|
|
Hello,
I will quickly say what I understand by "reliability ratio" just to make
sure we talk about the same things.
Let's say there is a data set on educational attainment for a set of
countries. For each country there is an entry on P1, the percentage of
population whose highest educational level is tertiary (university
education). There is reasons to believe that there is measurment errors in
that data set. For some countries the data come directly from
census/survey data, for the rest of the countries the data were estimated
by some procedure.
To get a measure of the measurment error in this data set, econometricians
need another data set on the same variable, let's call it P2 (the
percentage of population whose highest educational level is tertiary) and
the same countries but coming from a different source - to ensure
independence between the measurement errors in the two data sets. Now the
econometricians can estimate the "reliability ratio" of the data set1
R1=covariance(P1,P2)/variance(P1)
Ofcourse one can similarly estimate R2. I know a little bit about
reliability ratio, but I got lost while reading the spss help
on "reliability analysis": scales and items...
It is not a big deal for me to estimate my "reliability ratio", but I was
wondering maybe spss offers me a lot more to evaluate the reliability of a
data set. The data I have is on several levels of education (no schooling,
primary, secondary and tertiary) for each country and I have the data for
1990 and 1995.
I do not know if I should spend the time to try and understand the options
offered by spss on reliability analysis, I do not want to unnecesarily
read the 7 papers/books the spss help suggests.
Is anybody familiar with both "reliability ratio" and spss reliabilty
analysis? Can I use the reliability analysis offered by spss to evaluate
the reliability of the data set? Which in the end resumes to evaluating
the reliability of how good the estimates in the data set 1 are.
thanks,
roxana
|
Locked
