Data Analysis Plan

Attachment F – Analysis Plan

Since the primary goal of the research is to examine the workings of novel survey operations, the data analysis will focus on how well systems and people worked to affect the study components. Much of this analysis will be descriptive.

1. To examine the implementation of the administrative records component, we will document the following details, consistent with maintaining respondent confidentiality:

• the number of sample units for which records are judged to be sufficient to enumerate households;

• some characteristics of those sample units (e.g., household size, type of structure, proximity to one another);

• the reciprocal number of sample units for which records are judged to be insufficient to enumerate households;

• some characteristics of those sample units (e.g. household size, type of structure, proximity to one another);

• the ability of systems to remove from the NRFU workload those sample units to be enumerated using records;

• the ability of systems to adjust case assignments for interviewers following the removal of records-enumerated cases;

• the ability of field supervisors to adjust staffing resources after the removal of records-enumerated cases from the NRFU workload;

• the ways in which interviewers adjust contact plans in response to removal of records-enumerated cases from the NRFU workload.

2. To examine the implementation of the fixed contact strategy, we will document the following details:

• the number of phone numbers attached to sample units after removal of nonworking landlines;

• the ability of interviewers to transmit contact and outcome information daily;

• the number, timing and outcome of contact attempts by CAPI interviewers using each number;

• the number, timing and outcome of in-person contact attempts

3. To examine the implementation of the adaptive contact strategy, we will document the following details:

• the number of phone numbers attached to sample units after removal of nonworking landlines;

• the number, timing and outcome of contact attempts by CATI interviewers using each number;

• the ability of models to generate daily propensity scores for cases to be contacted in person;

• the ability of systems to deliver daily case priority designations to interviewers;

• the ability of interviewers to contact cases as directed;

• the ability of interviewers to transmit contact and outcome information and retrieve new case priorities daily;

• the number, timing and outcome of in-person contact attempts.

4. To examine operational differences between contact strategies in different administrative records conditions, we will examine the aforementioned contact strategy details within each administrative records treatment.

Beyond the documentation of operational matters to satisfy the primary goal of the study, we will examine comparisons of outcomes for the different experimental groups. These analyses will focus on cost, effort and data quality measures.

1. Cost and effort will be operationalized in four ways:

• overall cost, which includes the costs of CAPI and CATI interviewers,

• average cost per case,

• average cost per contact attempt, and

• average case completion per contact attempt

2. Data quality will be operationalized in four ways:

• item nonresponse rates,,

• percent proxy,

• percent partial completes, and

• final response rates

Within the limits imposed by the sample size, we will also examine the correspondence between the counts of household members obtained from interviews with counts obtained from administrative records for cases where the NRFU workload is not reduced using administrative records.