SPSS操作实例-重复测量

1重复测方差分析实例操作分析过程1．数据格式2．软件实验步骤3．结果解释与描述原始数据groupt0t1t2t3groupt0t1t2t311.171.140.891.7701.350.740.620.6110.660.580.750.6801.661.391.10.7510.630.670.690.9700.710.630.490.5211.82.22.132.0102.011.31.090.8910.890.870.820.9100.990.790.660.710.760.630.590.8601.710.90.710.6611.321.021.451.3900.670.610.60.411.570.620.590.700.780.630.630.4511.740.570.530.9200.930.840.70.7110.71.261.761.9601.451.180.970.8610.640.770.620.7902.090.541.10.9311.091.481.190.5200.770.520.730.6810.881.421.291.3801.840.541.270.5210.750.770.91.101.330.961.121.2911.580.890.410.7400.80.670.60.6310.930.950.991.2101.041.030.80.8910.660.570.81.1200.720.550.440.4911.110.970.941.3200.880.750.710.6700.891.231.281.3201.10.590.560.600.450.30.30.301.010.780.660.6601.671.040.660.5801.430.920.860.8300.960.780.580.5400.810.910.830.7100.810.760.620.5100.880.680.490.5300.750.630.540.5701.130.490.450.5802.020.641.161.4200.950.70.850.8801.781.180.730.8601.320.780.570.5501.791.240.780.7501.031.090.970.9502.030.431.220.7101.271.120.90.6700.650.580.540.4900.840.60.420.5501.0610.720.7301.270.890.660.721.1数据格式1.2软件实验步骤些处的描述过程输出无标准差，group=2时可用Analyze\Explorer过程实现描述，group=3时可用Analyze→GeneralLinealModel→Multivariate去实现描述。1.3结果解释与描述表1有无合并症患者LC含量（sx）例数重复测量时间麻醉前麻醉后20分钟电切手术30分钟手术结束时无合并症421.181±0.4470.808±0.2600.755±0.2490.706±0.233有合并症181.049±0.3980.966±0.4190.963±0.4511.131±0.438统计描述可以通过Analyze\Explorer过程实现，该过程较简单不赘述。SPSS.11_0统计分析教程.（高级篇）张文彤P37也就是说，在分析时，我们首先要判断，重复测量的不同时间点之间的结果是否存在相关性，也就是进行球形检验，即Mauchly'sTestofSphericity。如果P0.05，不符合Huynh-Feldt条件，说明重复测量数据之间存在相关性，不可按单因素方差分析方法处理，需要进行多变量方差分析。以多元检验结果为准。SPSS.11_0统计分析教程.（高级篇）张文彤P41如果P0.05，符合Huynh-Feldt条件，说明重复测量数据之间不存在相关性，可按单因素方差分析方法处理。SPSS.11_0统计分析教程.（高级篇）张文彤P40Table2Mauchly'sTestofSphericityWithinSubjectsEffectMauchly'sWApprox.Chi-SquaredfSig.EpsilonaGreenhouse-GeisserHuynh-FeldtLower-boundTime0.48041.5945.0000.6800.7170.333本例由Table2Mauchly'sTestofSphericity可知，P0.001，不符合Huynh-Feldt条件，说明重复测量数据之间存在相关性，不可按单因素方差分析方法处理，需要进行多变量方差分析，以多元检验结果为准。Table3MultivariateTestsEffectValueFHypothesisdfErrordfSig.TimePillai'sTrace.2716.952a3.00056.000.000Wilks'Lambda.7296.952a3.00056.000.000Hotelling'sTrace.3726.952a3.00056.000.000Roy'sLargestRoot.3726.952a3.00056.000.000Time*groupPillai'sTrace.2927.681a3.00056.000.000Wilks'Lambda.7087.681a3.00056.000.000Hotelling'sTrace.4117.681a3.00056.000.000Roy'sLargestRoot.4117.681a3.00056.000.000多变量方差分析结果如Table3MultivariateTests所示，四种检验结果Time和Time*group的P值均0.05，说明时间因素以及时间因素和分组的交互作用有统计学意义，即测量指标有随时间变化的趋势并且时间因素的作用随着分组的不同而不同。Table4TestsofBetween-SubjectsEffectsSourceTypeIIISumofSquaresdfMeanSquareFSig.Intercept179.9661179.966617.964.000group1.36411.3644.684.035Error16.89158.291表4组间效应的检验。F=4.684，P=0.035，说明不同合并症患者的LC含量有差别。综上，也就是说患者LC含量随手术时间的变化趋势不同，且这种变化趋势随着合并症（有/无）分组情况不同而不同。如下图所示。