Test statistics from a time series sample against more recent data
Budget: $30 – $250 CAD
A paper has an hypothesis that a time-series population "variable" X is represented by a constant. The estimate for the constant is the mean of a sample X (38 data points). Support for the hypothesis is that s.d. of the sample is low.
So the constant is supposed to hold out of sample, within some confidence interval.
The last few data points in the sample are all outside of the the s.d. for the mean and are rising.
New data arrives (3 or 4 years).
That data continues to rise and is (obviously) also outside the s.d. of the mean of the initial sample. The latest 3 data points appear to be >2, >5 and >4 s.d. above the sample mean.
Is the hypothesis still viable? Can we reject? Etc.
So the constant is supposed to hold out of sample, within some confidence interval.
The last few data points in the sample are all outside of the the s.d. for the mean and are rising.
New data arrives (3 or 4 years).
That data continues to rise and is (obviously) also outside the s.d. of the mean of the initial sample. The latest 3 data points appear to be >2, >5 and >4 s.d. above the sample mean.
Is the hypothesis still viable? Can we reject? Etc.