SkipperLin · 2020年03月04日
问题如下：
Suppose all residual autocorrelations are , where T is the sample size. Would these violate the confidence bands in an ACF plot?
选项：
解释：
No, the confidence bands are and so these would not. If many autocorrelations are jointly relatively large, then this series is likely autocorrelated. A Ljung-Box test would likely detect the joint autocorrelation.
请问一下confidence bands的公式在笔记上哪里呢
NO.PZ2020011101000015问题如下 Suppose all resiautocorrelations are 1.5/T1.5/ \sqrt T1.5/T, where T is the sample size. Woulthese violate the confinban in Aplot? No, the confinban are ±1.96/T\pm 1.96/ \sqrt T±1.96/T anso these woulnot. If many autocorrelations are jointly relatively large, then this series is likely autocorrelate A Ljung-Box test woullikely tethe joint autocorrelation.“resiautocorrelations are（1.5/跟号T）”这句话是说e1和e2，e2和e3等等，各个e之间的correlation结果是（1.5/跟号T）？而不是 H0:p1=p2=…pn=0，Ha至少一个p不等于0的假设检验结果是（1.5/跟号T）？
NO.PZ2020011101000015 问题如下 Suppose all resiautocorrelations are 1.5/T1.5/ \sqrt T1.5/T, where T is the sample size. Woulthese violate the confinban in Aplot? No, the confinban are ±1.96/T\pm 1.96/ \sqrt T±1.96/T anso these woulnot. If many autocorrelations are jointly relatively large, then this series is likely autocorrelate A Ljung-Box test woullikely tethe joint autocorrelation. 印象中好像有个公式方差/根号下n，计算样本标准误的吧。这里的1.5/ 根号下T、1.96根号下T，是什么意思？
这道题目能翻译下吗？要考察的点是什么啊？已经看了之前的问题和回答，还是不是很清楚题目在问啥，谢谢。
回答思路不是很懂，这里为什么要用t test呢？因为题干里提到all resiautocorrelation，为什么不用F test呢？
