Stats Project, Can Anyone Help

[sarahlow]

I've got a stats project to do for uni about obtaining accurate blood pressure readings. Can anyone have a quick look and maybe tell me where to start or what methods i should be using.
Question is:
A study was designed to access four techniques of measuring blood pressure in both children and adults. The four techniques were auscultation (A), Doppler (D). flush (F), and palpatation (P) and observations using these techniques were comapred with intra-arterial measurement which is considered accurate but unsuitable for routine use.
Observations using the four techniques were made by both a nurse and a doctor in a group of 30 children and adults.
The Aim Of The Study:
(i) which is the most accurate method?
(ii) Are there any differences between accuracy achieved by the nurse and the dostor?
(iii) Are there any differences between the accuracy obtained for the children and adults?


Any help on what methods i should be using would be muchly appreciated.

Sarah

[JohnM]

A good candidate for this type of study is factorial ANOVA (Analysis of Variance), using independent variables:

age (Child/Adult)
technique (4) and
assessor (nurse/doctor)

The dependent variable would be the difference between each technique and the intra-arterial measurement.

Overall, the most accurate measurement would be the one with the smallest difference...

After examining the ANOVA for significant effects, you could then do some post-hoc comparisons along the lines of your research questions that you listed.

The advantage of doing ANOVA is that you can look at interactions between the factors - i.e., do doctors/nurses make more accurate measurements on certain techniques; do certain techniques work better on children than on adults, etc.

[sarahlow]

thanks very much, am familiar with ANOVA so shouldn't be too much of a problem.

Sarah

[sarahlow]

still same project - my advisor said i might want to be doing two-way ANOVA (makes sense because results are related as measuements are taken from each person more than once). Don't know if your familiar with Mathematica but thats what we use here. So i was trying to fit the model Resp=subject+technique. (think it would be two-way with no interaction) she kind of said i was along the right lines but i have a grouping variable as well as a repeated factor. Any help with this would be great.

Sarah