Individual-level household survey dataset for one urban and one rural constituency in Kenya and Zambia, covering questions on media and communications habits, political behaviour and attitudes. The objective of the surveys was to obtain representative samples of two constituencies per country. Constituencies were selected according to their social and economic characteristics, in order to capture a wide variety of contexts. A random procedure was deployed in all stages of sampling, ensuring representativity of households and individuals of voting age in the four constituencies. The results of the survey can be generalised to the particular constituencies with a margin of error of approximately minus or plus 5% for a 95% confidence interval.Politics and Interactive Media in Africa (PIMA) examines whether and how Africans, particularly the poorest and least politically enfranchised, use new communication technologies to voice their opinion and to engage in a public debate on interactive broadcast media, and its effects on modes of political accountability. Through detailed qualitative case-studies in Kenya and Zambia, PIMA critically interrogates the heralded potential for digital communications and liberalised media sectors to promote more responsive and inclusive democratic governance, with a keen eye for turning project insights into relevance for policymakers, media houses, journalists and development organisations. By employing survey-based, qualitiative and ethnographic methods to comparatively analyse interactive radio and TV programmes in the context of electoral and everyday politics, we will probe whose voice counts, why and to what effects in these new digitally-enabled spaces of voice and accountability. The project takes into account local innovation in the use of ICTs and the interactions between different modes, venues and actors of information gathering and dissemination that are particularly prominent in African contexts. PIMA brings together researchers from the Universities of Cambridge, Nairobi and Zambia, working closely with select broadcast stations and other stakeholders.

Data collection for the PiMA surveys took place during May 2013 (Kenya) and June-July 2013 (Zambia). In Kenya, surveys were conducted in Ruaraka: a peri-urban constituency in the capital city Nairobi, with mixed demographics including one of the city’s major slums; and Seme: a rural constituency settled around Lake Victoria in a largely fisher-agricultural community in the western Kenyan city of Kisumu. In Zambia, the surveys were conducted in Mandevu: an urban constituency in the capital city Lusaka with a mixed demographic including some of the city’s major slum settlements; and Chipangali: a rural constituency in the country’s largely agricultural Eastern Province. The four samples were designed as representative cross-sections of all households in those constituencies. Although no claim is made that the constituencies themselves were representative of the wider national population, they were selected based on the possibility of capturing variation in terms of socio-economic factors, political context and media landscape. A multi-stage sampling approach was deployed in the four sites, which involved selecting geographically defined units of decreasing size at each stage. The main four stages of the sampling strategy were: (1) cluster sampling for selection of wards; (2) simple random sampling for selection of enumeration areas (EAs) within wards; (3) systematic random sampling for selection of households within EAs (“random walk”); and (4) simple random (Kenya), or stratified by age and gender (Zambia) sampling for selection of individuals within households. Because there were no available lists of voting individuals living in those constituencies based on census data, the population was grouped into units from which reliable data was available, such as EAs. The lists of EAs constituted the sampling frame from which the primary sampling units (PSUs) were randomly selected. In Stages 2 and 3, selection was performed with probabilities proportional to population size. The purpose was to guarantee that more populated areas (wards, EAs) had a proportionally higher probability of being included in the sample. Within each household, individuals were selected using a random procedure. By employing random techniques in all stages of sampling, and using sampling with probability proportional to the population, it may be assumed that all individuals of voting age (18 and over) living in those four constituencies had a known and above zero chance of being included in the sample. The results of the survey allow inferences to the voting population in the four constituencies (macro-units) with some degree of accuracy (but not to the two countries). The sample sizes are 760 for Kenya (383 for Ruaraka and 377 for Seme) and 688 for Zambia (327 for Mandevu and 361 for Chipangali). The margins of error for a 95% confidence level are no more than plus or minus 5% for both Ruaraka and Seme, 5.41% for Mandevu and 5.12% for Chipangali. The response rate for Kenya was 90.4% (84.6% for Ruaraka and 96.3% for Seme). The response rate for Zambia was not available because the team did not record the number and reasons of unsuccessful calls.