The research exercise to map Tanzania's small arms problem was done in collaboration between the Tanzanian Police Force, the Institute for Security Studies, Saferworld and the Security Research and Information Centre. A central component of this project was a socio-economic survey of almost 2,900 Tanzanian residents conducted in 12 of the country's 21 provinces. By conducting the survey two ends were met. Firstly, the police officers involved in the exercise were trained in surveying skills and gained an important insight into and understanding of the firearms problem in the country. Secondly, a comprehensive data set outlining the pertinent aspects of firearm proliferation in Tanzania was compiled.

The following section of the report will provide a brief review of the objectives of the study, the methodology and the demographic profile of the respondents.

Purpose of survey

The aim of the survey was to determine the nature and extent of firearm penetration and of the firearm-related concerns in Tanzania. Through understanding the attitudes of respondents towards firearms and their socio-economic conditions, information is provided that can inform the design and implementation of community-based projects.

The purpose of the survey is to provide information to assist communities in understanding the nature and dynamics of firearm proliferation in order that they may conceptualise solutions to problems associated with the presence of firearms.

Survey parameters

The survey was conducted between 1 and 18 August 2001. There were 41 fieldworkers trained and utilised in the completion of surveys. In total, 2,892 questionnaires were completed across 12 regions of Tanzania. On average, each questionnaire took 50 minutes to complete and each fieldworker completed 70 questionnaires. The completed questionnaires were then sent to the police headquarters in Dar es Salaam, where they were checked, coded and captured in a database.

Each fieldworker was given instructions to sample 10 respondents within eight different predetermined sample categories as defined by age and gender parameters (as outlined in the table below). In some cases, adaptations were made in terms of the total respondents per fieldworker. This was due to unforeseen circumstances such as time limitations resulting from extensive distances traveled to training workshops, family problems, and other factors.

Figure 1: Sample grid per fieldworker

The sampling technique utilised was stratified random sampling. Each fieldworker randomly approached respondents within specified age and gender categories, within a given geographic locality. Fieldworkers ensured that some of their questionnaires were completed in private settings (such as within the homestead) and in public settings (such as at places of transport, recreation, etc.) to reduce any bias that could emerge with regards to the firearm ownership questions. Overall, 39% of the surveys were undertaken in public settings and 62% in private settings.

Sample area

The 12 regions surveyed were chosen to ensure that the most affected regions in Tanzania were included. Given the objectives of the survey—to identify the most prevalent demand factors for firearms in Tanzania; to gain greater understanding of perceptions of small arms penetration in Tanzanian society; and to determine the possible agents for change that could be used in a national implementation plan—it was necessary that areas that had been, were currently, or potentially could be, impacted upon by small arms proliferation were surveyed. This would provide an opportunity for sound deductions to be made about the nature of any implementation plan to address the proliferation of small arms. In addition to surveying the most affected regions, it was also important to obtain a geographically representative spread across the country in order to minimise any sample bias. For those regions in which no surveys were conducted, it is partially possible to deduce trends from neighbouring regions.

Table 1: Place of interview

	Number of surveys conducted	Percentage of total sample
Arusha Dar es Salaam Kagera Kigoma Kilimanjaro Mbeja Morogoro Mwanza Pwani Ruvuma Tanga Zanzibar Total	80 459 240 240 238 240 232 377 160 223 240 163 2.892	3 16 8 8 8 8 8 13 6 8 8 6 100

The survey was conducted by junior police officers within the Tanzanian Police Force. The police officers were deployed within the various sample areas and trained in two central training venues, in Moshi (in Kilimanjaro Region) and Dar es Salaam (in Dar es Salaam Region).

