Adjusted Cross-Tabs Idea

From InterSciWiki
Jump to: navigation, search

Doug to Wiley Companion co-editors: Mike suggested that CoSSci could include his http://lucy.kent.ac.uk/EthnoAtlas/ which is pretty neat because of the coloring. I didn't get any other feedback. Mike also volunteered to do a second such interactive crosstab tool for SCCS, same deal (coloring again).

I have an addition idea were we to agree to include one or both in CoSSci and the Wiley Companion: If its on CoSSci imagine that Anthon and Malcolm worked out a simple version of Dow-Eff functions for a crosstab pair: i.e., it applies the Wy to the first part of 2SLS as a control for autocorrelation. The effects of Wy for language, distance, etc. could be calculated

Then Anthon and Malcolm might be able to work out a way that an "adjusted for autocorrelation" cross-tab would have the same row and column marginals but instead of integers each cell could have a one-digit decimal adjustment. Might be 2x2, 2x3, 4x6, etc.: whatever. But now the shading would appear as before but WOULD BE VERY DIFFERENT with autocorrelation effects.

Might take a bit to work this out. But imagine how useful that would be for students learning about autocorrelation effects. And simply taking a published cross-tab and replicating an analysis with an adjusted significance test for the adjusted cross-tab.