Question

我目前正在r中运行lmer（）以获得具有以下结构的数据集：

avgcobint=lmer(scale.logcptplus1 ~ scale.logdepth + scale.avgcobb + scale.logdepth:scale.avgcobb + (1|location) + (1|Fyear), data=cpt, REML=TRUE)

并收到以下有关固定效果的摘要：

Fixed effects:
                             Estimate Std. Error t value
(Intercept)                  -0.04868    0.15425  -0.316
scale.logdepth               -0.32668    0.06168  -5.297
scale.avgcobb                 0.11014    0.10478   1.051
scale.logdepth:scale.avgcobb  0.11689    0.06086   1.921

然后我用日志和缩放变量预测我的实际值估计值，用这个等式表示中值深度（14）...这是下图中正确的实际值线：

resultmeddepth14= -0.04868 + exp((-0.32668) * ((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))) + (0.11014 * ((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm))) + (0.11689 *(exp(((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))))*((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm)))

lmer real value estimates

但是当我使用这些等式添加+/-一个标准误差线时：

MINUS SE

resultmeddepthminus= (-0.04868-0.15425) + exp((-0.32668-0.06168) * ((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))) + ((0.11014-0.10478) * ((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm))) + ((0.11689-0.06086) *(exp(((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))))*((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm)))

PLUS SE

resultmeddepthplus= (-0.04868+0.15425) + exp((-0.32668+0.06168) * ((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))) + ((0.11014+0.10478) * ((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm))) + ((0.11689+0.06086) *(exp(((depth_median-mean(cpt$logdepth))/sd(cpt$logdepth))))*((cobb_range - mean(cpt$avgsummercobbleareakm))/sd(cpt$avgsummercobbleareakm)))

我有这张图： lmer real estimates +/- one SE

并且所有线的交点都是x值0.068341709

为什么SE线相交？这里发生了什么，我该如何解决？

标准误差线相交

0 个答案: