EViews.com

Hi,

I'm good with time series, but this is my first venture into cross-section and panel data.

I have series of logins and then transactions for a 6 month period. Naturally, the number of transactions on each login varies between logins. Login is text and numbers (Lnnnnnn). Transaction is a number and I've also made unique, so if there are two with value 1021, I've changed second to 1021.01 (fairly rare occurance)

To structure the data in Proc/Structure Resize current page (Undated with ID Series), I've entered Login and Transaction, but says "Illegal panel structure LOGIN is nested within TRANSACTION" but not sure how this can be the case, not least because Transaction is unique within whole range of workfile. Login and Transaction do uniquely identify each observation, but not quite sure what message is telling me.

Many thanks!

Hard to say without seeing the data.

I've attached sanitised version of the file, but still get same error.

I guess I'm not seeing the panel structure of your data. You need to have two series that combined uniquely identify each observation. You cannot have a situation where one series uniquely identifies the observations (since that isn't a panel, that's just a cross-sectional data set).

Yes, understood. I see your point. This is my first time with panel/cross-section so not yet coming naturally to me. I will revisit tomorrow and see how I get on. With many thanks.