Pula: Botswana Journal of African Studies, Vol. 14 No.1 (2000) Demographic factors and party preferences in Botswana: a 1999 voters' survey Zifha S. Mokomane Democracy Research Project University of Botswana Abstract In this anicle, data from a survey conducted in March 1999 by the University of Botswana Democracy Research Project in fifteen constituencies in Botswana is used to examine the relationship between party preference and selected demographic characteristics namely, age, sex, place of residence and education. Bivariate analysis results indicate the ruling BDP is the only pany that enjoys higher support among women and among rural dwellers. Opposition parties, on the other hand, have a stronger suppon base among middle-aged people in the 29- 49 years age bracket. The results also indicate that the level of education tends to have a major role in determining party preference. The last section of the article critically discusses these findings. Introduction The purpose of this article is to identify factors, or a combination of factors, that mold party preferences of the electorate in Botswana. In particular, the relationship between party preference and selected demographic characteristics - age, sex, rural/urban residence and education will be examined. Such an inquiry will not only enable us to account for the changes in the pattern of Batswana's electoral behaviour but it will also provide an insight about the potential course of democratisation in Botswana. Theoretical Framework Studies of electoral behaviour have noted that party identification or preference, a concept that refers to a psychological attachment to a political party may make it easier for an individual to understand and evaluate political information by providing a reference point. In this sense, party preference is a perceptual screen through which the voter perceives politics. For example, a person who prefers party X may respond one way toward a policy advocated by a Party Y candidate, but a different way if the same policy is advocated by a Party X candidate (Clotfelter and Prysby, 1980:33). A variety of factors that may include personal circumstances, social and economic conditions and political problems of a country as well as ways that parties divide over issues may induce these partisan attachments. Research conducted in many democracies around the world has found that people of different age levels tend to differ in their party preference. The general rule is that the left does better with younger groups while the old are usually conservative in their choice of parties (Harrop and Miller, 1987:203; Frears, 1991:137). This is probably because left- wing parties are usually more secular and of more recent vintage than parties of the right, and therefore are more attractive to younger people. This apparent difference in party preference by age, however, does not exist because growing older greatly affects partisan attitudes, but because people form their party identification in different political climate. Thus the relevant fact about a person's age, in terms of its relationship to party preference, is that it indicates when the person grew up, not how many years the person has lived (Coltfelter and Prysby, 1980:42). A difference also exists between urban and rural electorates in terms of political party preference. Urban areas are usually the location for major economic and political decisions and are usually located near resource-rich areas. By contrast, rural areas generally control few resources, are not as developed as urban centres and are often isolated from other 59 regions. Consequently, a political agenda that reflects the nee.ctsand the soci?~eco~omic problems of the urban areas is usually different from the traditional rural pohtlcallssues (Kalaycioglu, 1994:411)... Education influences party preference to the effect that the manner III which each political party articulate and present their ideologies may result in the differences in their preference by different educational categories (Nengwekhulu, 1985:7). The absence of formal education does, to a certain extent, influence the rate and level of absorption of the content of party ideologies, polices etc. and limited education has been associated with support for politicians who offer simple solutions to complex problems. These contrasts help to account for a persistent finding of the tendency for the best-educated section of the middle-class to offer less support to conservative parties than middle class people with more limited education (Nengwekhulu, 1985:8, Clotfelter and Prysby, 1980:31). In the years before the 1970s, women in many countries proved to be more conservative in their electoral behaviour than men such that they were more likely than men to support the major party of the right than of the left. A common explanation for this was that on average, women live longer than men and, as discussed earlier, older voters are often heavily conservative. However, as the role of women in society changes, so too does their electoral behaviour and, in most countries today, gender differences in voting as well as partisan attitudes are declining. The most obvious reason for this is the massive emergence of women onto