Kenya National Survey for Persons with Disabilities 2007
Sample Frame, Households [sf/hh]
The Kenya National Survey for Persons with Disabilities (KNSPWD) was a national sample survey - the first of its kind to be conducted in Kenya - designed to provide up-to-date information for planning, monitoring and evaluating the various activities, programmes and projects intended to improve the wellbeing of persons with disabilities. The survey covered more than 14,000 households in a total of 600 clusters (436 rural and 164 urban).
The survey interviewed persons with disabilities of all ages in sampled areas to get estimates of their numbers; distribution; and demographic, socio-economic and cultural characteristics. The survey also sought to know the nature, types and causes of disabilities; coping mechanisms; nature of services available to them; and community perceptions and attitudes towards PWDs.
The survey was undertaken by the National Coordinating Agency for Population and Development (NCAPD) in collaboration with the Kenya National Bureau of Statistics (KNBS); Ministry of Gender, Sports, Culture and Social Services (MGSCSS); Ministry of Health (MOH); and the Ministry of Education Science and Technology (MOEST). Other participants were United Disabled Persons of Kenya (UDPK); Kenya Programmes of Disabled Persons (KPDP); Association for the Physically Disabled of Kenya (ADPK); and Africa Mental Health Foundation (AMHF). Technical and financial support came from the Department for International Development (DFID), the World Bank and the United States Agency for International Development (USAID) under the Statistical Capacity Building Project (STATCAP) project. The United Nations Population Fund (UNFPA) provided support for the design of survey instruments.
Kind of Data
Sample survey data [ssd]
Producers and sponsors
National Coordinating Agency for Population and Development
Kenya National Bureau of Statistics
Department for International Development
United States Agency for International Development
United Nations Population Fund
While the survey intended to estimate the number of PWDs, it was realized that a significant proportion of these individuals reside in institutions, which are not part of the household sampling frame. However, a comprehensive list of institutions that existed did not form sufficient sampling frame for estimation of numbers of institution-based PWDs for the entire country. A mechanism had to be devised for incorporating these persons into the survey to supplement the data derived from the household-based survey.
The targeted survey population for the institutional based survey was defined as all people living in homes and occupying long-stay beds in public or private hospitals; or living in long-stay residential units for people with an intellectual, psychiatric/physical disability, vision or hearing impairments, or with multiple disabilities. The following types of institutions were covered:
· Hospitals (acute care, chronic care hospitals, nursing homes)
· Psychiatric institutions
· Treatment centres for persons with physical disabilities
· Residential special schools
· Private and non-private group homes
· Private and non-private children's homes
· Private and non-private residences for senior citizens (Mji wa wazee)
· Other residential institutions with people with disabilities
The sampling frame compiled for the institutional survey comprised all institutions indicated above. The frame included the name of the institution, type, number of individuals, location and type of disability. The frame was compiled from various sources, including MOH, MOEST, MSGSS and various organizations dealing with disabilities, among others.
In order to achieve representation, the institutions were first stratified according to location (provinces) and then by nature of disability. The institutions were further classified into two broad categories depending on nature and size (number of PWDs). All key institutions were sampled with certainty (that is, all selected in the sample). The remaining institutions within a province were arranged and serially listed by disability type and a systematic random sampling procedure used to select the sample.
A sample size of 102 institutions catering for different population sizes of PWDs was covered. Once the institutions were sampled, the next exercise involved selection of individuals for the survey. Five bands were created depending on the size of the sampled institution. The bands were: less than or equal to 30; 31-50; 51-100; 101-200; and above 200. A listing of all residents was compiled during the day of the interview and a systematic random sample drawn. Five respondents were selected from each of the sampled institutions with up to 30 PWDs, eight from those having 31-50, and ten from those having 51-100. For institutions having 100-200 PWDs, 15 were chosen, and from those having 201 and above, 20.
The KNSPWD household sample was constructed to allow for estimation of key indicators at the provincial level as well as of the urban and rural components separately. The survey utilized a multi-stage cluster sample design and was based on a master sample frame developed and maintained by KNBS. The master sampling frame is the National Sample Survey and Evaluation Programme (NASSEP) IV. It has 1,800 clusters (data collection area points) that were developed with probability proportional to size (PPS) from the enumeration areas (EAs) delineated during the 1999 Kenya Population and Housing Census. Of the 1,800 clusters, 1,260 are rural based and the other 540 are located in urban areas.
In the frame, the first stage involved selecting the census EAs using PPS and developing them into clusters. The process involved quick counting of the selected EA and dividing into segments depending on the measure of size (MOS). The MOS was defined as an average of 100 households, with lower and upper bounds of 50 and 149 households, respectively. The EAs that were segmented had only one segment selected randomly to form a cluster. The EAs that had fewer than 50 households were merged prior to the selection process.
