0

I wonder if there is a test for comparing means of two different time series (both measured at the same person or two different persons)? Would be very glad if someone can help me!

eski
  • 121
  • 2
  • 10
  • I believe your question is similar to this one: http://stats.stackexchange.com/questions/19103/how-to-statistically-compare-two-time-series – Sam Felix Feb 10 '14 at 15:03

1 Answers1

0

It would not make sense, unless the variable were stationary or an IID. If you can make the series at least stationary (stationary means, roughly speaking, with stable mean and dispersion. Sometimes you can do it by taking the first differences of the series), than you you can test for bivariate correlation.

Germaniawerks
  • 1,027
  • 1
  • 10
  • 15
  • thanks! I'm familiar with stationarity etc and I suspected an answer like this :). In case of having stationary time series is there a test especially for comparing means? (Correlation analysis is barely suitable I think ) – eski Feb 10 '14 at 15:45
  • After ensuring stationarity (for example by means of identifying suitable ARIMA-models), is it possible to simply conduct a t-test for two groups? Does it make sense/is it allowed if there is dependence within one group? Isn't it possible to treat one person as one population? – eski Feb 11 '14 at 14:43