the paid labour markets especially in routine white-collar jobs. This has increased trade union membership among women and in other ways has marked a break with traditional roles (Harrop & Miller, 1987:205). Methodology The data for this study are derived from a survey conducted by the Democracy Research Project of the University of Botswana in March 1999. The survey was conducted in fifteen constituencies namely Gaborone West, Gaborone South, Gaborone Central, Lobatse, Selibe-Phikwe, Francistown East, Francistown West, MauniChobe, Mogoditshane, Moshupa, Kanye, Shoshong, Serowe North, Kgatleng East and Ghanzi. Trained enumerators administered a structured questionnaire, which included both closed ended and open-ended questions on a number of issues regarding the 1999 general elections. A total of 4365 respondents aged 18 years and above, and who were citizens of Botswana were interviewed. The dependent variable of this study is party preference. Respondents were asked the question "If elections were held today, which party would you vote for?" Without probing, the respondent was expected to name one of the thirteen political parties that were registered in Botswana at the time of the survey. The independent variables are sex, age, place of residence and education. Bivariate techniques, which employ summary statistics ?f percentages and means, are used to demonstrate the magnitude of subgroup differences In party preference. Results !he basic socio-ec~n~mic and demographic characteristics of the respondents are provided In Tab!e 1. The maJonty of the respondents in the survey (57 percent) were female. Almost one-third were aged 18-24 years and 51 percent were urban residents In addition, 33 percent of the respondents had primary education while 60 percent reported that they were unemployed. Overall, the Botswana Democractic Party (BDP) was supported by 60 percent of the respondents while 'Others' had the lowest support (3 percent) among respondents. 60 a e : ercental!e Distribution TbilP of Resoondents bv Backl!round Characteristics Sex Male 42.9 1873 Female 56.7 2475 Age 18-24 29.0 1268 25-29 15.0 655 30-34 10.6 463 35-39 9.2 402 40-44 7.6 332 45-49 5.6 243 50-54 4.6 202 55+ 17.6 769 Place of Residence Urban 50.9 2223 Rural 49.1 2142 Education None 19.5 850 Non-formal 2.1 93 Primary 33.0 1440 Junior Secondary 26.8 1169 Senior Secondary 15.3 667 Tertiarv 2.7 120 Employment Status Unemoloved 60.3 2138 Emploved 39.7 1406 Party Preference BCP 8.1 268 BOP 60.0 1996 BNF 28.5 947 Others 3.4 114 Note: N's may vary because of missing data in some variables; BCP - Botswana Congress party, BNF - Botswana National Front. Others - All otber political parties except BCP, BOP and BNF. Source: 1999DemocracyResearchProjectVoters' Survey Table 2 presents data on respondents broken down by the selected demographic c~aracteristics. Results in Table 2 indicate that, contrary to recent literature that gender differences in partisan attitudes are declining in most countries around the world, men and women in Botswana seem to differ notably in their party preferences. Except for the BDP, a~l political parties enjoy higher support among women than among men. While the difference is quite significant among BCP, BDP and BNF supporters, it is less significant for supporters of 'Others' (3.8 percent and 3.2 percent for men and women respectively). It also emerges from this table that there is quite a strong association between age and ~arty preference. Although preference for the BDP is relatively higher in all age groups, it is much higher in the 18-24, 50-54 and 55+ age groups (62, 63 and 72 percent respectively). The BCP, on the other hand, has a stronghold among the 25-29 year olds and, to a lesser extent, among the 30-34 and 40-49 year olds. Support for this party is very low among people in the 35-39 age group and among older people aged more than 55 years. The BNF enjoys high support among people in the 25-39 year age bracket. Otherwise preference for the party is approximately the same level (in the 20 percent range) in all age groups above 40 years. Other parties have relatively very low support 61 across all age groups, with their highest support being 6 percent among people aged 50-54 years. In most of the age groups, their support is very low at around 3 percent. Table 2. Percentaee distribution of Parh Preference b Selected DemoeraPI hic Varia. bl es Variable BCP BDP BNF Others Total N Sex Male 11.1 51.9 33.2 3.8 100 1873 Female 5.8 65.8 25.2 3.2 100 2475 Al!e 18-24 8.0 61.5 27.5 3.0 100 1268 25-29 13.6 49.5 33.4 3.4 100 655 30-34 10.6 56.0 30.8 2.7 100 463 35-39 4.9 54.4 35.4 5.3 100 402 40-44 10.7 56.9 27.9 4.6 100 332 45-49 10.1 59.0 28.2 2.6 100 243 50-54 6.7 62.8 25.0 5.5 100 202 55+ 2.6 71.8 22.7 2.9 100 769 Place of Residence Urban 10.6 54.8 29.3 5.3 100 2223 Rural 5.7 64.9 278 1.7 100 2142 Education None 5.5 66.0 25.0 3.5 100 850 Non-formal 4.3 63.8 24.6 7.2 100 93 Primary 7.5 61.0 28.4 3.0 100 1440 Junior 9.0 55.7 32.0 3.3 100 1169 SecondarY Senior 10.1 59.2 27.4 3.2 100 667 Secondary Tertiary 17.6 47.3 28.4 6.8 100 120 Source: 1999 Democracy Research Project Voters' Survey Analysis of the data by rural/urban residence shows that the BDP is the only party that has a stronger