期末作業(yè)匯報(bào)一、兩獨(dú)立樣本：1、獨(dú)立性：滿足2、方差齊性檢驗(yàn)：?sdtesty,by(group)VarianceratiotestGroup|LObsMeanStd.Err.Std.Dev.[95%Conf.Interval]十0129.0225572.56525783.044006—1?1353211.1804351|L162.415993?76094263.04377.79408214.037904十combined|45.8735565?48067863.22449—.09518761.842301Ho:sd(0)=sd(1)TOC\o"1-5"\h\zF(28,15)observed=F_obs=1.000F(28,15)lowertail=F_L=1/F_obs=1.000F(28,15)uppertail=FU=Fobs=1.000Ha:sd(0)<sd(1)Ha:sd(0)~=sd(1)Ha:sd(0)>sd(1)P<F_obs=0.4818P<F_L+P>F_U=0.9997P>F_obs=0.5182

由軟件結(jié)果可知，P=0.9997>0.10，不能拒絕H/故可以認(rèn)為方差齊性。3、正態(tài)性檢驗(yàn)：.sktestyifgroup==0Skewness/KurtosistestsforNormalityjointVariable|Pr(Skewness)Pr(Kurtosis)adjchi2(2)Prob>chi21y|0.4530.8170.650.7241?sktestyifgroup==1joint

Variable|Pr(Skewness)Pr(Kurtosis)adjchi2(2)Prob>chi2+y10.0760.3974.150.1257由軟件結(jié)果可知，兩組的P值都大于0?05,故不能拒絕H0,可認(rèn)為兩組資料服從正態(tài)分布。4、t檢驗(yàn)：由于資料滿足獨(dú)立性、方差齊性和正態(tài)性三個(gè)條件，故選擇t檢驗(yàn)。ttesty,by(group)Two-samplettestwithequalvariancesGroup|LObsMeanStd.Err.Std.Dev.[95%Conf.Interval]十0129.0225572.56525783.044006-1?1353211.1804351|L162.415993?76094263.04377?79408214.037904十combined|L45.8735565?48067863.22449-.09518761.842301十diff|-2.393436?9479405—4.30514-.4817315Degreesoffreedom:43Ho:mean(0)-mean(1)=diff=0Ha:diff<0Ha:diff~=0Ha:diff>0t=一2.5249t=一2.5249t=一2?5249P<t=0.0077P>|t|=0.0153P>t=0.9923由軟件結(jié)果可知，P=0.0153<0.05,拒絕H0,接受H1,即兩種藥物治療慢性腎炎的療效不同；由于0組的樣本均數(shù)(0?0225572)小于1組的樣本均數(shù)(2.415993),故可推斷B藥的效果更好。5、秩和檢驗(yàn)：?ranksumy,by(group)Two-sampleWilcoxonrank-sum(Mann-Whitney)testgroup|obsranksumexpectedT-01295556671|L16480368十combined|45103510351778.670.001778.670.00adjustedvarianceadjustedvariance1778.67Ho:y(group==0)=y(group==1)z=一2?656Prob>|z|=0.0079二、三個(gè)獨(dú)立樣本：1、獨(dú)立性：滿足2、方差齊性檢驗(yàn):ygroup;predicte,residual;genee=abs(e);onewayeegroup:onewayeegroupSourceSSdfMSFProb>FBetweengroups.0322453732.0161226870.030.9690Withingroups30.224460959.512278998Total30.256706261.496011578Bartlett'stestforequalvariances:chi2(2)=0.9884Prob>chi2=0.610由軟件結(jié)果可知，P>0.10,方差齊性，下面進(jìn)行正態(tài)性檢驗(yàn)。3、正態(tài)性檢驗(yàn)：swilkyifgroup==1VariableShapiro-WilkWtestfornormaldata|ObsLWVzProb>zy十|220.925081.8981.2990.09691swilkyifgroup==2Variable|ObsWVzProb>zy11170.964780.744-0.5890.72222swilkyifgroup==3Variable|ObsWVzProb>zchi-squaredwithties=chi-squaredwithties=33.000with1d.f.y|230.899432.6311.9670.02460由軟件結(jié)果可知，第3組不滿足正態(tài)性的要求，故選用K-W檢驗(yàn)。4、K-W檢驗(yàn)：chi-squared=probability=?kwallisy,by(group)chi-squared=probability=Test:Equalityofpopulations(Kruskal-Wallistest)group_Obs_RankSum122276.00217506.003231171.0051.070with2d.f.0.0001chi-squaredwithties=51.070bability=0.0001由軟件結(jié)果可知，P<0?05,故三種藥治療慢性腎炎的有效率不全相同。5、兩兩比較：采用Bonferroni法，校正后的a=0?05/3=0.01667chi-squared=probability=kwallisyifgroup==1|group==2,by(group)chi-squared=probability=Test:Equalityofpopulations(Kruskal-Wallistest)group_Obs_RankSum122276.00217504.0021.574with1d.f.0.0001hi-squaredwithties=21.574with1d.f.robability=0.0001ifgroup==1group==3,by(group)ifgroup==1group==3,by(group)group_Obs_RankSumgroup_Obs_RankSum22253.00323782.00chi-squared=probability=33.000with1d.f.0.0001probability=0.0001ifgroup==2probability=0.0001ifgroup==2group==3,by(group)group_0bs_RankSum17155.00chi-squared=probability=23665.00chi-squared=probability=28.027with1d.f.0.0001chi-squaredwithties=28.027bability=0.0001由于每?jī)山M之間比較的P值都小于a，故可推斷三種藥治療慢性腎炎的有效率都不相同。6、方差分析：?onewayygroupSourceAnalysisofVarianceSSdfMSFProb>FBetweengroups383.7919092191.895955148.750.0000Withingroups76.1134674591.29005877Total459.905377617?5394324Bartlett'stestforequalvariances:chi2(2)=0.3200Prob>chi2=0.8527、兩兩比較：?lsdygroupLSD方法進(jìn)行兩兩均數(shù)比較

