PMA2017/Kenya Round 7 Indicators

Round 1 Sample Design

During the first round in Kenya (2014) a multi-stage cluster design with urban/rural and county as strata was used. The first stage of sampling was at the county level using probability proportional to size procedures to select 9 out of 47 counties: Nairobi, Kilifi, Nandi, Nyamira, Kiambu, Bungoma, Siaya, Kericho and Kitui. Within the nine selected counties, 120 enumeration areas (EAs) were selected proportional to size with urban/rural stratification. The sample was powered to generate national and urban/rural estimates of all woman mCPR with less than 3% margin of error.

In each selected EA, field supervisors randomly selected up to three private service delivery points (SDPs) to be interviewed by an RE using the SDP questionnaire. The field supervisors themselves administered the SDP questionnaires at an additional three public SDPs that serve each EA - the lowest, second-lowest and third-lowest level public health SDPs designated to serve each EA (a dispensary, a health center and a referral hospital), either at the sub-county or county level. 

Round 7 Sample Design 

PMA2018/Kenya-R7, the seventh round of data collection in Kenya, used the same enumeration areas (EAs) from 11 selected counties as in Rounds 5-6 with a new random selection of 42 households. In Round 5, a new selection of 120 EAs in the original nine counties (Nairobi, Kilifi, Nandi, Nyamira, Kiambu, Bungoma, Siaya, Kericho, and Kitui) was done by the Kenya National Bureau of Statistics (KNBS) by randomly selecting EAs from the list of EAs that were geographically adjacent to Round 3-4 EAs. In two additional counties, Kakamega and West Pokot, the Kenya National Bureau of Statistics drew 31 new EAs from the master sampling using the same strategy as in Round 1. These same counties and EAs were used in round 6 (2017) with a new random selection of 42 households within each EA. In round 6, three EAs in West Pokot had their boundaries extended with help from KNBS.

In Round 7, all households, health service delivery points and key landmarks in each EA were listed and mapped by the REs to create a frame for the second stage of the sampling process. Field supervisors randomly selected 42 households using a phone-based random number-generating application. A household roster was completed and all eligible women age 15-49 were approached and asked to provide informed consent (and assent if aged 15-17) to participate in the study.

The majority of public SDPs are repeated in each round, forming a panel survey. If an EA had more than three private SDPs identified during the listing process, then a new, random sample of three private SDPs is selected during each round.

The final sample included 6,097 households, 5,671 females and 419 health facilities (97.9%, 99.1% and 97.9% response rates respectively). Data collection was conducted between November and December 2018.

PMA2020 uses standardized questionnaires to gather data about households, individual females and health facilities that are comparable across program countries and consistent with existing national surveys. Prior to launching the survey in each country, local experts review and modify these questionnaires to ensure all questions are appropriate to each setting. Three questionnaires were used to collect data from the PMA2017/Kenya-R7 survey: the household questionnaire, the female questionnaire and the service delivery point questionnaire.

The household, female and health facility questionnaires were based on model questionnaires designed by PMA2020 staff at the Bill & Melinda Gates Institute for Population and Reproductive Health and fieldwork materials of the 2008-09 Kenya Demographic and Health Survey.

All PMA2020 questionnaires are administered using Open Data Kit (ODK) software and Android smartphones. The PMA2018/Kenya-R7 questionnaires appeared in Kiswahili in addition to English. Female resident enumerators in each enumeration area (EA) administered the household and female questionnaires in selected households.

The household questionnaire gathers basic information about the household, such as ownership of livestock and durable goods, as well as characteristics of the dwelling unit, including wall, floor and roof materials, water sources, and sanitation facilities. This information is used to construct a wealth index.

The first section of the household questionnaire, the household roster, lists basic demographic information about all usual members of the household and visitors who stayed with the household the night before the interview. This roster is used to identify eligible respondents for the female questionnaire. In addition to the roster, the household questionnaire also gathers data that are used to measure key water, sanitation, and hygiene (WASH) indicators, including regular sources and uses of WASH facilities used by household members.

The female questionnaire is used to collect information from all women age 15 to 49 who were listed on the household roster at selected households. The female questionnaire gathers specific information on: education; fertility and fertility preferences; family planning access, choice and use; quality of family planning services; and exposure to family planning messaging in the media.

The SDP questionnaire collected information about the provision and quality of reproductive health services and products, integration of health services, and water and sanitation within the SDP.

Training

The PMA2018/Kenya-R7 fieldwork started with a refresher training which was held in October 2018. PMA2020/Kenya's implementing partner, led the training with (remote) support from PMA2020 staff from the Bill & Melinda Gates Institute for Population and Reproductive Health of the Johns Hopkins Bloomberg School of Public Health.

The objective of the refresher training was to address the gaps and errors identified during any previous rounds of data collection, to understand the questionnaire changes for Round 7, to refresh the knowledge and skills on questionnaire content and the art of asking questions through paired interviews. In addition, field staff were also reminded of key survey protocols they needed to abide by, including consent administration and research ethics.

Throughout the refresher training, REs and supervisors were evaluated based on their performance on several written and phone-based assessments and class participation. The RE trainings were conducted primarily in English, some small group sessions were conducted in Kiswahili. All training participants were given instructions on survey changes to the tools since the previous round. The REs and supervisors were all evaluated based on their performance on phone-based assessments.

Data Collection & Processing

Data collection was conducted between November and December 2018. Unlike traditional paper-and-pencil surveys, PMA2020 uses ODK Collect, an open-source software application, to collect data on mobile phones. All the questionnaires were programmed using this software and installed onto all project smartphones. The ODK questionnaire forms are programmed with automatic skip-patterns and built-in response constraints to reduce data entry errors.

The ODK application enabled REs and supervisors to collect and transfer survey data to a central ODK Aggregate cloud server. This instantaneous aggregation of data also allowed for concurrent data processing and course corrections while PMA2020 was still active in the field. Throughout data collection, central staff at ICRH-K in Kenya and the data manager at the Bill & Melinda Gates Institute at Johns Hopkins School of Public Health in Baltimore, Maryland routinely monitored the incoming data and notified field staff of any potential errors, missing data or problems found with form submissions on the central server. The use of mobile phones combined data collection and data entry into one step; therefore, data entry was completed when the last interview form was uploaded at the end of data collection.

Once all data were on the server, data analysts cleaned and de-identified the data, applied survey weights, and prepared the final data set for analysis using Stata® software.

The table shows response rates for household and female respondents by residence (rural/urban) for PMA2018/Kenya-R7. A total of 6,342 households were selected for the Round 7 survey; 6,230 households were found to be occupied at the time of the fieldwork. Of the occupied households, 6,097 (97.9%) consented to a household-level interview. The response rate at the household level was higher in rural (98.9%) than in urban (96.3%) areas.

In the occupied households that provided an interview, a total of 5,722 eligible women age 15 to 49 years were identified. Overall, 99.1% of the eligible women were available and consented to the interview. The female response rate was slightly higher in the rural (99.1%) relative to the urban (99.0%) enumeration areas (EAs). Only de facto females are included in the analyses; the final completed de facto female sample size was 5,671.

During the survey, 428 SDPs were identified of which 419 SDPs completed the survey (97.9% response rate).

To view the sample errors for the PMA2020 indicators described above, download the full SOI report here. For more information about PMA2020 indicators, including estimate type and base population, click here.