Brady T. West

Sampling Weights and Data Analysis for the Medical Monitoring Project Provider Survey

Research Project Description

Task 1.0. Contract Kick-off (mid- to late October)
Participate in the Project Kickoff conference call. In advance meet one or more times with the Altarum project team to coordinate call contents and review initial agenda. Present an overview of weighting design considerations. Lead any subsequent discussions about obtaining required inputs from CDC and exploring alternative approaches. Identify client responsibilities and action items pursuant to implementing other tasks.

Task 3.0. Weighting Analysis Plan (late October to late December)
Guide the creation of the Weighting Analysis Plan. The weighting analysis plan will provide a detailed work description of the steps we will take to adjust the collected survey results data for effects of the sample design and to identify and mitigate sources of non-response bias with appropriate post-field weight adjustments. The plan will also describe our approach to documenting methods and code to accomplish this task, as well as for transferring the task deliverables to CDC via approved secure methods. Create Plan outline, provide key content sections, and support Altarum staff in writing tasks.

Generate Adjusted Response Rate Calculations
Provide advice on most appropriate calculations(s) of response rates. Using disposition codes assigned during data collection, as well as original sample information, advise on alternative methods for calculating response rates. Advise on appropriate treatment of presumed ineligibles respondents among non-response group.

Generate Sampling Weights
Generate weights to accommodate clustered sample design. The sample design used in the 2013 MMP Provider Survey reflects the structure of the MMP program itself, with a hierarchy defined as Project Areas and facilities within Project Areas. Provide advice on construction of appropriate weights and instructions on how/where to employ them in subsequent analysis.
Conduct Non-response Analysis to Identify Characteristics Associated with Provider

Non-response
Provide assistance/consultation on non-response analysis techniques. With reference to the implemented weighting scheme and general study design, provide advice and consultation to Altarum team in conducting a non-response analysis.
Create Appropriate Strata and Cluster Variables
Review construction of strata and cluster variables. Review Altarum team processes to ensure that the calculation of the variance is correct given the complex survey design and that any subsequent analysis utilizes a consistent approach to applying sample weights and properly subsetting the results data.

Generate Analytic Variables as Specified by CDC
Review and advise on the construction of additional variables. Altarum will generate analytic variables as requested by the CDC. Typical variable construction could include scales or composites formed from several related survey items, simple recodes such as date of birth recoded into convenient age groups, or the transformation of ordered response sets into vectors of dummy variables.

Task 5.0. Data Analysis and Reporting (8 subtasks) (January to October/November 2014)
Advise and participate in analytic subtasks. Each of the eight analytical and reporting tasks has several related purposes including supporting briefings and presentations as well as contributing to formal reports and publications. The analysis and reporting path cannot be fully specified at the outset because early and intermediate results and evolving understanding will inform later steps. CDC researchers will provide the substantive insights to guide this process, and each analysis will require meetings and discussions along the trajectory to final results. Provide consultation on each of the analytic within a phase to review the applicability and appropriateness of the weighting design to the analytical goals of the subtask. Review and advise on specific analytic techniques utilized within each subtask.

Funding: Centers for Disease Control and Prevention (subcontract: Subcontract No SC-13-025)

Funding Period: 9/30/2013 to 12/31/2015

PSC In The News

RSS Feed icon

Raghunathan et al. found behavioral treatment reduces urinary incontinence

Sarah Miller quoted in The New York Times

More News

Highlights

Total Survey Error: A Framework for High Quality Survey Design - a workshop by Brady West and Paul Schulz, Oct 23

Kowalski elected to the Board of Directors of the American Society of Health Economists (ASHEcon)

More Highlights


Connect with PSC follow PSC on Twitter Like PSC on Facebook