Chapter 3, #1-3

*Question 1...
* 95% CI for Age = 1.43 +or- 1.96(.46) = (.5284, 2.3316)
* Does NOT cross zero, so Null is REJECTED.
* 95% CI for Weight = 25.9 +or- 1.96(31) = (-34.86, 86.66)
* Does cross zero, so the Null is NOT REJECTED.
* The CI of the Age factor is quite tight, so the magnitude of effect for every one unit chage in age is quite a bit. The Weight factor CI is quite large, so there has to be an extremely large change in weight to see an effect of magnitude.
* Problem 2. Explore the relationship between dollars and the other covariates listed above. Fit a model that you feel best captures this relationship.
use "(data location)\3.ex.Funding.dta", clear
regress dollars incid preval hospdays mort yrslost disabil
regress dollars incid preval hospdays mort yrslost
regress dollars preval hospdays mort yrslost
regress dollars hospdays mort yrslost
regress dollars hospdays yrslost
* Problem 3. Perform a forward stepwise linear regression of log(dollars) against the following potental covariates: log(incid), log(preval), log(hospdays), log(mort), log(yrslost) and log(disabil).
generate logdol = log( dollars)
generate logincid = log(incid)
generate logprev = log( preval)
generate loghosp = log( hospdays)
generate logmort = log(mort)
generate logyrs = log( yrslost)
generate logdis = log( disabil)
* Threshold = .1
sw regress logdol logincid logprev loghosp logmort logyrs logdis, forward pe(.1)
* Threshold = .2
sw regress logdol logincid logprev loghosp logmort logyrs logdis, pr(.2)
* The covariates selected are logdis loghosp for the cutoff = .1 and logyrs logdis loghosp logmort for the cutoff = .2