Annex B. Methodology of the SIGI Tanzania
The SIGI Tanzania country study is conducted by the OECD Development Centre, in partnership with UN Women Tanzania, the National Bureau of Statistics (NBS) and the Office of the Chief Government Statistician (OCGS) of Zanzibar, and in close collaboration with the Ministry of Community Development, Gender, Women and Special Groups of the United Republic of Tanzania and the Ministry of Health, Social Welfare, Elderly, Gender and Children of the Revolutionary Government of Zanzibar. The SIGI Tanzania aims to improve the rights and well-being of women and girls in Tanzania and to advance gender equality through the elimination of discrimination in social institutions.
The purpose of the SIGI Tanzania is to build robust evidence on gender equality focusing on social norms and practices and to support national partners in strengthening national statistical and analytical capacities. Similarly, SIGI Tanzania aims at integrating the evidence generated into future policies targeting gender equality and women’s empowerment. To achieve this objective, SIGI Tanzania applies the global SIGI conceptual framework developed by the OECD Development Centre to the national level and sub-national level with the aim to strengthen national policy making by focusing on the root causes of gender inequalities.
To build a robust, evidence-based analysis of discriminatory social norms, attitudes and practices and to produce the required data, two approaches were employed:
A quantitative component based on a household survey statistically representative at the national and regional levels.
A qualitative component based on a qualitative assessment that draw on focus group discussions and key informant interviews.
The SIGI household survey was conducted between 10 December 2020 and 11 January 2021 by NBS and OCGS staff. The survey consisted of two separate questionnaires, a household questionnaire and an individual questionnaire. In each selected household, the household questionnaire was first applied and answered by any member of the household aged more than 18 years, with a preference for the head of household or its spouse/partner. Immediately after completing the household questionnaire, two eligible individuals (one woman and one man) for the individual questionnaires were identified and participated in the individual interviews. When a selected eligible individual was not present at the time of the interview, the assigned enumerator had to do three call back visits per absent respondent on the same day of the interview.
Interviews were conducted in Kiswahili and data collection was performed using Computer Assisted Personal Interview (CAPI) whereby CSPro program was installed in the tablets. After completion of an interview, interviewers were able to upload the collected information to the server via the Internet. Fieldwork supervisors verified the information collected before sending it to the server. On average, household questionnaires were completed in 20 minutes and individual questionnaires were completed in 52 minutes.
Sampling and fieldwork strategy
The sample for the SIGI Tanzania household survey was a stratified sample selected in three stages, which adopted some of the enumeration areas (EAs) of the 2017/18 Household Budget Survey (2017/18 HBS) and used the list of enumeration areas from the 2012 Population and Housing Census (2012 PHC) frame. In total, 61 sampling strata were formed. Stratification was achieved by separating each region into urban and rural areas, each area in each of the regions forming a sampling stratum. In Dar es Salaam, only one stratum was formed as it is a purely urban region. Samples of EAs were selected independently in each sampling stratum, by a probability proportion to size (PPS). Implicit stratification and proportional allocation were achieved at each of the lower administrative unit levels by sorting the sampling frame within the explicit stratum according to administrative unit in different levels before sample selection at the stage.
In the first stage, 217 EAs were selected with probability proportional to the EA size in terms of households and with independent selection in each sampling stratum. Among the selected 217 EAs, 71 EAs were located in urban areas and 146 EAs were in rural areas. The numbers of urban and rural EAs selected reflected the distribution of urban and rural EAs in the sampling frame of EAs. With a fixed number of 20 households selected per cluster, the total number of households selected for the SIGI Tanzania household survey was expected to be 4 340, divided between 1 420 households in urban areas and 2 920 households in rural areas.
In the second stage, a fixed number of 20 households (including 12 households which were interviewed in the 2017/18 HBS and 8 additional households) were randomly selected from each selected EA. Selected households were visited and interviewed; household heads or any most knowledgeable household member of age 18 years or above were eligible for the household interviews. The interviewers were asked to interview only the pre-selected households; no replacement was allowed for non-respondent households to prevent bias. The interviewers were asked to make at most three call backs for nonresponse in order to reduce nonresponse bias.
In the third stage, two individuals (one woman and one man) aged 15 years or older were selected for individual interviews among the eligible members of the household. Eligible members were all the members of the household aged 15 years and older who were available on the day of the survey. Selection of the man and the woman to be interviewed was done automatically by the tablet, following a Kish grid technique.
Results of the survey
A total of 4 340 households were selected for the survey out of which 4 339 households were found and successfully interviewed, achieving a household response rate of nearly 100%. For individuals, a total of 8 680 individuals were expected to be interviewed, of which 7 124 were found within the selected households. Among these 7 124 individuals, 53 individuals did not provide the consent for individual interview, 3 individuals provided the consent but stopped the interview and 7 068 were successfully interviewed, translating into a response rate of 99.2% (Table A B.1).
Data cleaning, validation and analysis
In the wake of the completion of data collection, data cleaning and weighting processes were performed for five months between March and July 2021.
Analysis of the dataset by OECD and drafting of the SIGI Country Report for Tanzania started beginning of September 2021 and concluded end of November 2021. In parallel, the SIGI Tanzania Survey Report was produced by NBS and OCGS with the support of UN Women Tanzania.
The SIGI Tanzania qualitative assessment was conducted by an independent consultant, Selemani Mbuyita, in February and March 2021 using qualitative methods of data collection. The information collected complements and assists the interpretation of data and findings collected from the SIGI household survey. The methods included focus group discussions (FGD) and key-informant interviews (KII). FGDs included community members of various demographic groups, including:
KIIs were conducted with political/administrative leaders, religious leaders and influential persons, all drawn from the selected sites. KIIs were used to collect information on gender responsiveness of systems and social institutions including women and adolescent girls’ participation in community decision-making processes, women/girls status in relation to decision making at household level as a result of values held by social institutions, ownership of resources and women empowerment activities. They were also asked about men’s involvement and roles as well as the availability of places where adolescents would be able to receive social and psychosocial services.
Three regions – Dodoma, Shinyanga and Zanzibar (Ungunja) – were selected. Three studied sites (one per region) were randomly selected among the EAs covered by the SIGI household survey. All FGDs and KIIs participants were drawn from the clusters and villages selected:
Overall, 92% of all planned interviews in Dodoma, Shinyanga and Zanzibar were conducted (Table A B.2).