輸入效應(yīng)變量分組變量

mean1-mean2=-2.742864P-value=0.000000mean1-mean3=一5?836804P-value=0.000000mean2-mean3=-3.093940P-value=0.000000?onewayygroup,bSourceAnalysisofVarianceSSdfMSFProb>FBetweengroups383.7919092191.895955148.750.0000Withingroups76.1134674591.29005877Total459.905377617?5394324Bartlett'stestforequalvariances:chi2(2)=0.3200Prob>chi2=0.852Comparisonofybygroup(Bonferroni)RowMean-|ColMean|412+2|112.742860.0003|15.83680.0003.093940.000三、'配對(duì)資料1、獨(dú)立性：數(shù)據(jù)獨(dú)立，先考慮配對(duì)t檢驗(yàn)2、正態(tài)性檢驗(yàn)：gend=x1-x2swilkdShapiro-WilkWtestfornormaldataVariable|ObsWVzProb>z+d|320.983270.558-1?2110.88707由軟件結(jié)果可知，差值服從正態(tài)分布，選用配對(duì)t檢驗(yàn)。3、配對(duì)t檢驗(yàn)：ttestd=0Variable|ObsMeanStd.Err.Std.Dev.[95%Conf.Interval]L十d|32-.3875.75401814.265371-1.925331.15033Degreesoffreedom:31Ho:mean(d)=0Ha:mean<0Ha:mean~=0Ha:mean>0

t=—0.5139t=—0.5139t=—0.5139P<t=0.3055P>|t|=0.6110P>t=0.6945由軟件結(jié)果可知，P值＞0?05,不能拒絕H0,故不能認(rèn)為這2種藥物治療高血脂的療效不同。4、符號(hào)秩和檢驗(yàn)：?signrankxl=x2Wilcoxonsigned-ranktestsign|obssumranksexpected+一positive|14236263?5negative|17291263.5zero|L111十a(chǎn)ll|32528528unadjustedvariance2860.00adjustmentforties—0.88adjustmentforzeros—0.25adjustedvariance2858.88Ho:x1=x2z=-0.514Prob>|z|=0.6070本題可直接采用秩和檢驗(yàn)：ranksumy,by(group)group|obsranksumexpected十0I40138913601I27889918十combined|6722782278unadjustedvariance6120.00adjustmentforties—2070.00adjustedvariance4050.00Ho:y(group==0)=y(group==1)z=0.456Prob>|z|=0.6486

