I've found an inverted u-shaped relation between the dependet variable "roa" and the variable of my main interest "varx".
For doing that, I used the following code:
Code:
reg roa2013 /// Dependent variable varx2013 /// Variable of main interest varxsq2013 /// Variable of main interest squared // Controls controla2013 /// controlb2013 /// controlc2013 /// controld2013 /// controle2013 /// controlf2013 *Utest command utest varx2013 varxsq2013
Code:
reg roa2013 /// Dependent variable c.varx2013##c.varx2013##c.moderator /// Variable of main interest // Controls controla2013 /// controlb2013 /// controlc2013 /// controld2013 /// controle2013 /// controlf2013
The interaction varx2013##c.moderator as well as varx2013##varx2013##c.moderator (and the quadatic term) is significant.
I have now two questions:
1) Because of the fvvarlist-operators I'm using in the second regression, I cannot take advantage of the utest command anymore. Is there another way to test the significance of the inverted u-shaped relation and it's moderator (incl. turning point)?
2) How can I scatter the two u-shapes (with and without moderator) in one graph?
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input float roa2013 double varx2013 float(moderator controlb2013 controlc2013 controld2013 controle2013 controlf2013 controlg2013) .05226809 5 5.666667 18.14469 .06278002 4065218.5 .3193709 1 .061403 .0553228 4 5 14.84563 .06730169 319434.25 .54305065 1 .05809223 .08281977 5.5 5 10.947507 .11188485 4836470.5 .8503829 0 .09462444 .11069837 4 2 12.29784 .09459518 29807718 .900606 0 .05524298 .0805665 5.5 5 11.919324 .1036812 1741782.6 .6284503 1 .05541043 .11150632 7 6 10.62099 .10003947 1741782.6 .6284503 1 .05541043 .09160694 2.5 6 10.154166 .1723896 5872063 .3670754 0 .0001397046 -.13085464 4 6 8.91211 .033733662 3748229.5 .7868435 0 -.016308693 .05229507 5.5 4 12.829374 .063645296 29807718 .900606 0 .05524298 .029555695 2.5 5 17.536222 .020356985 976984.8 .007668674 . .04100607 .02923595 5.5 5 17.217607 .05679471 1224742.6 .04854649 1 .033227947 -.18810207 5.5 3 16.669004 -.3212181 47613800 .7451095 1 .05309721 .016424885 2.5 6 10.947678 .2246402 4065218.5 .3193709 0 .061403 .1704918 5.5 5.5 12.142302 .2218278 35624924 .3757325 1 .064520545 .03112654 5.5 7 13.287418 .071752846 4836470.5 .8503829 1 .09462444 .04407731 7 7 12.174742 .07259794 35624924 .3757325 1 .064520545
0 Response to Inverted u-shaped relation: how to deal with a moderator
Post a Comment