[gov2001-l] identifying the problem - Gov2001

smarkus＠fas.harvard.edu

11 May 11 May

5:09 p.m.

i think it's multicollinearity only if there's a perfect linear relationship b/w X2 and X1. it's more like endogeneity among independent variables. it seems that an interaction term X1*X2 would be handy here. i may be wrong.. s t a n **************************** Stanislav Markus Ph.D. Candidate Harvard University Department of Government e: smarkus(a)fas.harvard.edu t: 617.513.5407 -----Original Message----- From: gov2001-l-admin(a)fas.harvard.edu [mailto:gov2001-l-admin@fas.harvard.edu] On Behalf Of Olivia Lau Sent: Sunday, May 11, 2003 4:29 PM To: gov2001-l(a)fas.harvard.edu Subject: [gov2001-l] identifying the problem I have a problem. I have a dependent variable Y, and two independent variables X1 and X2, such that X1 -> Y X1 -> X2 X2 -> Y Is this multicollinearlity? I'm at a loss for words, someone please help! Thanks, Olivia. _______________________________________________ gov2001-l mailing list gov2001-l(a)fas.harvard.edu http://www.fas.harvard.edu/mailman/listinfo/gov2001-l

Reply

olau＠fas.harvard.edu

10:40 p.m.

More specifically, I am asking, what exactly is the consequence of controlling for a consequence (X2) of the key explanartory variable (X1)? Including a consequence of the key explanatory variable may illustrate the causal mechanisms through which the key expanatory variable acts. (Yes, I'm aware that the causal mechanism is not causally prior and hence not an omitted variable.) There doesn't seem to be any loss of efficiency by including X2 as well as X1. Thanks, Olivia. On Sun, 11 May 2003, Kosuke Imai wrote:

...

Not sure exactly what you mean, but it seems that the problem is endogeneity. Kosuke On Sun, 11 May 2003, Olivia Lau wrote:

I have a problem. I have a dependent variable Y, and two independent variables X1 and X2, such that X1 -> Y X1 -> X2 X2 -> Y Is this multicollinearlity? I'm at a loss for words, someone please help! Thanks, Olivia. _______________________________________________ gov2001-l mailing list gov2001-l(a)fas.harvard.edu http://www.fas.harvard.edu/mailman/listinfo/gov2001-l

Reply

Gary King

6:20 p.m.

controlling for the consequences of a key causal variable (although perhaps ok for finding the mechanism by which the key variable has its effect) biases the effect. it could decrease efficiency also since you wind up estaimting more params. Gary On Sun, 11 May 2003, Olivia Lau wrote:

...

More specifically, I am asking, what exactly is the consequence of controlling for a consequence (X2) of the key explanartory variable (X1)? Including a consequence of the key explanatory variable may illustrate the causal mechanisms through which the key expanatory variable acts. (Yes, I'm aware that the causal mechanism is not causally prior and hence not an omitted variable.) There doesn't seem to be any loss of efficiency by including X2 as well as X1. Thanks, Olivia. On Sun, 11 May 2003, Kosuke Imai wrote:

Not sure exactly what you mean, but it seems that the problem is endogeneity. Kosuke On Sun, 11 May 2003, Olivia Lau wrote:

I have a problem. I have a dependent variable Y, and two independent variables X1 and X2, such that X1 -> Y X1 -> X2 X2 -> Y Is this multicollinearlity? I'm at a loss for words, someone please help! Thanks, Olivia. _______________________________________________ gov2001-l mailing list gov2001-l(a)fas.harvard.edu http://www.fas.harvard.edu/mailman/listinfo/gov2001-l

_______________________________________________ gov2001-l mailing list gov2001-l(a)fas.harvard.edu http://www.fas.harvard.edu/mailman/listinfo/gov2001-l

Reply