Table 2: Number of questionnaires completed per fieldworker, by region

Sample region	Fieldworker identification code	Total number of surveys per fieldworker (n=2,892)
Arusha	27	80
Dar es Salaam	1 13 19 20 22 23 41	60 80 80 27 80 80 52
Kagera	25 26 32	81 81 81
Kigomo	24 31 33	80 81 81
Kilimanjaro	34 35 36	79 82 72
Mbeja	7 11 15	80 80 80
Morogoro	8 9 14	80 80 72
Mwanza	29 30 37 38 39 40	67 67 53 67 56 67
Pwani	3 6	80 80
Ruvuma	16 17 18	73 80 70
Tanga	2 4 12	80 80 80
Zanzibar	5 10	80 83

Each fieldworker was given a quota of surveys to complete. The table above outlines the number of questionnaires per region completed by each of the 41 fieldworkers.

Police officers were specifically requested to conduct the survey and made significant contribution to the success of the mapping exercise, particularly given the tight budgetary constraints within which the project was operating. However, financial considerations were not the primary rationale for their involvement in the process.

The police were used as surveyors for a number of important reasons:

At the start of the survey, many of the police officers did not see the relevance of the survey to their work, but once the connection to victim profiling, detective or investigation skills-development and improving an understanding of the dynamics of insecurity were made, the utility of the exercise was praised.

The police officers on the course were extremely junior. Many of them had never been on a training course besides their basic training and saw their involvement in the exercise as an opportunity for advancement within the police. Police officers from the more far-flung regions felt particularly marginalised from police processes and were glad to be involved in the exercise.

The quality of the survey completion was excellent. Most surveys were returned with extensive qualitative reports summarising the impressions and findings of the respondents. These were submitted to the Tanzanian police for their purposes.

It could be questioned to what extent the utilisation of police officers to conduct the survey would bias the study. It would be naive to try claim that the respondents would not be influenced by the fact that police officers were undertaking the survey. The biggest impact on the results of a survey of this type with police interviewers will be the reluctance to talk frankly and openly because of general suspicion of the police's motives, a general societal perception of the police and a desire not to implicate oneself or one's friends in any criminal activity. With these considerations in mind certain measures were employed in the survey design to off-set this impact including: the extensive nature of the survey, covering a comprehensive range of issues; the placing of sensitive firearm questions among enquiries about a range of other socio-economic information helping to make the survey non-threatening; and the pitching of potentially incriminating questions at a community rather than an individual level.

The margin of error in using the police to conduct the survey was not calculated. However, when reading the report, one should bear in mind that there may be a bias, especially with regard to the security-related questions and those questions implying illegal activities. This bias is not unusual in other quantitative surveys in which threatening questions are contained, and under-representation is expected to occur (especially on firearm ownership questions).

Survey design

In meeting the objectives of the survey, it was necessary that the survey be designed to accurately record the nature and extent of firearm penetration within Tanzania and the perceptions of the people of Tanzania towards small arms and to identify possible agents of change for the national implementation plan. The survey utilised questions which experience had shown were both reliable and consistent socio-economic, as well as security and firearm penetration, indicators within the African context.

A range of design techniques were used to ensure the reliability of the survey and to compensate for possible respondent bias resulting from the fact that police officers were conducting the survey. These design techniques included, amongst others, removing all threatening questions (for example, those that could implicate the respondent or their friends/family in illegal activities); locating questions about firearm ownership and illegal activities within a cluster of other socio-economic indicators; and ensuring that there was a fair balance between the various sections of the questionnaire in order that those relating to security were less obvious. The surveys were bilingual (in English and Swahili).

The survey was conducted as a partnership between the Tanzanian Police Force, the Institute for Security Studies, Saferworld, and the Security Research and Information Centre. Each organisation provided the resources as available in a collaborative manner. The survey was made possible by the fact that the Tanzanian Police offered to cover the main cost factor, namely the fieldwork staff.

Data considerations

The data represented in this report is not weighted. The responses presented assume an even age and gender distribution within each region. Usually, the percentage value indicates the varying proportion of the responses to each question by region. Throughout this report, summary tables of the means (averages) are presented. These tables indicate the mean response per region for a particular question. A mean key table is provided at the end of each mean report table to assist with the interpretation of the data. The mean minimum and maximum value is presented.

