Joint policy on variance estimation and statistical reporting standards on NHANES III and CSFII reports: HNIS/NCHS Analytic Working Group recommendations

HERO ID

1005567

Reference Type

Technical Report

Year

1993

Language

English

HERO ID 1005567
Year 1993
Title Joint policy on variance estimation and statistical reporting standards on NHANES III and CSFII reports: HNIS/NCHS Analytic Working Group recommendations
Book Title Analytic and reporting guidelines: The third National Health and Nutrition Examination Survey, NHANES III (1988-94)
Authoring Organization National Center for Health Statistics
Publisher Text Centers for Disease Control and Prevention
City Hyattsville, MD
Page Numbers 39-45
Abstract Below is a summary of the recommendations reached by the Methodological Subcommittee of the HNIS/NCHS Analytic Working Group on the issues of variance estimation and statistical reporting standards. Specific recommendations are underlined, whereas suggested practices are italicized. The implementation of these recommendations and suggestions will vary from survey to survey and, perhaps, from estimate to estimate. Nevertheless, official agency publications should contain a "statistical notes" section describing the variance estimation and statistical reporting standards used therein. The design-based approach to the estimation and analysis of survey data is assumed here. Unlike model-dependent alternatives, the design-based approach makes few assumptions about the nature of the data being summarized and/or analyzed. Two aspects of the sampling design must be taken into account when using this approach: the sample weights and the complex sample design (stratified, multi-stage sampling). Weights are used in the this approach when estimating mean, medians, and other descriptive statistics as well as analytical statistics like regression coefficients. Both weights and indicators of stratum and primary sampling unit (PSU) membership are used when estimating variances and testing for statistical significance. In general, using statistical weights that reflect the probability of selection and propensity of response for sampled individuals will affect parameter estimates, while incorporating the attributes of the complex sample design (i.e., clustering and stratification) will affect estimated standard errors and thereby test statistics and confidence intervals. The recommendations for presentation of statistical data that follow arise from the issue of sampling variability, and reflect the random way (in the rigorous statistical sense) in which the sample was selected. Although beyond the scope of this report, a consideration of nonsampling issues such as measurement error, nonresponse bias, and other methodological biases are necessary for any thorough interpretation and evaluation of the validity of survey findings.
Url https://wwwn.cdc.gov/nchs/data/nhanes3/manuals/nh3gui.pdf
Is Certified Translation No
Dupe Override No
Comments Available from: Agricultural Research Service, Survey Systems/Food Consumption Laboratory, 4700 River Road, Unit 83, Riverdale, MD
Is Public Yes
Language Text English
Relationship(s)