I am trying to find the impact of foreign exchange exposure on various factors. I have 12 independent variables including the foreign exchange measures. And there are 6 dependent variables.
First I tried Manova analysis but was not able to come to any significant results with some structural unsuitability of the data to this analysis.
Then I tried working on multiple regression for each of the dependent variables against the 12 dependent variables. This means I am trying to fit 6 regression equation. This seems to work fine and my initial regression shows promising results. However, when I check the assumptions, the data seem to violate the outlier assumption and the normality of the error term assumption.
For the outlier, I solved it considerably by trimming the outliers manually following the robust regression method.
But I have no clue how to meet the normality of error term assumptions.
Could you guide on this, please?
Look at this here on RG:
Article Full-text available
OUTLIERS IN TIME SERIES DATA
- January 2011
- International Journal of Agricultural and Statistics Sciences 7(7):685-91
- 📷Deneshkumar V
- 📷Senthamarai Kannan Kaliyaperumal
In fact, you have a complex scenario, im not familliar with this topic;however, you should simplify it before proceeding to the analysis. Why? MANOVA for more than 2 or 3 dependent variable become difficult to interpret. It's preferable to draw the correlation between dependent and independent and even between dependent, like the figures provided in factor analysis and structural equation modeling, but before that, are these factors are correlated with each other? based on a theoretical framework. Also, it is better to reduce number (combine the scores) of factor if they we're measuring similar item.
After you determine the appropriate statistic method, you should deal with assumptions: regarding to the outlier you can run analysis with and without the outlier. In regard to normality you can look for method robust to normality violation like baysian, etc.
these papers may help;
Article TO DETERMINE SKEWNESS, MEAN AND DEVIATION WITH A NEW APPROAC...
Article PERCENTILE BASED HISTOGRAM BIN WIDTH
- Badges
- Science topic
- Similar topics
- Psychology