Inter- and intra-subject reliability

[tbing]

Hi, I would like to measure the inter- and intra-subject reliability of 9 subjects in preparing three different thicknesses of liquids (i.e. mild, moderate & extreme). Subjects were asked to prepare each sample 3 times (i.e mildly-thick x 3, moderately-thick x 3, and extremely-thick x 3). Here's the data I get:


Previous study used replication reliability analysis (see attachment) but I can't find any info about it.
Variability_among_speech-language_pathologists.pdf

Inter-subject reliability: Variability between the subjects in preparing the liquids
Intra-subject reliability: Variability within the subject in preparing the liquids

My questions are:
1. How can I measure the inter- and intra-subject reliability using SPSS?
2. Is it correct to use intraclass correlation (ICC) to measure inter-subject reliability? If so, how to key in the data in SPSS?

Thank you.

Bing

[tbing]

Hi,

Can anyone please help me? I have no idea on how to analyze the data.
Thanks.

Bing

[Art Kendall]

I was unable to read your post as actual data.  If appears to be a raster scan.
the Nabble version of the list allows you to attach your SPSS file.

List members might try to wok with your data, but I doubt if any list member would key it all

You should be able to do a doubly repeated ANOVA and get the variance components.

Are the three thickness mixtures of the same ingredients?

Are the cell entries the all viscometer readings? does a lower number indicate a more viscous fluid?


Am I reading correctly that each subject made 9 attempts, i.e., 3 attempts at 3 viscosities.  

In the newsgroup you said there were 19 subjects.  

I suggest that you start with a 2-way scatter plot. Use 3 different symbols for the 3 attempts and 3 colors for the 3 desired viscosities.

Another visualization would connect each triad of dots.