The number of respondents for each question is indicated with 'n=…'. In some cases, the number of respondents is less than the total sample for that category of respondents. This occurs in those cases where data is omitted. Omitted data or missing information occurs due to a variety of reasons, including:

the respondent's refusal to answer the question;


the question not being applicable to the respondent;


the respondent not knowing the answer; or


human error (for example, data capture error or fieldworker omissions).

Methodological assumptions

Undertaking a quantitative survey is based upon a number of critical assumptions. In this case, it is assumed that the community have a need for a survey such as this both in form and content.

However, the community may not necessarily have a need for, or appreciation of, the survey, or adequate resources to ensure its completion. The survey was conducted on the assumption that there was a need to provide information to assist communities in Tanzania to understand the dynamics of firearm proliferation. The need for the survey was identified by the government of Tanzania in response to their growing concern about the number of firearms moving within and through their territory and the destructive and harmful impact that the presence of firearms appeared to be having.

There is an assumption that the community appreciates quantitative data. There are ranges of methodologies available for the collection of data. Often those more interactive and qualitative methodologies, such as participatory research, focus groups and observations, provide more textured information—especially in light of the need to understand the micro-processes at play. Quantitative surveys (as utilised in this study) are critiqued for 'boxing' respondents into predetermined categories, for not allowing for nuances and for lending themselves towards blueprint planning, as classically used in top-down planning approaches. In an attempt to overcome some of the limitations of the quantitative survey there are some remarks that need to be made. The research team is well aware of the limitation of the quantitative survey, and saw this survey as a first step to scan the extent of the problem of firearm proliferation and understand how firearm penetration relates to a set of broader socio-economic indicators. Should a community wish to undertake actions to reduce firearm proliferation, they would be able to draw upon their own experience to identify possible programmes, with or without this study.

The utility of the survey lies in its broad nature—as applied across 12 regions—exploring a range of security, social and economic indicators. It provides a scope of the dynamics of firearm proliferation within Tanzania.

Sample overview

This section provides a profile of the respondents. As determined by the sampling technique, 53% of the respondents were male and 47% were female. On average, the respondents were 41 years old, with the youngest respondent being 18 years and the oldest 97 years old.

Figure 2: Age and gender profile respondents (n=2,876)

It was found that 27% of the respondents were single, and 50% of the respondents were married (the rest being either divorced, separated, widowed or living with a partner).

Figure 3: Marital status of respondents (n=2,889)

The size of the households varied from single person households to those larger households with more than 14 people. The majority of the households comprised between four to seven people.

Figure 4: Household size (n=2,891)

In the sampling methodology, fieldworkers were encouraged to sample people that were Tanzanian residents. It was found that all but seven respondents were Tanzanian. The other nationalities sampled included respondents from Burundi, Kenya, Mozambique, Rwanda and India.

The parentage of the respondent's mother and father was obtained in the survey in order to see the extent of integration within the region. The majority of the respondents had two Tanzanian parents. Among the respondents that did not have Tanzanian parents there were a range of nationalities comprising the parentage of the respondents.

Table 3: Nationality of respondent, and their parents

	Nationality of respondent's father	Nationality of respondent's mother	Nationality of respondent
Tanzanian Burundi England India Italy Kenya Malawi Mozambique Rwanda Somalia South Africa South Korea Yemen Total	2,871 0 0 2 1 2 2 9 3 0 1 1 0 2,892	2,845 7 1 2 1 10 5 14 4 1 0 1 1 2,892	2,885 1 0 1 0 1 0 1 3 0 0 0 0 2,892

The level of education of the respondents varied. Overall, 11% of the respondents had received no formal education and a further 64% of the respondents had received no further training since leaving school.

It was found that 36% of the respondents were employed part-time (6%), in the informal sector (15%) or in the formal sector (15%). Among the sample, 19% were unemployed.

Figure 5: Vocation of respondents (n=2,883)

Price | Subscribe