This report describes the findings from the external evaluation of the Swaziland component of the
Southern African Regional Social and Behavior Change Communication Program (BCCP). The program,
implemented in eight countries in Southern Africa with funding from the British Department for
International Development (DfID), aims to reduce HIV infection by increasing health awareness and by
facilitating social and behavioral change through the use of both mass media and community-based
activities. In Swaziland, the program is implemented by Lusweti /Soul City and the Southern African HIV
and AIDS Dissemination Service (SAfAIDS).
EVALUATION OBJECTIVE
The main objective of the evaluation is to assess the net effect of exposure to specific components of
the program on key indicators of HIV knowledge, attitudes, and behaviors, after controlling for other
factors or programs that might also concurrently influence or determine those outcomes. The results of
the study will also be used for a separate analysis of the cost-effectiveness of the program.
DATA
The evaluation is based on a nationally representative survey of male and females aged 15-49. The
survey was implemented by the Social Impact and Policy Analysis Corporation (SIAPAC) with technical
support from Tulane University. The survey sample was drawn with the assistance of the Swaziland
Central Statistical Office (CSO), using a three-stage sampling design that involved stratification of the
population into urban, rural, and border areas. Within each of those domains, areas of concentrated
programmatic activities were identified and over-sampled to increase the statistical power for
measuring the effects of these localized interventions.
The data collection instrument was developed from the questionnaire used for a similar evaluation in
Malawi and adapted to the Swaziland context by Tulane, SIAPAC, Lusweti and SAfAIDS. The instrument
covers the eight health areas targeted by the program (multiple/concurrent sexual partnerships, other
HIV risk factors, HIV communication, condom use, HIV testing, HIV treatment, HIV stigma, and genderbased
violence). Approval for the study was granted by the Scientific and Ethics Committee of the
Ministry of Health of Swaziland and by the Institutional Review Board of the Tulane Human Research
Protection Program. Following extensive training in survey procedures and objectives, questionnaire
content, and ethical conduct of research, fieldwork was conducted in 125 enumeration areas (EAs) by
eight field teams, comprised of a supervisor and four enumerators. In total, 3,972 interviews were
successfully completed.
METHODS
This evaluation uses a post-only cross-sectional design, given the national scope of the program.
Multivariate statistical methods are used to control for differences between individuals who are
exposed to the intervention and those individuals who are not exposed. Two different estimation
methods are used to determine the existence of program effects: (1) multivariate regression analysis, 10
and 2) propensity score matching (PSM). All analyses are weighted to account for the multi-stage
sampling design.