Behavioural observation- statistics

Yura

New Member
#1
I‘m stuck with a statistics question on a behavioural Observation project.
I have 27 participants, 12 are using device A, 15 device B. they were all observed during a period of approximately 2 hours on how often they used the device, and crucially, for how long the do specific operations (squeeze, roll, ...).
Some participants used the device/toy a lot, some infrequently (between 20 and 100 usage occasions). we would like to see if device A was used for longer periods than device B.
my question is now, do I compare all usage occasions across all participants, with device as grouping variable (which ignores that some data is from the same participants), or do I average for each participans and compare the averages (which ignores that i have around 2000 device usages).
I‘ve done both, results are pretty much the same (one device is used a lot longer for our target manipulation than the other), but which is the correct approach?
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
You could report it both ways, that wouldn't hurt anything. You could also look at the number (count) of times they used the device.

Were subjects randomized to which device they received? If not, you could have self-selection confounding the results.
 

Yura

New Member
#3
Thanks a lot! Participants were assigned a device, so that bit is fine. They used the device approximately the same number of times, and for the same length - but what exactly they did was quite different. device group A (12 participants, about 1000 observations) used our target action a lot longer than device group B (15 participants, about 1200 observations). Participants were not instructed to use the device, it was just provided to them, hence we lost some participants in both groups who just never picked it up.