Say for example we are trying to assess the impact of X on Y. Y in turn consists of Parameters K,L,M,N,O,P. K,L,M,N,O,P being qualitative in nature. More appropriately if we are trying to assess the impact of a particular government scheme on poverty alleviation of its beneficiaries of the scheme. Which best statistical techniques can be used to rightly assess the impact?
The measurements are quantitative as well as qualitative, however highly subjective. For qualitative parameters we are using a 10 Point Likert Scale.
However other quantitative variables are string variables.
You may use non-parametric correlation coefficient, however it depends on type of the data (both in ordinal scale). Other possibility is to use contingency table with Chi square test, with - sometimes - discretization of quantitative data.
if 'Y' is a categorical data ı think we can use logistic regression or the other lineer regression...
Thanks Aleksander, Wen Wang, Erhan Tecim. I have elaborated my Research Problem in another Post. Kindly offer your suggestions. Please find the Link
https://www.researchgate.net/topic/Statistics/post/Which_statistical_technique_is_appropriate_for_Impact_Studies
You may use Ordinary Least Squares (OLS) with Dummy Variable, where the impact of X on Y and on the qualitative variables (i.e. K,L,M,N,O,P) can be measured by this technique. However, the OLS depends on the size of your number of observation.
First it is better to define some indexes for the variables that are qualitative in nature. then you may use the least squares method by the software EVIEWS. It gives you some other useful statistics such as R square, Durbin Watson stat, Fisher Stat, etc. It also provides some tests for you to make sure the estimation is legitmate. You just enter the DATA in the software and it does all the estimation for you.
Hi Audil. I hope its not too late to answer your question. If your qualititative data is intervalised/categorised may I suggest then please consider the following. DO NOT use Pearson's chi-squared statistic. When you have an X (predictor) variable and a Y (response) variable, the chi-squared statistic does not pick up on the direction on the association (only that the association exists). Instead consider the Goodman-Kruskal tau index, although you will need to convert it to a C-statistic (very easily done) to test the statistical significance of the "asymmetric" association between X and Y. If you want a visual representation of the intervals/categories you can use non-symmetrical correspondence analysis.
-Thanks Eric... I am not very well about Goodman-Kruskal tau index,,, Can you please explain it a bit further...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics