• No results found

3.7.1 – Sample study area

Sampling is the selection of a part to represent the whole. It takes place in selection of research topic, research site, the people to be studied, concepts and variables which are used, the data collected, methods employed and the relationships in which the analysis is followed69. The decision to conduct the research in Harare alone was largely determined on the degree of researchers safety and financial constraints as most politically related violence was occurring in remote areas outside the city.

Harare was also selected for various other reasons. It is a dynamic city and currently presents significant disparities between the rich and the poor due to the economic crisis making it an ideal sample. I had full supervision from the University of Zimbabwe, where I was allocated office space and had to report to the office constantly hence kept the travelling to an absolute minimal due to financial constraints and fuel shortages.

Fig 3.4 shows the areas in which the research was conducted. This map was tailor made only for the purpose of this research at the Surveyor General Office in Harare.

69Peil et al, 1982

Fig 3.4 Sample study area

(Source: Surveyor General Office in Harare)

In Zimbabwe, urban housing types are categorised simplistically by ‗density‘. Harare can conveniently be divided into the city centre, 'high-density' suburbs and 'low-density' suburbs. Low Density Areas (LDAs) were generally the former white areas with freehold tenure and good quality large houses. They comprise for the most part large bungalows, on large plots of ground, situated along tree-lined avenues.

The low-income housing areas were termed High Density Areas (HDAs) and included the colonial era ‗townships‘, where the houses were formerly rented and built by the Government or municipality, such as Mbare and Highfields, and post colonial site-and-

service settlements like Kuwadzana and Budiriro. The high- density suburbs have small plots, crammed together, with little space for gardens and trees. Houses are typically simple four-room buildings, but there is variation of size and type. Lodgers may rent a single room, or a rough wooden shack outside the main house, or perhaps even half a room. In the more densely populated areas there are around thirty persons living on each plot, with only one official dwelling unit and normally only one toilet.

The research used the above map for the study area by identifying:-

 Boundaries of residential an areas

 Upper and Lower income residential areas

 Major target groups in the area.

 Commercial activities, particularly those related to HIV transmission, including major sources of employment, nightclubs, bars and other liqueur outlets

 Nightlife (bars/hotel/nightclub) areas

 Health, education, religion, social and NGO services in an area

 Security and crime in an area

 Major highways and public transportation arteries as sexual risk behaviour is often more common along such routes

As the random sampling method was adopted to administer questionnaires, all the above criteria helped in the selection of the sample. Boundaries of residential areas helped in identifying the high and low income earners and their perception on HIV prevention.

Commercial, health, education, religion and major highways enabled the identification of different categories of the work force including commercial sex workers who are most at

risk populations in the contraction of the virus. The research was conducted in the following areas:-

 High Density Areas ( Mabvuku, Kuwadzana and Epworth70)

 Low Density (Mount Pleasant, Avondale, Borrowdale)

 Central Business District CBD including bus stations, departmental stores and shoppers.

 Parirenyatwa Hospital

 The University of Zimbabwe

Fig 3.5 Peri–urban Areas of Harare

70 Fig 3.5 Above shows the location of peri –urban areas such as Epworth which is one of the areas the research was conducted.

3.7.2 – Sample Population

The challenge in sampling is the difficulty in making it representative in sexual behavioural change studies, as the determinants of the epidemiology of HIV vary across time and space in Zimbabwe, and as the exact factors relating to each audience of transmission are unique.

Mapping of risk behaviour by population groups is essential for the effective targeting of prevention efforts hence population sampling was aimed at obtaining unbiased and representative samples. However, the problem with sampling is the difficulty of making it representative hence I took the following approach:-

 I visited the Central Statistic Office71 in Harare for information on the Enumeration Zones in the city, the population, demographic status and the gender ratios between men and women by using the 2002 Census Report, which was the latest at the time the research was conducted as the census only takes place every 10 years.

- The City Council of Harare was also approached for information regarding the traditional land use patterns. This was fundamental in mapping out the residential areas considered as affluent and those considered poor.