由軟件結(jié)果可知，P值＞0.05，故不能認(rèn)為兩種藥物的療效有差別。五、本題可直接采用符號(hào)秩和檢驗(yàn):?signrankyl=y2sign|Lobssumranksexpected十positive|7689.51576negative|252462.51576zero|L8234033403十a(chǎn)ll|11465556555unadjustedvariance125091.25adjustmentforties-682.00adjustmentforzeros-46791.25adjustedvariance77618.00Ho:y1=y2z=—3.182Prob>|z|=0.0015由軟件結(jié)果可知，P值<0.05,故兩種藥物的療效不同。六、隨機(jī)區(qū)組1、數(shù)據(jù)轉(zhuǎn)換：stackx1blockx2blockx3block,into(yblock)2、方差齊性檢驗(yàn)：anovaygroupblock;predicte,residual;genee=abs(e);anovaeegroupblock:?anovaeegroupblockNumberofobs=99R—squared=0.4773RootMSE=1.04435AdjR—squared=0.1996Total|133.547507981.36272967由軟件結(jié)果可知，P<0.05,故方差不齊，采用Friedman檢驗(yàn)。3、Friedman檢驗(yàn)：friedmanblb2b3b4b5b6b7b8b9bl0bllbl2bl3bl4bl5bl6bl7bl8bl9b20b21b22b23b24b25b26b27b28b29b30b31b32b33Friedman=23.8636Kendall=0.3616p-value=0.0000由軟件結(jié)果可知，P<0.05,故這3種藥物治療該疾病的療效不同。4、方差分析：?anovaygroupblockNumberofobs=99R-squared=0.6244RootMSE=2.31064AdjR-squared=0.4248Source|LPartialSSdfMSFProb>F十Model|567.9478983416.70434993?130.0000group|171.420602285.710301116.050.0000block|396.5272953212.3914782.320.00211Residual|L341.699399645.33905311十Total|909.647296989.28211527七、相關(guān)性檢驗(yàn)：?pwcorrxy,sig1Lxy十x|||1.0000y10.66341.000010.0000由軟件結(jié)果可知，r=0.6634,P=0?0000<0.05,故pH0,因此，兩個(gè)指標(biāo)之間存在相關(guān)性。八、回歸分析:

?regyxSource|SSdfMSNumberofobs=17(1aa\—46十F(1,44)—?Model|93832.6765193832.6765Prob>F—0.0000Residual|28.477749344?647221576R-squared—0.9997AQQQ7十Adjrsquared—0.9997Total|93861.1542452085.80343RootMSE—.8045y|Coef?Std.Err.tP>|t|[95%Conf.Interval]+x|4.961854.0130315380.760.0004.935594.988117_cons|2.488645.226470910.990?0002.0322222.945067由軟件結(jié)果可知，y=2.49+4.96x。P值小于0.05,故該回歸方程成立。九、本題釆用Logrank檢驗(yàn)：?ststestgroupfailure_d:deadanalysistime_t:timeLog-ranktestforequalityofsurvivorfunctions|EventsEventsgroup|observedexpected10|3120.27

111727.731Total|4848.00chi2(1)=10.84Pr>chi2=0.0010由軟件結(jié)果可知，兩種藥物的療效不同。十、樣本率與已知總體率的比較:?prtesti5060.15,countOne-sampletestofproportionx:Numberofobs=50Variable|LMeanStd.Err.zP>|z|[95%Conf.Interval]十X1.12?04595652.611160.0090.0299269.2100731Ho:proportion(x)=