I am analyzing the effect of marital status on life satisfaction for a period of 32 years of balance panel data with missing observations. I am controlling for log income, sex, age, educational level, employment level, region, kids in the family. I am wondering if I should add or not squared age into the regression or not.
My arguments for are that age might me related to the dependent variable - life satisfaction. However, that is exactly the relationship I ant to analyze, how happiness changes with age for example. Life satisfaction is categorical variable running from 0 (not satisfied) to 10 (completely satisfied).
At first step I did transform the variable into dummy 1 if reported levels over 7 and 0 otherwise. Then I did a 'utest' which showed that I should not be adding ages squared.
xtlogit newlifesatis_ c.age age2, fe
utest age age2, prefix(newlifesatis_)
Further I did a 'vif' variance inflation factor test to check for multicollinearity and all variable came with coefficient smaller than 5. However, when running the regression with and without age squared I am getting a bit higher R-squared (0.26 compared to 0.2640) and the significance of the variables doesn't change.
Below is the sample of my dataset. Please, advise if I am wrong. I don't want to bias my regression.
Kind regards,
Gabriela
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input byte(lifesatis marstatus newd11105) float(log_HH_income log_educ) int age float age2 byte(region sathealth empl_level) int syear 8 1 1 10.289362 2.70805 54 2916 1 8 1 1984 8 1 1 9.700759 2.70805 55 3025 1 8 1 1985 10 1 1 10.054663 2.70805 56 3136 1 7 1 1986 8 1 1 8.497194 2.70805 57 3249 1 5 1 1987 8 1 1 9.633514 2.70805 58 3364 1 8 1 1988 8 1 1 9.845593 2.70805 59 3481 1 7 1 1989 . . . . . . . . . . 1990 . . . . . . . . . . 1991 . . . . . . . . . . 1992 . . . . . . . . . . 1993 . . . . . . . . . . 1994 . . . . . . . . . . 1995 . . . . . . . . . . 1996 . . . . . . . . . . 1997 . . . . . . . . . . 1998 . . . . . . . . . . 1999 . . . . . . . . . . 2000 . . . . . . . . . . 2001 . . . . . . . . . . 2002 . . . . . . . . . . 2003 . . . . . . . . . . 2004 . . . . . . . . . . 2005 . . . . . . . . . . 2006 . . . . . . . . . . 2007 . . . . . . . . . . 2008 . . . . . . . . . . 2009 . . . . . . . . . . 2010 . . . . . . . . . . 2011 . . . . . . . . . . 2012 . . . . . . . . . . 2013 . . . . . . . . . . 2014 . . . . . . . . . . 2015 8 1 1 10.289362 2.1972246 44 1936 1 7 3 1984 8 1 1 9.700759 2.1972246 45 2025 1 8 3 1985 9 1 1 10.054663 2.1972246 46 2116 1 7 3 1986 8 1 1 8.497194 2.1972246 47 2209 1 8 3 1987 9 1 1 9.633514 2.1972246 48 2304 1 8 3 1988 10 1 1 9.845593 2.1972246 49 2401 1 10 3 1989 . . . . . . . . . . 1990 . . . . . . . . . . 1991 . . . . . . . . . . 1992 . . . . . . . . . . 1993 . . . . . . . . . . 1994 . . . . . . . . . . 1995 . . . . . . . . . . 1996 . . . . . . . . . . 1997 . . . . . . . . . . 1998 . . . . . . . . . . 1999 . . . . . . . . . . 2000 . . . . . . . . . . 2001 . . . . . . . . . . 2002 . . . . . . . . . . 2003 . . . . . . . . . . 2004 . . . . . . . . . . 2005 . . . . . . . . . . 2006 . . . . . . . . . . 2007 . . . . . . . . . . 2008 . . . . . . . . . . 2009 . . . . . . . . . . 2010 . . . . . . . . . . 2011 . . . . . . . . . . 2012 . . . . . . . . . . 2013 . . . . . . . . . . 2014 . . . . . . . . . . 2015 8 2 3 10.289362 2.484907 21 441 1 8 1 1984 8 2 1 9.89278 2.484907 22 484 1 10 1 1985 7 2 1 9.326166 2.484907 23 529 1 9 1 1986 7 2 1 10.35974 2.484907 24 576 1 10 1 1987 . . . . . . . . . . 1988 . . . . . . . . . . 1989 . . . . . . . . . . 1990 . . . . . . . . . . 1991 . . . . . . . . . . 1992 . . . . . . . . . . 1993 . . . . . . . . . . 1994 . . . . . . . . . . 1995 . . . . . . . . . . 1996 . . . . . . . . . . 1997 . . . . . . . . . . 1998 . . . . . . . . . . 1999 . . . . . . . . . . 2000 . . . . . . . . . . 2001 . . . . . . . . . . 2002 . . . . . . . . . . 2003 . . . . . . . . . . 2004 . . . . . . . . . . 2005 . . . . . . . . . . 2006 . . . . . . . . . . 2007 . . . . . . . . . . 2008 . . . . . . . . . . 2009 . . . . . . . . . . 2010 . . . . . . . . . . 2011 . . . . . . . . . . 2012 . . . . . . . . . . 2013 . . . . . . . . . . 2014 . . . . . . . . . . 2015 10 . 1 7.90802 2.397895 58 3364 1 10 3 1984 10 . 1 7.805475 2.397895 59 3481 1 5 3 1985 9 . 1 8.028781 2.397895 60 3600 1 9 3 1986 10 . 1 . 2.397895 61 3721 1 10 3 1987 end label values lifesatis p11101 label def p11101 10 "[10] Completely satisfied 10", modify label values marstatus d11104 label def d11104 1 "[1] Married 1", modify label def d11104 2 "[2] Single 2", modify label values region l11102 label def l11102 1 "[1] West-Germany 1", modify label values sathealth m11125 label def m11125 10 "[10] Completely satisfied 10", modify label values empl_level e11103 label def e11103 1 "[1] Full Time 1", modify label def e11103 3 "[3] Not Working 3", modify
0 Response to Should I add age^2 when working with panel data or not?
Post a Comment