During the creation of NASSEP IV, other than each of the 69 districts being a stratum, the six major urban areas (Nairobi, Mombasa, Kisumu, Nakuru, Eldoret and Thika) were further stratified into five income classes: upper, lower upper, middle, lower middle and lower. The aim was to ensure that different social classes within these areas were well represented in any time sample that was drawn.
The second sampling stage involved selecting clusters for the KNSPWD from all the clusters in the NASSEP IV master sampling frame. A total of 600 clusters (436 rural and 164 urban) was sampled from all the districts in the country with boundaries as defined in the 1999 Kenya population and housing census. The third stage of selection involved systematically sampling 25 households from each cluster, hence producing 15,000 households in total.
Mt. Elgon district was excluded from the survey because of persistent insecurity in the area. The effect of exclusion of the district in the sample is minimal since it contributes 0.5% of the population according to 1999 census.
A total of 14,569 households were covered. Response rate for the households was 97.4% (urban - 96.6% and rural - 97.7%).
Response rate for the individual reproductive health questionnaire (females 12-49 years) was 93.8% (rural - 94.3% and urban - 92.3).
Response rate for the individual persons with disabilities was 96% (rural - 96.2% and urban - 95.4%).
49.6% of the respondents were males while 50.4% were females.
Two-thirds of the household members had attained primary level of education; those with secondary level education were 21%.
The proportion of urban respondents with educational attainment above primary level was 46% while that of the rural residents was 22%.
The prevalence of orphanhood was slightly higher in the rural areas (11%) than in urban areas (9%).
Nyanza Province had the highest proportion of orphans (20%), Central Province had the lowest (8%).
To produce unbiased estimates, sampling weights were calculated as the inverse or reciprocal of all the selection probabilities at all the stages mentioned above. The weights were derived from the processes involved in the creation of NASSEP IV. Further, the weights were adjusted to cover household and individual non-responses. Post stratification adjustments were done to align with the population projections before the weights were finally normalized.
All the results from the household survey presented in the report, except for the response rates, are based on weighted data.
Dates of Data Collection
Data Collection Mode
Models of questionnaires and survey instruments developed by the World Health Organization (WHO), Washington Group Consortium and organizations in other countries were tailored to the Kenyan context. The purpose was not only to make the instruments responsive to the country situation, but also to ensure that the results would be comparable to those from other countries.
With input from a wide range of people who have worked in the area of disability, and who have conducted national surveys, a workshop was held to develop and adopt the following instruments for Kenya:
· Household questionnaire: Designed to collect background information at the household level for all the usual members as well as any visitors who slept in the household the night before the interview. This questionnaire was also used to screen PWDs by type to identify those who were eligible for the individual disability questionnaire. This instrument was administered to the most knowledgeable person in the household on the day of the visit.
· Individual questionnaire: Administered to any PWDs who had been identified using the household questionnaire. The questionnaire included the following key sections: activity limitation; environmental factors; situation analysis; support services; education; employment and income; immediate surroundings; assistive devices; attitudes towards disability; and health and general well-being
· Reproductive health questionnaire: Administered to all eligible females aged 12 to 49 who were living with any form of disability. It collected information on reproductive health.
· Institutional questionnaire: Administered to the heads of the various categories of institutions serving PWDs. Randomly selected PWDs in these institutions were interviewed using the individual questionnaire.
· Focus group discussion guide: Used to collect qualitative information from a group of 6-10 members within each of the sampled clusters. The groups comprised PWDs, community leaders, service providers, opinion leaders and teachers. The focus group discussions collected information on knowledge, attitudes and beliefs of community members about PWDs and the different services available for PWDs in the different communities. Likewise, focus group discussions were used to collect qualitative information about problems faced by PWDs, their coping mechanisms and their access to essential basic services, as well as an overview of community perceptions of PWDs and views on how best to address the needs of PWDs.
NCAPD availed ten computers for data processing. The computers were installed with Statistical packages for Social Sciences (SPSS), Nudist, and Census and Survey Processing System (CSPRO) software programmes. Programmes for data entry and analysis were written, edited and tested. Mechanisms were put in place for adequate data quality control checks. A team of 13 data entry clerks and a supervisor worked on the data processing. This included receiving the questionnaires, office editing and coding for open ended questions; data entry; and verification. The data processing commenced two weeks after the start of fieldwork so that the data entry team would have sufficient questionnaires for data entry and verification. The team underwent a four-day training in data entry before being deployed.
Qualitative data analysis used QRS software using the Nudist programme (N6). This software facilitates accessing, managing, shaping and analysing textual data. Transcribed field notes on focus group discussions were imported into the qualitative software as text units, which were worked on according to topic of interest. All common themes were coded and categorized in a standard way using the “tree nodes” structure so that comparisons could be made across subsamples and easily analysed.