Transfere them from date formate in any integer formate (such as secons) and calculate either the root mean squared error, or a correlation coefficient.

Anyway, what's wrong with overall agreement?

well, considering the overall percentage of agreement (ie. time_1 = time_2) I only have 68.9%. And in this case does not take into account differences in, for example, 1 second. I think that in these data may not be an error of reliability but a redundancy in function of the measuring instrument (stopwatch).

Well, then use something like a standard error: TS1 and TS2 have a standard error of XXX sec. This can then be compared to the standard error of the stop watch (which should be known). If the standard error between the two time series is smaller than the standard error of the stop watch, the differences are by random nature. Anyway, if you don't know how large the error in measurment is you can't specify if the differences are by change or singificant.

Ok, thank you. Good suggestion...

I made a Pearson Correlation and I find 0,967. I think that is not bad!

I calculate the root mean squared error (RMSE) via "Analyze > Forecasting > Create Models..." is this the right way?

I find RMSE= 3,721. what does this mean?

The RMSE tells you that the deviation between the two TS is 3.7 seconds on average.

If your error in measurement (therefore the stop watch error) is smaller 3.7 you can assume a difference between time series. Anyway, this is not a statistical test! Just a first "insight" into your data.

There is some theoretical support for this? Any reference that I can cite to justify this approach?

Any first grade statistics book

Hello, I've just tried to open your spreadsheet, but the message I get is that this file originated from an internet location and hence maybe unsafe so this is in protected view. Sorry about that but the firewall works at least.

This is an abstract where physical activity data was compared with pedometers and accelerometer data in test - retest circumstances so you could contact the author details below.

Reliability and validity of a modified self-administered version of the Active Australia physical activity survey in a sample of mid-age women

Wendy J. Brown1,

Nicola W. Burton1,

Alison L. Marshall2,

Yvette D. Miller3

Article first published online: 9 DEC 2008

DOI: 10.1111/j.1753-6405.2008.00305.x

© The Authors. Journal Compilation © 2008 Public Health Association of Australia

Issue

Australian and New Zealand Journal of Public Health

Volume 32, Issue 6, pages 535–541, December 2008

Additional Information(Show All)

How to CiteAuthor InformationPublication History

Author Information

e-mail: [email protected]