I read a post the other day (I believe by Fedoja) that suggested we do a regression analysis to determine what factors are significant in JW growth are decline. Basically is a strong economy correlated with a JW drop in numbers? I have statgraphics and I know how to crate forecast models so if you guys can get me the stats in Excel format I can run the analysis. I will then print the results here. In the end we should have a reasonable forecast model for JW growth and decline.
Stats needed: JW numbers if possible broken down by month and by country
Economic variables: I am thinking unemployment rate, GDP, and Inflation (I got the USA’s stats). These need to be broken down.
Any other Factors: The more factors you can think of to go into the model the better, the RA will determine if they are significant or not. So if there are any other variables you think are significant to JW growth are decline let me know.
I have a couple of thoughts. To make the task easier, you might want to use yearly averages for, say, the last twenty years. I don't think monthly variations would account for much, but then, you never know. I think you would want to calculate both multiple correlation and multiple regression. The correlation would tell you the amount of variance in JW growth/decline accounted for by the combined economic variables. Mulitple regression would tell you how much weight each economic variable had. One other economic variable you could include would be median family income in constant dollars.
Thanks for the ideas. In my analysis I will auto correlate the dependent variable (JW growth) and cross correlate the independents (economic factors) to determine significance. Once I have the significant factors then I will run a RA (using the significant factors) to determine the model. Using yearly data would make it eaiser so I will probably do that.
Do you have any idea where I can get the median family income in constant dollars.
I have OECD data for practically all economic indicators all over the world, but unfortunately only up to 1997 or so. I'll check if I can find anything more recent.
