I have some data, which are measured with some measuring instruments also i have taken the subjective feel for those measurements also. The objective data and subjective feedback are not hand in hand. So here i want to find a correlation between objective data and subjective feedback. I am a beginner in the field of statistics so need help!
The correlation coefficients work with numbers, without knowing if your data were objectively or subjectively assessed. You have to transform your subjective data in the same "currency" your objective data have. Then you calculate Spearman och Pearson correlation coefficient depending on your sample size. If it is less rtha 30 you take spearman, if it is > 30 you can use the other one.
As long as you can quantify your results, you can easily perform correlation analysis by first considering the distribution of normality of the data. You can perform asymmetry and kurtosis for this. Then, if it is normal, use Pearson's correlation, if not Spearman.
You can code your subjective data and then correlate it with your objective data.
General rule is that you cannot mix ordinal and nominal data in statistics. However, Descriptive statistical methods like Chi Square allows you to mix them up. This is not a correlation but it provides acceptable levels of statistical measure of differentiation between the groups by categories.
Hi!
My dissertation focuses QoL measuments in the psychiatric field. QoL is most often measured by subjective ratings, but since patients often have limited capacities in different ways I construed an interviewer based instrument as well where the interviewer rates the QoL in the same scales as the subjective instrument. The interviewer had a manual with specified end- and middlepoints for the different scales. Nine interviews where tape-recorded and rated by independent listerners. The inter-rater reliability was good.
My approach allowed me to compare the patients subjective ratings with the ratings done by the interviewer. The results showed that the patients own ratings did not differentiate between healthier patients and stepwise more severely ill patients, but the intervjuer rarings did. Thus, the level of QoL could not be detected by the subjective rating, but indeed with the interviewer rating. The correlation between the parallell ratings was acceptable but not so useful since different patient-groups and even healthy people get the same level of QoL.
- Badges
- Science topic
- Similar topics
- Instrumentation Engineering
- Instruments