Please share some information on how to measure and test the heterogeneity in a single variable.
Thanks in advance.
I am not sure exactly what you're looking for, but some common measures of distribution should be helpful, such as variance, kurtosis, skewness, and mean vs median. I am a fan of visual displays of the data, so I would most likely start with a histogram or scatter plot and then look at the numerical summaries of the variable.
Again, it would be helpful to know exactly what you're looking to achieve, as this can help focus the response to something specific.
I agree with Ariel Linden, you can use these methods for assessing homogeneity. Some statistical packages like SPSS allows you to measure homogeneity for t-test and ANOVA through Levene's test.
It will be helpful if you explain the nature of your study and intended analysis.
Hello,
I guess the subject you are looking for is "Correlations Analysis", in which the independent variable have a relationship form 1 to -1 with the dependent variable, furthermore, with the use of standard deviation you can determine if the set of data gives a certain pattern "homogeneous" or just the opposite "heterogeneous".
Naveen,
Generally, chi-squared (χ2, or Chi2) test is an efficient way to measure the data heterogeneity.
I would suggest look at the attached page to learn more.
In Geostatistics, we have a Heterogeneity Index or Dykstra – Parsons Coefficient that measure the heterogeneity for a vector data:
You need to obtain cumulative probability distribution function for a parameter, then you select two quantiles: P50 & P84.1
The HI=(P50-P84.1/P50)*100
If it is < 50%, it is low heterogeneous
If it is > 50%, it is highly heterogeneous
For more details, you may read the attached article.
http://handbook.cochrane.org/chapter_9/9_5_2_identifying_and_measuring_heterogeneity.htm
geolib.geo.auth.gr/index.php/sasg/article/download/7543/7300
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs