Sociodemographic characteristics of the sample
The participants were asked to indicate their age, marital status, income, employment, religion, housing quality, educational status, number of people in the house, migration status, rural-urban linkages, ownership of property in rural areas, and alcohol and smoking habits. The sample consisted of 388 respondents who were aged 18–88 years. About 36% of the respondents were men, and 64% were women. Their mean age was 38.6 years (SD=16.2).
The largest group of the respondents was single (43.5%), and others were cohabiting (16.8%), married (22%), separated (3%), widowed (9%), or divorced (2.5%). The population was ethnically heterogeneous. Forty-seven percent of the respondents were Bakalanga. The remaining respondents are distributed among the Bangwato (14%), Bazezuru (8.5%), Batawana (8.5%), Bahurutshe (7%), and other ethnic groups (9%).
Table shows the distribution of the respondents by age, income, marital status, educational status, employment status, number of children, and number of people in the household and religious affiliation.
Respondents by sociodemographic characteristics by sex
Thirty-one percent of the respondents had no children, 23% had one or two child(ren), 15% had three or four children, 22% had 5–7 children, and one respondent had 12 children. The mean number of children was 2.3 (SD=1.2). Approximately 35% of the respondents had no religious affiliation compared to 65% who had some religious affiliation. The respondents were distributed among African Independent Churches (38%), Pentecostal Churches (15%), Protestant Churches (6%), and traditional religions (2.8%).
Most (85%) respondents had some form of formal education. Approximately 35% had primary education only, 27% had Junior Certificate only, and 20% had an O’ Level Certificate (Form 5). Only three had participated in the adult education programme and had attained only primer 3 level.
One-third (29%) of the respondents were unemployed. Most of the unemployed were women. About 20% of the respondents were involved in micro-enterprises or informal sector. Women were most likely to be in the micro-enterprises/informal sector work, which includes food services, hairdressing, selling fruits and vegetables, and dress-making. Most of these activities provide income for the family. Eleven percent were engaged in professional jobs, such as accountancy, teaching, nursing, and public administration. Blue-collar workers, as they are normally called, such as drivers, messengers, and cleaners, accounted for 14% of the respondents.
The respondents were asked to indicate their monthly income. Their mean monthly income was P 801.32 (US$114), with a median of P 425 (US$ 61). About 27% did not have an income. We categorized income data into five groups (No income; P 400 and less; P 401–1,000; P 1,001–3,000; and P 3,001–6,000). In Botswana, anyone earning below P 400 is considered to be below the poverty datum-line. Twenty-three percent of the respondents were earning income below the poverty-datum level. Only seven respondents earned between P 4,400 and P 8,400. There were more women earning income below the poverty-datum level than men (17% compared to 6%).
Area of residence
Francistown has a number of low-income areas. For the purposes of the study, the areas were divided into five categories. Gerald Estates and Pelotelele were put in the same category because they are made up of people who are mainly ex-squatters, most of whom were evicted from Somerset West, Kanana, Matjimenyenga, and PWD. Table shows the distribution of the respondents by area of residence.
Environmental quality and social capital by area of residence
A majority (68%) of the respondents were living in houses of Self Help Housing Agency (SHHA). The remaining respondents were distributed among those who were renting (27%), those who rented some parts of their houses for money (2%), and those who were sharing the house with somebody else (4%). Most (83%) houses were built of masonry, 54% of which belonged to women. Twenty-six percent of the houses were supplied with electricity. A majority (65%) of the houses used paraffin and candles compared to 26% who used gas lamps for lighting. About 45% of the houses were built of cement while the remaining houses were built of stone, bricks, and other materials. About 36% of the respondents fetched water from the street-pumps while 40% had water piped to their yards. Only 14% had in-house sanitation, and 17% had a flush toilet. Eighty-four percent of low-income urban residents still use pit-latrines compared to 9% who use waterborne system.
According to the Urban Development Plan 10 for Francistown City Council, the population density of Francistown is higher in lower-income areas, such as Monarch and Gerald Estates, compared to higher-income areas, such as Molapo Estates and Ntshe. This is mainly attributed to affordability levels on various income groups. The high population density in low-income areas has posed problems for development, and services provided by the Government are extended far beyond their planned boundaries. Subsequently, these highly-populated areas still source water from public standpipes and rely on the use of pit-latrines.
The results for zero-order correlations are discussed by hypothesis. Hypothesis 1 was that the environmental quality will positively influence the overall health status and quality of life and the physical, psychological and level of independence domains of health. There was a strong and positive correlation between the environmental quality and the overall health-related quality of life (r=0.71, p<0.001), a moderate and positive correlation with physical health (r=0.36, p<0.001), a strong and positive correlation with psychological well-being (r=0.64, p<0.001), and a moderate but positive correlation with level of independence (r=0.43, p<0.001). The higher the perceived environmental quality, the better were the reported health outcomes.
It was further hypothesized that social capital would be positively correlated with the overall health status and quality of life and with the physical, psychological and level of independence domains of health. There was a weak but positive relationship between social capital and the overall health status and quality of life (r=0.10, p<0.06). There was no relationship between social capital and the level of independence (r=0.01, p=0.93). The only significant, albeit weak, relationship was between social capital and psychological well-being (r=0.18, p<0.006), suggesting that people who lived in environments they considered to have high levels of social capital tended to also report higher levels of psychological well-being. An unexpected finding was an inverse and weak relationship between physical health and social capital (r=−0.17, p<0.001). This result would suggest that when social capital is high, people perceive or experience poor physical health.
Test of moderation of social capital
Multiple regression analyses were used for testing the moderation model by examining the hypothesized variables (environmental quality) and the hypothesized moderator variables (social capital) respectively, after controlling for antecedent variables. To test for the moderation effects, first, the control variables (age, migration, housing, job-category, and income) were entered in the equation. Second, the dependent variable was regressed on the independent variable to determine the main effects of the independent variable on the dependent variable. Third, the dependent variable was separately regressed on the moderator variable to determine the main effects of the moderator variable on the dependent variable. Finally, after controlling for main effects, the dependent variable was regressed on the interaction product term to determine its interaction effects. Where there is a statistically significant main effect of the independent variable on the dependent variable and a statistically significant interaction effect of the independent variable and the moderator, it was taken that there is evidence of moderation. When an interaction was found, social support was trichotomized in low, medium, and high to determine the nature of this interaction.
The moderation hypothesis was that social capital would moderate the relationship between the poor environmental quality and the overall health status and quality of life and the physical, psychological and level of independence domains of health. Significant main effects were found for the environmental quality and overall health status and quality of life (beta=0.61, p<0.001), physical well-being (beta=0.47, p<0.001), psychological well-being (beta=0.51, p=0.001), and level of independence (beta=0.23, p<0.001). The main effect for social capital on the overall health status and quality of life was not significant (beta=0.08, p=0.34). Significant main effects were found for social capital on physical health outcomes (beta=−0.17, p<0.001) and psychological well-being (beta=0.16, p=0.001). The main effect for social capital on the level of independence was also not significant (beta=0.02, p<0.90).
There was no significant interaction effect of social capital and the environmental quality for the overall health status and quality of life (beta=0.24, p=0.22) and for psychological well-being (beta=0.03, p=0.15), suggesting that social capital does not buffer the effects of the poor environmental quality on the overall health status and quality of life and on psychological well-being.
There was, however, a positive significant interaction effect of the environmental quality and social capital for physical well-being (beta=0.62, p=0.02). The buffering hypothesis was supported. Social capital buffers the impact of the poor environmental quality on physical well-being. People who lived in communities which they considered to have high social capital tended also to report better physical health outcomes. The buffering hypothesis suggests that when individuals live in communities that they perceive to have high social capital, they tend to have better physical health outcomes even when their environments were poor. Low social capital was also beneficial for people who reported poor environmental quality. The results showed that the interaction made a significant contribution to the R2 (R2=0.01; F (5,382)=30.882; p<0.001). The product term involving the environmental quality and social capital produced a significant regression coefficient (t=1.980; p<0.001). The R2 for the total model is 0.29 F (5,382)=3.922 (p<0.001). This indicates that 29% of the variance in physical health outcomes could be explained by the two variables. This indicates that social capital is a moderator of the poor environmental quality. Table shows the results of moderated hierarchical regression analysis for the environmental quality, health-related quality of life, and social capital.
Moderated hierarchical regression analysis results for environmental quality, health-related quality of life, and social capital
There was also a significant interaction effect of the environmental quality and social capital on the level of independence (beta=0.34, p=0.02). These results suggest that high social capital buffers the impact of the poor environmental quality on the low level of independence. At high levels of social capital, the effects of the poor environmental quality on the level of independence are reduced, such that those people will report high levels of independence. The buffering hypothesis of social capital on the environmental quality was partially supported.