Independent variable: Financial liberalization
Dep. var.: level of Islamic Finance and Banking Development
Control variables: Percentage of Muslims in the country, World income Gini Index, GDP level, Index of corruption perception (developed by Transparency International).
I expect some control variables to correlate with FL because FL usually follows similar patterns with economic growth.
You're talking about multicollinearity. As I understand it, multicollinearity may not bias your results but it will increase the possible standard errors. If your software lets you look at variance inflation factors, see how bad they are. Beyond 5 is considered suspicious.
You're talking about multicollinearity. As I understand it, multicollinearity may not bias your results but it will increase the possible standard errors. If your software lets you look at variance inflation factors, see how bad they are. Beyond 5 is considered suspicious.
I am very new to multivariate statistics as you see (because I could not spot what you said) I just read what collinearity means. Thank you a lot. I use R Deducer I am sure there are tools to check variance. At least it shows F statistics and with more experience and help of ResearchGate :) I hope to learn to do it properly.
This is from the statsoft book.
Multicollinearity and Matrix Ill-Conditioning
This is a common problem in many correlation analyses. Imagine that you have two predictors (X variables) of a person's height: (1) weight in pounds and (2) weight in ounces. Obviously, our two predictors are completely redundant; weight is one and the same variable, regardless of whether it is measured in pounds or ounces. Trying to decide which one of the two measures is a better predictor of height would be rather silly; however, this is exactly what you would try to do if you were to perform a multiple regression analysis with height as the dependent (Y) variable and the two measures of weight as the independent (X) variables. When there are very many variables involved, it is often not immediately apparent that this problem exists, and it may only manifest itself after several variables have already been entered into the regression equation. Nevertheless, when this problem occurs it means that at least one of the predictor variables is (practically) completely redundant with other predictors. There are many statistical indicators of this type of redundancy (tolerances, semi-partial R, etc., as well as some remedies (e.g., Ridge regression).
The spss can eject the variable which seems redundant (if I'm not mistaken). But does collinearity means always that a variable is redundant? I can imagine cases of collinearity in which two variables can perform next to one another within a model without one being redundant.
@Lyubov
In linear regression, nonlinear, univariate or multivariate, as you said @Mitchell, has an effect on the estimation of the dependent variables, over or under estimating it.
the problem is to consider, what are the variables which control and independents.
I think we should make a further analysis on the variables used, as I see it, it seems that the independent variable your sample, it is not and could depend on some control.
Hi Lyubov, you're very lucky, because these days there are great statistics resources on the Internet. The ones at UCLA are especially well known, but many things are on Wikipedia, and many faculty at universities maintain web pages on their favorite topics and specialties. So now that you know terms like "multicollinearity" and "variance inflation factor" you can just go ahead and do Google searches on them and find lots of information. You will find plenty to read, and when you find particularly good sites that you like, you can bookmark them and go back to read about other topics. Having been traine, long, long ago, sometimes I don't think of Googling as soon as I should, but there are great things out there that I use, such as effect size calculator, or statistical power calculators, that when I find ones that I like (and trust) that I bookmark on my browser. You can also try searching the topics in the search window here on ResearchGate, but I think Google does a better job in prioritizing search results.
On the topic of your question, statistically, there usually isn't any difference between a control variable and an independent variable (predictor) so if they are collinear, each affects the other. Some software apparently lets you designate control variables, but as far as I know the difference is only in the presentation, not in the statistical model itself. Bob
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics