Definitions of the Target Population and Study Eligibility
In general, the target population for the NCD Mobile Phone Survey should include all eligible individuals, 18 years and older, with MPNs within the country’s MPN series (refer to the Mobile Phone Technology Manual for definition). To propose an exclusion, a country should explicitly define which MPN ranges are included in and excluded from the target population and provide an estimate of the percent of the population that resides in those ranges being excluded. This percentage will provide an indication of the potential bias effect of any real exclusions on estimates that are otherwise intended to be national in scope.
The target population for the NCD Mobile Phone Survey should include men and women who are 18 years of age and older. This definition is important when completing the demographic questions, which determines who is eligible to complete the NCD questions.
In general, a sampling frame is a list of all eligible members of the target population from which the sample could be drawn. For the NCD Mobile Phone Survey, the type of list depends on the study design:
Option 1. Two-phase sampling of mobile phone users from an implicit frame of MPNs, with stratified sampling disproportionate to the mobile phone user population in the second phase.
- An implicit frame is an implied sampling frame produced from the range of all possible MPNs in the country, some of which will not currently be in service. This frame will be constructed in collaboration with the country’s telecommunication regulatory agency.
- For this study design, a comprehensive list of all survey-eligible individuals in a country likely does not exist, making sampling from the preferred frame virtually impossible. Consequently, an indirect list frame should be used to select the sample. This list is implied by the known set of possible MPNs that could be assigned by providers to their customers.
- This option is preferred whenever a reputable national listing of MPNs currently in service is unavailable.
Option 2. Two-phase sampling of mobile phone users from an explicit frame of MPNs, with stratified sampling disproportionate to the MP user population in the second phase.
- An explicit frame is a list of MPNs provided by the telecommunications operators. This frame may or may not be a complete list of MPNs subscribing to the telecommunications operator’s service.
- From this frame, the first-phase sample of MPNs is selected from a provider-produced list of current MP subscribers.
Option 3. Split-sample household-level integration with the STEPS stratified multi-stage cluster sample.
- A cluster frame is a list of entities that provide indirect links to multiple individuals in the target population by grouping them somehow (e.g., a list of a city’s schools, each being a grouping of students to be sampled, or a list of the room’s file drawers, each being a grouping of files to be sampled). This method is called a cluster sample because a sample of clusters should be selected first, and then the sample of individuals is identified from the individuals linked to the selected clusters.
- For a planned STEPS household survey, a cluster sample of geographic areas within the country is randomly selected, a complete list of all households within each selected area is constructed, a sample of households is randomly selected from each list, and one individual is randomly selected for interview from a randomly designated subset of the selected households in each sample area. This type of multi-stage cluster design assumes that each selected survey-eligible individual can be linked to a household in the country.
- Mobile phone users in selected STEPS households that are not earmarked for data collection in STEPs are recruited for participation in the mobile phone survey. If recruited participants do not have a mobile phone, they will be provided one, along with instructions on how to use them for the survey.
- This sample design option is dependent on a household survey being conducted and assumes that respondent provides a mobile phone number and that the provided mobile phone number does not change from the time it was collected to the time the mobile phone survey was initiated. Any biases, measurable or unmeasurable, in the design of the household survey would be reflected in the mobile phone survey estimates.
The rest of this document describes methods for implementing options 1 or 2. Instructions for option 3 will be provided after data collection in the initial countries is complete.