In general principle, it is best to have as large a sample as possible to increase the possibilities of analysis, increase representation of the population and decrease the sampling error. As a precaution however, I did not collect more than I could handle, hence used a realistic figure of 364 questionnaires derived from using the following equations,

71 Now named Zimbabwe Statistical Agency (ZIMSTAT)

Determination of the number of Questionnaires administered

A formula of sample size determination was used to determine the number of questionnaires to establish using an attribute of P as the target population of 15 -49 year olds as a percentage of the Harare Population;

Where : -

N = sample size desired

Z = standard normal deviation72 set @ 1.96.

(corresponds to 95% confidence interval)

P = proportion target population with desired characteristics/ attribute (15 – 49 age groups)

Q = 1-p (a compliment of p, that is p + q = 1) D = degree of accuracy required,

(set at 0.05)

Therefore, our P = sexually active age group of 15 -49 years olds

According to the 2002 Central Statistics Office (CSO) data in Harare, the total sum of this age group is 880 378. Therefore, the proportion of the population of this age groups from the total population of 1 435 784 in Harare is:-

= 1 435 784 880 378

72 The standard deviation measures the spread of the data about the mean value. That is, it measures the variability of the data or how spread my data is.

= 61.31688

= 61.3%

= 61.3 for use as a proportion of the population 100

= P = 0.613%

Where : - z= 1.96 p = 0.613

q= (1.0 – 0.613) = 0.387 d = 0.05

=

= 364.53864384

= 364 Questionnaires administered

Determination of the number of Male and Female correspondents – Sex Ratio The total number of 15-49 year old in Harare at the time of the study was 880 378 With: - 442 987 Males

437 391 Females

The %age of males of the total population -

= 0.503178 × 100 = 50.3%

The %age of males of the total population -

= 0.496822 × 100 = 49.7%

Stratification of the sample

The number of questionnaires administered as per sex ratio – 364× 0.503178 = 183. 398 male respondents

364 × 0.496822 = 181.0816 female respondents

Therefore 183 (male) + 181 (female) = 364 questionnaires

The respondents were selected through purposive73 random sampling74 which is the selection of characteristics rather than individuals mainly for security reasons as the research was much safer to conduct in open public spaces as opposed to the systematic or stratified sampling in households. Another reason of using random sampling is that at the time the survey was conducted, Harare was experiencing severe food shortages where numerous long queues in for different commodities including food and fuel could be observed. This enabled us to effectively select participants who were very much interested in the research as this took off the frustration of queuing for several hours. The

73Respondents were selected through simple purposive sampling for reasons of security and convenience.

74 A simple random sample is selected so that all samples have an equal chance of being selected from the entire population.

advantage of this method is that each individual in a given population has an equal chance of being selected.

3.7.3 Background characteristics of the respondents.

Table 3.1, Tables 3.2, Fig 3.6 and Fig 3.7 show the socio-demographic characteristics of the sample population interviewed using questionnaires in Harare. Please note, there was one incomplete questionnaire after the research, hence the total number of questionnaires analysed is 363.

Fig 3.6 Age distribution of the total sample population (Source: Survey Data)

As Zimbabwe shows a classic age distribution of AIDS cases in which most cases are found among sexually active people, the study used the random sampling of 15 to 49 year

old age group which is the sexually active group with higher HIV prevalence rates. More than 70% of AIDS cases are among adults between the ages of 20 and 49. They are the economically productive part of the population hence deaths constitute an important economic burden. The smallest number of AIDS cases is in the 5 to 14 year old age groups emphasising the fact that the main mode of transmission is through sexual contact or perinatal transmission.

Table 3.2 Marital Status of the sample population

Variable Frequency %age

Education is an important factor in influencing an individual‘s attitude and outlook on various aspects of life, hence the educational level of respondents was among the most important variable in this research because it is associated with many factors that have a significant impact on health seeking behaviour, reproductive behaviour, use of contraception, and the health of children. Overall, the majority of Zimbabweans have attained some education, and there is very little difference by sex in educational attainment.

Literacy is widely acknowledged as benefiting both individuals and society. It is also associated with a number of positive health outcomes.