View Full Version : Accelerometers for the measurement of physical activity

02-01-2006, 07:18 AM
Hello all,

I am having trouble understanding/explaining some data from an accelerometer
reliability study we completed. In the study we shook (oscillated in the
vertical plane via a mechanical shaker plate) 40 accelerometers of model x then
we repeated the protocol with 40 accelerometers of model y. The same six
randomly assigned testing conditions were used (see list below).

The unit order is as follows: g-force, m/s^2, Hz, displacement (m)
0.50 4.90 2.50 0.0198
0.50 4.90 2.00 0.0311
0.50 4.90 1.50 0.0552
1.00 9.81 2.50 0.0398
1.00 9.81 2.00 0.0621
1.25 12.26 2.50 0.0497

The trouble is, I don't know how to best order the conditions when presenting
the data. My first choice was to present the conditions (from 1-6; least
intense to most intense) based on the product of acceleration and frequency.
This made intuitive sense because the count output from model x behaved in this
nice ~ linear manner.

However, model y followed no such pattern. In fact, the conditions for model y
seem to be better represented by the product of acceleration and displacement
as this metric describes nicely (~linear from low to high) the count output.

Although I am not an engineer nor a biomechanics guru, I believe there is
something happening here that implies a validity issue with at least one of the

If anyone has any pearls of wisdom I would appreciate their feedback (either by
email or a phone call).

Note: Remember, the count output from these type of accelerometers is in
arbitrary units as they are propriatary to the specific companies (i.e., not
directly comparable across models).

Kind regards,

Dale Esliger
PhD Candidate
College of Kinesiology
University of Saskatchewan, Canada
306-966-1099 (office)
306-665-5921 (home)
Physical Activity Complex, Room 379