:R195.1 :A doi: /j.issn

Transcription

1 1, 2,3* (1., ; 2., ; 3., * :,E -mail:lphu812@sina.com), ; ; ; ; :R195.1 :A doi: /j.issn , 2,3* (1.,,, , ; 2.,, , ; 3., , * :, - : 812@. ) Complex sampling is the sampling plan of other sampling methods or their combination, except a simple random sampling of one stage in the process of sampling.this paper presented a macro overview of the characteristics of complex sampling data, the main points of the difference analysis and multiple regression analysis based on the complex survey data, and the key points of multiple regression analysis of survey survival data.the paper could provide references for the researchers to better understand and implement the analysis of complex sampling data. Complex survey; Feature; Sampling weight; Analytical techniques of statistics ; Multiple regression analysis 1 1.1,, [1], :, ; ;, [2],,, [3], 1.2 : (2015AA020102) 410,,,,,, [4] [5], 95%,, ( ), ;,,,, 1.3 [4],,,,

2 ,, [4],,, [4],,, : = [4],,,, [4] 1.4,,,,,,,,, [6], SAS SUMMARY FREQ MEANS REG,SAS 9.0 SURVEYMEANS SURVEYFREQ SURVEYREG SURVEYLOGISTIC SURVEYPHREG [7] 1.5,,,,, [8] Taylor ( Taylor Series Linearization, TSL):, Taylor, [9-10],,,,,, :,, ( Balanced Repeated Replication, BRR ) ( Jackknife Repeated Replication, Jackknife ) Bootstrap Jackknife : k,,,, [11],,BRR,Jackknife, BRR : L,, 2L, 2L,, [8,12] BRR, (PSU) PSU Jackknife Bootstrap,BRR,, Bootstrap : ( ), Jackknife BRR SURVEYFREQ PROC SURVEYFREQ,,,PROC SURVEYFREQ, TSL BRR Jackknife SURVEYFREQ 411

3 (, ),PROC SURVEYFREQ Rao -Scott ;,PROC SURVEYFREQ Rao -Scott Rao -Scott Wald Wald PROC SURVEYFREQ : PROC SURVEYFREQ < options STRATA variables < /option TABLES requests < /options PROC SURVEYFREQ,, PROC SUR- VEYFREQ TABLES, STRATA CLUSTER WEIGHT REPWEIGHTS BRR Jackknife,BY BY SURVEYFREQ FREQ PROC SURVEYFREQ, VARMETHOD =TAYLOR, VARMETHOD =BRR, VARMETHOD =BRR ( fay = c) ( c ), VARMETHOD =JACK- KNIFE, CLUSTER REP- WEIGHTS STRATA 2.2 SURVEYMEANS SURVEYMEANS PROC SUR- VEYMEANS, PROC SURVEYMEANS TSL BRR, 412 PROC SURVEYMEANS : PROC SURVEYMEANS < options >< statistic -keywords CLASS variables ; DOMAIN variables < variable _ variable variable _variable _ variable... >< /option RATIO < label > variables /variables ; STRATA variables < /option VAR variables ; PROC SURVEYMEANS, VAR CLASS STRATA CLUSTER DOMAIN,RATIO,WEIGHT,REPWEIGHTS BRR Jackknife,BY BY SURVEYMEANS MEANS PROC SURVEYMEANS, VARMETHOD = BRR (fay =c)(c ), CLUSTER DOMAIN REPWEIGHTS STRATA SURVEYREG PROC SURVEYREG, - PROC SURVEYREG, -,PROC SURVEYREG

4 TSL BRR PROC SURVEYREG : PROC SURVEYREG < options CLASS variables ; CONTRAST label effect values <...effect values >< /options >; DOMAIN variables <variable_variable variable_variable_variable...>; EFFECT name = effect -type ( variables < /options > ) ; ESTIMATE < label > estimate -specification < /options LSMEANS < model -effects >< /options LSMESTIMATE model -effect lsmestimate -specification < /options >; MODEL dependent = < effects >< /options OUTPUT < keyword < =variable -name >...keyword < =variable -name >>< /option SLICE model -effect < /options STORE <OUT =>item -store -name < /LABEL = label STRATA variables < /options TEST < model -effects >< /options PROC SURVEYREG MODEL,, CLASS, CLASS MODEL, CONTRAST ESTIMATE, MODEL CONTRAST ESTIMATE CLASS CLUSTER CONTRAST EFFECT ESTI- MATE LSMEANS LSMESTIMATE REPWEIGHTS SLICE STRATA TEST, MOD- EL WEIGHT STORE OUTPUT CLASS,CLUSTER,DOMAIN, MODEL,REPWEIGHTS BRR Jackknife SURVEYREG REG PROC SURVEYREG, VARMETHOD = TAY- LOR, VARMETHOD =BRR, VARMETHOD =BRR (fay =c) ( c ), CLUSTER DOMAIN STRATA 3.2 SURVEYLOGISTIC SURVEYLOGISTIC Fisher Newton -Raphson,, ordinallogistic probit log -log logit,, TSL BRR PROC SURVEYLOGISTIC : PROC SURVEYLOGISTIC < options CLASS variable < ( v -options) >< variable < ( v -options) >... >< /v -options CONTRAST label effect values <,...effect values >< /options DOMAIN variables < variable _ variable variable _ variable _ variable... EFFECT name = effect -type ( variables < /options > ) ; ESTIMATE < label >estimate -specification < /options >; FREQ variable ; LSMEANS < model -effects >< /options LSMESTIMATE model -effect lsmestimate -specification < /options MODEL events /trials = < effects < /options > MODEL variable < ( v -options) > = < effects >< /options OUTPUT <OUT =SAS -data -set ><options >< /option >; SLICE model -effect < /options STORE <OUT =>item -store -name < /LABEL = label STRATA variables < /option < label: > TEST equation1 <,..., equationk >< /options UNITS independent1 = list1 <...independentk = listk >< / option CLASS CLUSTER CONTRAST EFFECT ESTIMATE LSMEANS LSMESTIMATE,REPWEIGHTS SLICE STRATE TEST, MODEL WEIGHT STORE OUTPUT UNITS, CLASS MODEL,CONTRAST MODEL BY,CLASS,CLUSTER, DOMAIN,MODEL 413

5 , REPWEIGHTS BRR Jackknife SURVEYLOGISTIC LOGISTIC SURVEYLOGISTIC, VARMETHOD BRR (fay =c)(c ), DOMAIN REPWEIGHTS 4 SURVEYPHREG Cox,Cox,, SURVEYPHREG,, TSL BRR PROC SURVEYPHREG : PROC SURVEYPHREG < options CLASS variable < ( options) ><...variable < ( options) >>< /options DOMAIN variables < variable _ variable variable _ variable _ variable... ESTIMATE < label > estimate -specification < /options FREQ variable ; LSMEANS < model -effects >< /options LSMESTIMATE model -effect lsmestimate -specification < /options > ; MODEL response < *censor( list) > = effects < /options NLOPTIONS < options OUTPUT <OUT =SAS -data -set ><keyword =name...keyword =name >< /options SLICE model -effect < /options STRATA variables < /option STORE <OUT =>item -store -name < /LABEL = label TEST < model -effects >< /options PROC SURVEYPHREG MODEL, CLASS MODEL, = MODEL,CLASS,STRATA,CLUSTER,WEIGHT,NLOPTIONS, REPWEIGHTS BRR Jackknife,DOMAIN,BY SURVEYPHREG PHREG PROC SURVEYPHREG, VARMETHOD = BRR (fay =c)( c ), DOMAIN REWEIGHTS NLOPTIONS [1],,,. [ J].,2015,32(4) : , 726. [2] Osborne JW.Best practices in using large, complex samples: the importance of using appropriate weights and design effect compensation [ J ]. Practical Assessment, Research and Evaluation, 2011, 16(12) :1-7. [3] Anderson KM, Wilson PW, Odell PM, et al.an updated coronary risk profile.a statement for health professionals[ J].Circulation, 1991, 83(1) : [4],. [ J]., 2015, 39(5) : [5] Sharon L.Sampling: Design and Analysis[ M].Boston: Thomson Brooks Cole, 2009: [6] SAS Institute Inc.SAS /STAT 9.3 User s Guide [ M].Cary, NC: SAS Institute Inc, 2011: [7],. logistic [ J]., 2008, 25(6) : [8],,. [J]., 2012, 29(5) : [9],. [ J]., 2008, 25(4) : [10] West BT.Statistical and methodological issues in the analysis of complex sample survey data: practical guidance for trauma researchers[ J].J Trauma Stress, 2008, 21(5) : [11] KrewskiD, Rao JNK.Inference from stratified samples: properties of the linearization, jackknife and balanced repeated replication methods[ J].Ann Stat, 1981, 9(5) : [12]. [ J]., 2011, 28(2) : ( : ) ( : ) 414