Using sample for estimating equation
Posted: Mon Apr 16, 2012 9:09 am
Hi,
I am using panel data but one variable is clearly including several outliers. I would like to estimate my model without the 1st and 99th percentiles of this variable.
Do I need to define the sample beforehand or the estimation window? And how do I define this sample?
Another question about the resulting sample: are the percentiles calculated per period or for the pooled data? The same goes for the omitted data, is 2% omitted of the entire dataset as a whole or 2% per year?
Thanks in advance,
Roy
I am using panel data but one variable is clearly including several outliers. I would like to estimate my model without the 1st and 99th percentiles of this variable.
Do I need to define the sample beforehand or the estimation window? And how do I define this sample?
Another question about the resulting sample: are the percentiles calculated per period or for the pooled data? The same goes for the omitted data, is 2% omitted of the entire dataset as a whole or 2% per year?
Thanks in advance,
Roy