IPUMS PMA Survey Notes

Survey Types

IPUMS PMA comprises several different type of surveys, collected by the same organization and using similar data collection techniques. A basic understanding of the structure and characteristics of each survey is essential for leveraging these data well. Below are descriptions of each file found on the IPUMS PMA online data dissemination system.

Household and Female surveys

The core of IPUMS PMA are the household and female surveys conducted by Performance Monitoring for Action (PMA) in 11 countries since 2013. These are cross-sectional surveys of sampled households and women of childbearing age (15 to 49) who reside in the household. The household respondent provides information about the wealth, water sources, sanitation facilities, and other housing characteristics. The female interview includes questions about family planning use, fertility preferences, fertility history, and other topics.

Data file structure: Each record is a person within a household. Women of childbearing age have household characteristics attached to their record. These files include nonrespondents to both the household questionnaire and female questionnaire. By default, the online dissemination system selects only female respondents, but users may choose to include household members and/or nonrespondents in their data extracts.

How to get these data: On the online data dissemination system, choose the Family Planning - Person unit of analysis.

Service Delivery Point (SDP) surveys

In parallel with the Household and Female surveys, enumerators interviewed a knowledgeable staff person at up to 3 public and up to 3 private health facilities in the same enumeration areas (EAID) as the household and female surveys in the same years. This allows researchers to use microdata to describe the service delivery environment in which the women live. See our user note on how to use SDP data with household and female data.

Data file structure: Each record is a health facility. In order to use these data with the household and female files, researchers need to calculate summary statistics at the EAID level and merge to a household and female file of the same country(ies) and round(s).

How to get these data: On the online data dissemination system, choose the Family Planning - Service Delivery Point unit of analysis.

Maternal and Newborn Health

In Ethiopia, PMA collects data on maternal and newborn health. Enumerators conducted household roster surveys of all households in sampled enumeration areas to find households with pregnant women. Then, a female screening survey was conducted for each pregnant woman. Enumerators followed up in person at 7 days and 6 weeks after the woman's delivery, and then either by phone or in person at 6 months after delivery.

Data file structure: Each record is an infant (both live and stillbirth). There are several cases of twins, in which case the attributes of the mother and the household are attached on each infant's record. Therefore, in each file, mothers of twins will be represented twice. There are also a small number of cases of households with more than one pregnant women. These households will also be represented more than once.

How to get these data: On the online data dissemination system, choose the Infant unit of analysis.

Nutrition

In 2017 and 2018, PMA conducted Nutrition surveys in Burkina Faso and Kenya, at both the person level and at the service delivery point level.

Household and Female

Data file structure: Each record is a person within a household. Enumerators collected household roster information from randomly selected households from sampled Enumeration Areas (EAs).

Round 1 (2017)

Households were eligible for the household survey if there was at least one female aged 10 to 49 and at least one child under age 2 living in the household. Females aged 10 to 49 were eligible for the female survey if they lived in a household with a child under the age of 2. Children under the age of 5 living in a household with a child under the age of 2 and a female aged 10 to 49 were eligible for the female-child nutrition survey. The female aged 10 to 49 answered the survey for the child under 5.

Depending on the child's age and other factors, each child was eligible for a different set of nutrition questions:

  • The first group was the youngest biological child in the care of the female under the age of 2 (these variables have a prefix INF and can be found in the variable groups Infant Nutrition and Infant Food Consumption).
  • The second group was children between age 2 and 5 who were the youngest child in the care of the female aged 10 to 49 (these variables have a prefix YK and can be found in the variable groups Young Child Nutrition).
  • The last group was children aged 2 to 5 who were not the youngest biological child in the care of the female (these variables have a prefix OTK and can be found in the variable groups Other Young Child Nutrition).
  • See the variable ELIGTYPE for more details. Users should note that children under 2 who were not the youngest child in the care of the female aged 10-49 were not eligible for the nutrition questionnaire.

    The original data was disseminated in three files - the household roster, female survey, and children nutrition survey. IPUMS restructured these data into a single data file using the following steps:

    1. Attached data from the female survey onto each of the women's records on the household roster using a individual female ID number (FCQINSTID) provided in the original data.
    2. Characteristics from the female's record was merged onto the child's record also using the female ID number, which represented the mother or caretaker of the child.
    3. Using children's birth month, birth year, and household identifier, the data from the child survey was attached to the corresponding record of the child in the household roster. This step is important because the child's gender was on the household roster, but not the Child file.
    4. Attached household level characteristics (water sources, wealth, etc) to children from the Child file that were not matched to records in the household roster.
    5. PMA had also constructed records for children who had died, according to the responses from the female-child survey. These records contained the birth year and death year of the child. IPUMS attached this data to the woman's record and omitted the records of the deceased children from the file.
    Round 2 (2018)

    All households were eligible for the household survey. All children under age 5 were eligible for the child survey. The primary caretaker of the child answered the questions for the child. Twenty-five percent of the households surveyed (see variable FQSELECTED) for the household roster were randomly selected for the female survey, even if a female aged 10 to 49 was not a part of the household. Females age 10 to 49 in these selected households were eligible to respond to the female survey. The data in this round were disseminated in one file per country. Unlike Round 1, the mother or caretaker's identification number is not present on the child record, so it is not possible to attach mother's characteristics onto the child's record.

    How to get these data: On the data dissemination system, choose the Person - Nutrition unit of analysis.

    Service Delivery Point

    Data file structure: Each record is a health facility. In order to use these data with the household and female files, researchers need to calculate summary statistics at the EAID level and merge to the Nutrition household and female file of the same country and round by EAID. See our memo about using household and service delivery point (SDP) data together.

    How to get these data: On the data dissemination system, choose the Service Delivery Point - Nutrition unit of analysis.

    COVID-19 surveys

    The PMA COVID-19 survey is a follow-up telephone survey administered to women who participated in an in-person baseline survey for a broader panel study. This baseline survey was collected between November 2019 and Februrary 2020 - prior to the appearance of COVID-19 in most countries.

    When the outbreak of COVID-19 grew into a global pandemic in the spring of 2020, PMA representatives partnered with the Ministries of Health in DRC, Kenya, Burkina Faso, and Nigeria to design a shorter - approximately 30 minute - survey responding directly to the effect of COVID-19 on women and their households.

    Women were selected for the baseline survey if they were age 15-49 and resided in a household screened at random from a sample cluster represented by EAID. All women surveyed at baseline where eligible to participate in the COVID-19 follow-up, provided that they 1) agreed to the interview, and 2) owned or had access to a telephone.

    You'll find survey weights adjusted for the probability that a given woman had access to a telephone recorded in the new variable CVQWEIGHT. This weight is normalized for the target population of each sample (note that two of the samples are not nationally representative):

    You'll find more detail about the construction of PMA COVID-19 survey weights here. For information about response rates for each sample, check out sample-specific Dataset Notes.

    To link female records between the baseline surveys and the COVID-19 surveys, drop household members who are not eligible women (use CONSENTFQ) and merge by FQINSTID. If you are using Stata, use the following code to trim trailing zeros and change the format of FQINSTID in order to merge:

    
    recast str50 fqinstid
    replace fqinstid = trim(fqinstid)
    
    

    Client Exit Interviews

    The client exit interview surveys were introduced in 2019 in select countries. Interviewers approached women leaving family planning clinics to ask about the services they received during their visit. Only women aged 15-49 were eligible to be interviewed, and are only interviewed once in a sample. Most questions on the survey were only given to women who had received either information on family planning or a method. Questions include whether the woman received her preferred method, whether advantages or disadvantages of certain methods were discussed, and how the visit experience was (length wait time, polite staff, etc).

    The sampling frame for client exit interviews begins with the random sampling of primary sampling units called EAs, which are small geographic areas of about 200-300 households. Households are sampled from these EAs for the household and female family planning surveys, and up to three public and three private health facilities in these areas are also visited for the service delivery point surveys. A subset of these facilities that provide family planning services were chosen if they reached a certain threshold of clients per day, and interviewers approached women who left the facility after an appointment to complete a client exit interview.

    The data in the client exit interviews can be attached to the service delivery point (SDP) data in the same year and country using the variable FACILITYID. For some countries, there is a follow up survey for the respondents within a 6 month period. These data will be available in 2022.

    How to get these data: On the data dissemination system, choose the Person - Client Exit Interview unit of analysis. The default case selection is for women aged 15-49. When selecting All Cases, you include observations of women who were not eligible for interview.