SOLICITATION NOTICE
R -- ORM-Propensity
- Notice Date
- 8/15/2006
- Notice Type
- Solicitation Notice
- NAICS
- 541690
— Other Scientific and Technical Consulting Services
- Contracting Office
- Department of Health and Human Services, Center for Disease Control and Prevention, Procurement and Grants Office (Atlanta), 2920 Brandywine Road, Room 3000, Atlanta, GA, 30341-4146
- ZIP Code
- 30341-4146
- Solicitation Number
- Reference-Number-00HCPC13-2006-35340
- Response Due
- 8/30/2006
- Archive Date
- 9/14/2006
- Description
- The United States Government intends to procure technical expertise for the development of propensity score models. STATEMENT OF WORK Propensity Score Method for Nonignorable Nonresponse Adjustment for health survey data. BACKGROUND INFORMATION: NCHS confronts an increasing problem of non-response, partial or unit, especially for National Health and Nutrition Examination Surveys. Consequently missing information for survey data causes problems in contingency table or other data analysis. When the non-responses are estimated by modeling and incomplete counts are distributed to proper cells, the analysis with distributed data will be much more improved than the analysis only with observed data. It is important to develop proper models for non-response estimation and there are needs for more research on how to distribute the counts of unknown cells properly. Past research in this area addressed the Hierarchical Bayesian model for the non-ignorable non-responses in the small area problems. However it is complex and difficult to use. We now propose propensity score method for non-ignorable nonresponses and it differs from the usual Hierarchical Bayesian method. New method is used to allocate respondents and nonrespondents to adjustment cells using covariates related to nonrespondents; uncertainty of these adjustment cells are also taken into consideration. The analysis of the response variables is distribution-free (i.e., does not depend on model). This method will be applied to NCHS data systems to minimize nonresponse bias. OBJECTIVE: The objective is to provide the technical expertise for developing propensity score models to suggest ways to solve non-response in small area problems for NCHS data. Development of computing procedures for the use of the Markov chain Monte Carlo simulation and analysis applications at NCHS are required. SCOPE OF WORK: The vendor will provide all labor, materials, and equipment required to complete the task outlined below. The vendor may require travel or communications with CDC to obtain data needed to illustrate the methodology. TASKS TO BE PERFORMED: The vendor shall perform the following tasks in order to achieve the objective; the following activities will be carried out under the contract. 1. Provide technical expertise in covariates selection for the non-response or small area estimation and appropriate propensity models should be most efficient. 2. Model selection procedure will be carried out to select the better fitting covariates among a set of competing ones. This will involve simulation and variance measures. 3. On the selected covariates, residual analysis will be done to assess the goodness of fit using cross validation. This is both a mathematical and computation task for the model investigation since the components (i.e., conditional means and variances) do not exist in a closed form. 4. Investigate robustness to the assumptions associated with the propensity score models. 5. Perform simulation studies to assess the accuracy of the selected propensity model using quantities like mean squared error. 6. Use an appropriate variable (e.g., body mass index, bone mineral density, activity limitation status) from the NCHS survey data to demonstrate the procedures. 7. Outline a set of criteria for the comparison of alternative methods 8. Make recommendations for further research regarding analysis of NCHS data with non-ignorable non-responses. GOVERNMENT FURNISHED MATERIALS: N/A PERIOD OF PERFORMANCE: The period of performance shall commence upon award of purchase order and shall continue through the next 6 months and completed no later than March 31, 2007. DELIVERABLES The vendor shall provide the following deliverables to the Project Officer as a result of this contract: 1. The preliminary plan of general propensity model for analyzing data with non-ignorable non-response data (Task items 1-6) by 3 months after the award of purchase order, but no later than November 30, 2006 2. The final report that includes a detailed outline of criteria for comparing and evaluating alternative methods. Recommendation of further research on these topics. Complete a report or draft of a paper with the result of this research by the 6 month after the contract awarded (Task items 7-8), but no later than March 20, 2007. The Government intends to evaluate offers and award a contract without discussions with Offerors providing the requested service. Therefore, the initial offer should contain the offeror?s best terms. The quotation shall include the price, prompt payment terms, the F.O.B. Point(Destination or Origin), Taxpayer Identification Number(TIN), Dun & Bradstreet Number(DUNS),and descriptive literature of the service quoted. The quotation must reference Solicitation number 00HCPC13-2006-35340. Quotations will be due fifteen(15) calendar days from the publication date of this synopsis All responsible sources may submit a quotation, which if timely received, shall be considered by the agency. Quotations may be faxed to (770)488-2670/2671, Attn. Ivory Reid or emailed to ireid@cdc.gov All vendors must be registered in the Central Contracting Registry(CCR) prior to award of a Federal contract. CCR registration can be completed online at http://www.ccr.gov.
- Place of Performance
- Address: 3311 Toledo Road, Room 1217, Hyattsville, MD
- Zip Code: 20782
- Zip Code: 20782
- Record
- SN01115115-W 20060817/060815220321 (fbodaily.com)
- Source
-
FedBizOpps Link to This Notice
(may not be valid after Archive Date)
| FSG Index | This Issue's Index | Today's FBO Daily Index Page |