, especially of indigenous people and the term has also been used to refer to urban slum areas and enclosed developments within towns and cities (https://en.wikipedia.org/wiki/Kampong), which is often quite crowded, and mixed with the general population. Through networks with mamis, we were able to find quiet places or separate rooms to conduct the interviews. The majority of the interviews were conducted in the waria’s bed rooms, a smaller number in the salons, and the remaining in the Primary Health Center (PHC) or the mami’s living rooms.
Questionnaire
Background characteristics
The socio-demographics gathered were age, educational level, cultural background, persons currently living with the respondent, type of job and duration of stay in Jakarta. Sexual behavior was assessed with questions on commercial sex practices, current sex partners (only men/mostly men/men and women equally/mostly women/only women), the type of sex they most often engaged in (anal sex as the receptive partner/insertive partner/both as receptive and insertive partner/receiving oral sex/giving oral sex/insertive vaginal sex and others), condom use as the receptive partner in the past week, and condom use as the insertive partner in the past month.
The central part of the questionnaire was based on the application of the TPB on the four behaviors. The four examined behaviors are, visiting Sexually Transmitted Infection (STI) services regularly, adherence to the STI treatment, taking an HIV test and picking up the result of HIV test.
Attitude
The attitudes were assessed by using a 7-point semantic differential scale: unhealthy (1) – healthy (7); harmful (1) – beneficial (7); boring (1) – exciting (7); inconvenient (1) - fun/enjoyable (7); something that can be overlooked (1) - something I have to do (7); foolish (1) – wise (7). The items for each behavior category include the following: An example item for visiting STI services regularly: “If I go to STI services once every 2 or 3 months, this is…”; An example item for adherence to STI treatment: “If I receive drugs or get the prescription from the physician, finishing all drugs accordingly is…”; An example item for taking an HIV test: “If I take an HIV test at least once a year, this is…”; An example item for picking up the result of HIV test:,“If I pick up the results, this is…”. Concerning each behavior separately, the average item score was computed to be used as the scale score: The higher the score, the more positive the attitude towards the specific behavior. The Cronbach’s alphas of the four scales ranged from .70 to .88.
Subjective norms
The subjective norms were measured by using a 7-point semantic differential scale ranging from should not (1) to should (7). Items were related to normative beliefs regarding regular partner, friends, mamis and outreach workers. The items for each behavior category were the following: “According to the following people, I should or should not visit STI services once every 2 or 3 months”, “According to the following people, I should or should not finish all drugs received from or prescribed by the physician”, “According to the following people, I should or should not take an HIV test at least once a year”, “According to the following people, I should or should not pick up the results of HIV test (after taking an HIV test). Concerning each behavior separately, the average item score was computed to be used as the scale score: The higher the score, the more positive the subjective norms towards the specific behavior. The Cronbach’s alphas of the four scales ranged from .86 to .90.
Perceived behavioral control
PBC was measured by using a 7-point semantic differential scale. The first scale ranged from very difficult (1), difficult (2), somewhat difficult (3), not easy (4), somewhat easy (5), easy (6), to very easy (7). The second scale ranged from not sure at all to be able (1) to very sure to be able (7), and the last one was completely uncertain to be able to ensure (1) to will certainly be able to ensure (7). Each behavior was assessed with three items. Example items consisted of two questions and one statement: “In the next 6 months, how easy or difficult will it be for you to visit STI services?”, “In the next 6 months, if I want to, I am sure that I will be able to finish all drugs received from or prescribed by the physician”, “In the next 6 months, how sure are you that you will be able to take an HIV test?”. Concerning each behavior separately, the average item score was computed to be used as the scale score: The higher the score, the easier engaging in the specific behavior was perceived to be. The Cronbach’s alphas of the four scales ranged from .79 to .87.
Actual behavior
Each of the four behaviors was assessed with one item. They were: (1) “In the last 6 months, how many times have you visited STI services?” The responses indicated the number of times visiting STI services in the last 6 months; (2) “When you received medicine or got a prescription from the doctor, did you take or buy and finish it all?” The possible responses were never (1), seldom (2), sometimes (3), regularly (4), often (5), very often (6) and always (7); (3) “Did you take an HIV test?” (Yes/No); and (4) “Did you pick up the result of HIV test?” (Yes/No).
Data analyses
Because two of the four behaviors were assessed with a binary variable, we decided to use logistic regression analysis for all four behaviors. Therefore, visiting STI services and adherence to STI treatment was also dichotomized (0 versus more than 0, and not always versus always, respectively). The covariates chosen were age, education and main job. The first analysis examined the relationships of these covariates with each of the DVs. Only when a covariate had a significant relationship with the DVs it was included in the main test: The three IVs – attitude, subjective norms and PBC – predicting each of the DVs. When one or more covariates were included, we present the statistics of the improvement of the statistical model by the set of three main predictors. When no covariate was included, we only present the statistics of the three main predictors.
To test for mediation effects, Preacher and Hayes’ SPSS Macro PROCESS (version 17) was used. Because this is a fairly novel method, it will be presented here in some detail. PROCESS has been described by Preacher and Hayes (2008) as a method for testing multiple mediators. This procedure yields unstandardized path coefficients for a multiple mediator model and estimates 95 % confidence intervals (CI) of the indirect (= mediated) effect using a bootstrapping sample procedure. Assessing an indirect effect through a bootstrapping sample procedure is more reliable than testing significance of the mediation effect, because the sampling distribution of the indirect effects is normal only for large samples. The mediation analysis used here followed the product of coefficients approach, and thus focused on the indirect effects rather than the individual paths. Using PROCESS for conducting the mediation test does not require normality in the selected variables [21].
The Mediate test was used here to examine whether 1) the relationships of subjective norms (IV) with behavior (DVs) were mediated by attitude and PBC, and whether; 2) the relationships of attitude (IV) with behavior (DVs) were mediated by subjective norms and PBC. The results below indicate mediation when the Confidence Interval of the indirect path by a 95 % bias corrected bootstrap (based on 5000 bootstrap samples) does not contain zero.
Since the model has more than one mediator, it is called a single-step multiple mediator model (see Figs. 1 and 2), in which the total effect of the model (the relationships of attitude or subjective norms with behaviors are symbolized by c) is equal to the direct effect of attitude or subjective norms on behavior which are symbolized by c’; plus the sum of the indirect effect of attitude on behaviors through subjective norms (a
1
b
1
) or PBC (a
2
b
2
), or plus the sum of the indirect effect of subjective norms on behaviors through attitude (a
1
b
1
) or PBC (a
2
b
2
). That is, c = c’ + a
1
b
1
+ a
2
b
2.