support base in the rural areas (65 percent compared to 55 percent in urban areas). The BCP, BNF and others, on the other hand, have stronger support bases in the urban areas. While there is no significant difference between the proportions of the BNF's urban and rural support (29 and 28 percent respectively), the BCP's support in rural areas (6 percent) is almost half of its urban support (I I percent). In the same vein, Others' urban support is almost three times their rural support. Data analysis by educational level reveals that the BCP's support increases with educational attainment. The party has very low support (less than 10 percent) among people who have junior secondary or lower education, the lowest being 4 percent among people with non-formal education. Its support is 10 percent and 18 percent among senior secondary and tertiary graduates respectively. Conversely, support for the BDP decreases steadily with educational attainment. From 66 percent among those who have no education, its support decreased to about 64 percent among those with non-formal education and it kept decreasing until it reached its lowest level at about 47 percent among those with post-secondary education. The BNF on the other hand, has the highest support (32 percent) among people with junior secondary education. Otherwise in all other age groups, its support is rather consistent in the 25-28 percent range. Other parties have their support base among those people with non-formal education (7.2 percent) and those with tertiary education (6.8 percent). For all other age groups, the support is much lower and constant at about 3 percent. Discussion In this article, the relationship between party preference and selected demographic charact~ristics has been explored. Bivariate analyses indicate that contrary to findings ~n recent lIterature, there exists a major difference between Batswana women and men III terms of political party preference. That women are more inclined than men to support the 62 BDP than opposition parties is consistent with earlier literature which argues that women are usually more conservative and more likely to support the major party of the right. These results may also be an indication that mobilised voting, where women follow the political behaviour of their male relatives such fathers, husbands and brothers is decreasing among women and in its place there is a new breed of women whose political acts are independent from that of their male relatives. Although the BDP has significantly higher support than all other parties across all ages, the general pattern is that opposition parties are more likely to be preferred by younger people while the BDP enjoys more support among the older generation. This may be due to the fact that the young are normally more educated and therefore more receptive of new and radical ideas (associated with most opposition political parties) whilst the older generation is attracted to a more moderate and conservative outlook-something which the BDP is associated with. Additional factors may be that most young people in the sample were urban residents and, for reasons to be discussed later, they are more likely to support opposition parties than the BDP. The analysis also uncovered a very interesting finding of the significantly high preference for the BDP by young people aged 18-24 years. This finding is interesting for a number of reasons. Firstly, the finding contradicts predictions made by some Botswana politics scholars in the late 1980s and early 1990s that if the voting age is reduced to 18 years (which was done in 1998), the opposition's, especially the BNF's, share of popular support will be increased phenomenally. Secondly, with the prevailing high unemployment rates in this age group, the assumption would be that the youth in this age group would show disillusionment with the ruling party and would, in the hope of getting jobs and a better life, prefer the opposition parties more than the ruling party. Finally, the finding does not reflect the fact that majority of the youth in this age group are urban dwellers and are fairly educated with most of them having junior secondary education. Although this finding may be a reflection of the youth's disillusionment and discouragement with opposition parties caused mainly by the split of the BNF in 1998, the situation is important and requires further research and explanation. It is clear from the results that urban dwellers are more inclined to support opposition parties while majority of rural dwellers tends to support the ruling BDP. Since the literature argues that older people tend to be conservative in their choice of political parties than younger people, the high proportion of old people (aged 50 years and above) residing in rural areas compared to urban areas may explain this pattern. With respect to opposition parties, the common view that the BCP is elitist, and for the educated urbanites, may explain the party's very low support among the less educated rural dwellers compared to the urban residents. On the other hand, parties that fall under 'Others' have been associated with radical, socialist ideas which tend to appeal more to the young and educated section of the