Simulating the case it is found that the result for total universe "A and B" does not resemble the SEM shape of the sectors. This analysis is important in econometrics where urban areas incomes, population fractions and SEM distributions are very different from rural ones. Even if the SEM is the same, the factors U and p produce also unexpected results. The file attached shows it for two Pareto's distributions. If this happends for just one SEM, would this effect many multivariate analysis methods and predictive theories based in only one SEM´s family?
Dear Emilio, do you study the effect of addition of two or more random variables? Because that problem is well studied in statistics.
Demetris. I have just started to explore the theme you mention, looking at Wikypedia there are some notes about doubtful premises and methods. My math level is not good enough to understand matrix algorithms, so I have done first few simulations for two or three variables with other simpler methods. Another interesting place is the International Journal of Forecasting at http://www.sciencedirect.com/science/journal/01692070/29. "The illusion of predictability: How regression statistics mislead experts" (Soyer, Emre; Hogarth Robin M.) 2012. They sustain that problem is still not well studied nor well understood, neither well applied. So I need ideas that help me to clarify this important point. Thanks, emilio
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling