Hi
I am looking for some advice regarding the appropriate code to use to assess agreement between 5 measurements taken in a single subject which results in correlated data. I am using Stata 14.
Individuals have had 5 separate measurements taken at different anatomical points along a blood vessel.
I have 40 such individuals.
I would like to see if there is a difference between the measurements in these individuals and the overall agreement between measurements.
My data is set up as: individual (1-40), site (1-5), measurement (continuous) with data in long form.
Through my reading, I came across a paper that did a similar analysis.
They state in their statistical methods: "The agreement between parameters describing the P, Pr and Px waveforms was summarised using the Intraclass Correlation Coefficient (ICC) function (Stata 13.1, StataCorp, Texas) derived from a linear mixed model analysis with participant as a random effect and location included as ordinal fixed effect rather than a continuous variable. This is appropriate to our data because the aortic sites were defined relative to anatomical locations, not measured distances."
In trying to replicate this, the code that I have used is:
mixed measurement site || id:
and then estat icc
I am wondering if:
1. This is the correct code for my question
2. Should i be using the expression i.site as it is a factor variable?
The regression output is attached below:
In terms of interpreting the output, does this mean that the intercept is 0.5378991 (which is the grand mean) and that for each change in site as it moves from 1 to 5, there is a 0.0134852 increase at each site in the measurement outcome which is significant (p<0.001).
Thank you for your help
The link to the paper above is: https://www.ncbi.nlm.nih.gov/pmc/art...emss-73076.pdf
Best wishes
Nitesh
I am looking for some advice regarding the appropriate code to use to assess agreement between 5 measurements taken in a single subject which results in correlated data. I am using Stata 14.
Individuals have had 5 separate measurements taken at different anatomical points along a blood vessel.
I have 40 such individuals.
I would like to see if there is a difference between the measurements in these individuals and the overall agreement between measurements.
My data is set up as: individual (1-40), site (1-5), measurement (continuous) with data in long form.
Through my reading, I came across a paper that did a similar analysis.
They state in their statistical methods: "The agreement between parameters describing the P, Pr and Px waveforms was summarised using the Intraclass Correlation Coefficient (ICC) function (Stata 13.1, StataCorp, Texas) derived from a linear mixed model analysis with participant as a random effect and location included as ordinal fixed effect rather than a continuous variable. This is appropriate to our data because the aortic sites were defined relative to anatomical locations, not measured distances."
In trying to replicate this, the code that I have used is:
mixed measurement site || id:
and then estat icc
I am wondering if:
1. This is the correct code for my question
2. Should i be using the expression i.site as it is a factor variable?
The regression output is attached below:
In terms of interpreting the output, does this mean that the intercept is 0.5378991 (which is the grand mean) and that for each change in site as it moves from 1 to 5, there is a 0.0134852 increase at each site in the measurement outcome which is significant (p<0.001).
Thank you for your help
The link to the paper above is: https://www.ncbi.nlm.nih.gov/pmc/art...emss-73076.pdf
Best wishes
Nitesh
Comment