society, majority of whom are urban dwellers. This may explain why 'Others' enjoy higher support in urban areas than in rural areas. Unlike all other parties, the BNF's support does not seem to be affected by the population's age structure and little difference is observed between its urban and rural support. This is, probably because the party has more or less constant support across all ages and educational categories. Another plausible reason for the observed pattern of party preference by urban/rural residence is Botswana's high urban growth rates. According to Campbell (1995), the average annual urban growth rates increased from 10.1 percent in 1964-71 through 11.8 percent in 1971-81 to 13.8 percent in 1981-91 with rural-urban migration playing a major contributory role in this high urbanisation rate. Consequently, most of the urban areas in the country are inundated with migrants who face various problems such as high and worsening unemployment rates, retrenchments, crime, shortage of accommodation etc. A political agenda that reflected the needs and the socio-economic problems of the increasingly overcrowded urban areas eventually emerged in the 1990s and major urban areas in Botswana stated to experience increasing political participation of both the 63 conventional and protest form. An example is the 1994 general elections when the opposition BNF won all the traditionally BDP parliamentary seats in the urban areas of Gaborone, Selibe-Phikwe, Lobatse, Mogoditshane and Francistown West. Thus the urban political agenda, which is different from the traditional rural, or modernising rural, political issues in substance may be responsible for the apparent disparity between urban and rural dwellers in terms of party preference. Consistent with the literature, the level of a person's education seems to have a strong impact on his/her party preference. The general pattern is that support for the BCP increases with education while the opposite is true for the BDP. Although some 'less educated' areas in towns (e.g. Old Naledi in Gaborone and Botshabelo in Selibe-Phikwe) have historically been anti-BDP while 'educated' constituencies such as Gaborone Central has been a BDP stronghold, this finding manifests a national pattern and may be explained by the fact that the less educated section of society tends to be old people residing in rural areas. As seen earlier, majority of these are BDP supporters. It is not very surprising that support for other parties is relatively higher at tertiary levels because these parties are usually associated with complex ideologies that make sense only to few, highly educated and relatively young, people. It is however surprising and very interesting to note that these parties' highest support is found among people with non-formal education-a finding which calls for further research. As with other variables, the BNF enjoys consistent support across all educational categories showing that people, regardless of educational attainment, appreciate it equally. Conclusion In sum, the analyses show people do not make political party choices as isolated individuals. Rather, these choices are embedded in a variety of societal processes and structures and the position of people within these structures have an effect on their choice of political parties. References Botswana Government. 1970. Report to on the General Elections: 1969: Gaborone Botswana Government. Report to the Minister of State on the General Elections, 1974: Gaborone. Botswana Government. Report to the Minister of Public Service and 1nformation on the General Elections, 1979: Gaborone. Botswana Government. Report to the Minister of Presidential Affairs and PublicAdministration on the General Elections, 1989: Gaborone. Botswana Government. Report to the Minister of Presidential Affairs and Public Administration on the General Elections, 1994: Gaborone. Campbell, E.K. 1995. 'Population Distribution and Urbanisation,' in Proceedings of the 1991 Population and Housing Census Dissemination Seminar, May 1995. Central Statistics Office: Gaborone. Clotfelter, 1. and Prysby, C. L. 1980. Political Choices: A Study of Election and Voters. Holt, Rinehart and Winston: New York. Frears,1. 1991. Parties and Voters in France. St. Martin's Press: New York. Gordon, Martin and Miller, William L. 1987. Elections and Voters: A Comparative 1ntroduction. MaCmillan Education: London. Kalaycioglu, E. 1994. 'Elections and party Preferences in Turkey: Changes and Continuities in the 1990s' Comparative Political Studies, Yol. 27, No.3. Klecka, W. R. 1980. D~scriminant Analysis. Sage Publications: Beverly Hills. Moyo, J. N. ~992. Voting for Democracy: Electoral Politics in Zimbabwe, University of Zimbabwe PublicatIOns: Harare Nengwe.khul~, R. 1985. 'The 1984 General Elections in Lobatse-Barolong Farms Constituency' UmveTSltyof Botswana Election Study Group: Gaborone. Somolekae, G. 1?89. 'Do Batswana think and act as democrats' in 1. Holm & P. Molutsi (eds.) Democracy In Botswana. Macmillan Botswana Publishing Company: Gaborone. Tordoff, W. 198~. 'Democracy in Botswana: an external assessment' in J. Holm & P.Molutsi (eds.) Democracy In Botswana. Macmillan Botswana Publishing Company: Gaborone' in Final report 64