Appropriate test for measurements repeated at two levels

[Richard Devey]

I have a dataset containing heart rate measurements for two groups of
subjects over four time periods. I need to assess if there are differences
in heart rate between groups and over time. If I had one measurement per
time period I could use repeated measures or the friedman test. The
complication is that within each time period the subjects' heart rate is
measured at intervals as resistance is increased. The number of
measurements within the given time period can differ depending on the
fitness of the subject e.g. subject 1 might have five heart rate
measurements with increasing resistance while subject 2, fitter than
subject 1, might have ten measurements. I could take a mean of these
measurements to get one value per time period and proceed with repeated
measures analysis. However, I'd be interested to know if there is a test
which can take into account (1) the two groups (2) the four time periods
and (3) the multiple measures of heart rate within each of the four time
periods. Any suggestions would be greatly appreciated.

Richard Devey
University of Johannesburg
South Africa

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

Your measurements are nested within time periods, and time periods are nested within subjects.  So you could run a multilevel model (via MIXED).  With this approach, there is no need to have equal numbers of measurements per time period (or time periods per subject), as there would be for repeated measures ANOVA.  The book by Singer & Willett is very helpful, and most of their examples are demonstrated on the UCLA Computing website:

    http://www.ats.ucla.edu/stat/examples/alda.htm

HTH.
 

Richard Devey wrote

I have a dataset containing heart rate measurements for two groups of
subjects over four time periods. I need to assess if there are differences
in heart rate between groups and over time. If I had one measurement per
time period I could use repeated measures or the friedman test. The
complication is that within each time period the subjects' heart rate is
measured at intervals as resistance is increased. The number of
measurements within the given time period can differ depending on the
fitness of the subject e.g. subject 1 might have five heart rate
measurements with increasing resistance while subject 2, fitter than
subject 1, might have ten measurements. I could take a mean of these
measurements to get one value per time period and proceed with repeated
measures analysis. However, I'd be interested to know if there is a test
which can take into account (1) the two groups (2) the four time periods
and (3) the multiple measures of heart rate within each of the four time
periods. Any suggestions would be greatly appreciated.

Richard Devey
University of Johannesburg
South Africa

=====================
To manage your subscription to SPSSX-L, send a message to
[e-mail] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD