CROSSTAB Example #8 RTI International （交叉表例8 RTI国际）.pdf

CROSSTAB Example #8 SUDAAN Statements and Results Illustrated  Stratum-specific Chi-square (CHISQ) Test  Stratum-adjusted Cochran-Mantel-Haenszel (CMH) Test  ANOVA-type (ACMH) Test  ALL Test option  DISPLAY option Input Data Set(s): NHANES3S3.SAS7bdat Example This example illustrates the variety of hypotheses and test statistics now available on the TEST statement in CROSSTAB. Solution The data set consists of adults aged 17 and older from NHANES III , a cross-sectional sample survey of the civilian, non-institutionalized population aged 2 months or older, fielded during 1988-1994. All variables in this example are from the home interview component of NHANES III, and all six years of data are analyzed. Thus, the sample weight variable is WTPFQX6, and the stratification and PSU variables are SDPSTRA6 and SDPPSU6, respectively. The SAS-Callable SUDAAN code used for this example is displayed in Exhibit 1. This example uses the NHANES III dataset to evaluate whether having been diagnosed with arthritis (HAC1A, 1=yes, 2=no) is significantly associated with the self-reported condition of someone’s natural teeth (TEETHSTAT, collapsed from HAQ1 as excellent/very good, good, fair, or poor/no natural teeth). We also want to evaluate this hypothesis after adjusting for age group (17-49 vs. 50+) and self-reported general health status (HLTHSTAT, collapsed from HAB1 as excellent/very good, good, or fair/poor). In the SAS-Callable SUDAAN code below (Exhibit 1), the following statements are used: ■ TABLES statement requests: 1) a single 2-way table of arthritis-by-teeth status, and 2) a stratified (by age and health status) two-way table of arthritis-by-teeth status. All of these variables are listed on the CLASS statement. ■ TEST statement requests the stratum-specific CHISQ test of general association for each table request—based on (observed – expected), analogous to