2SLS regression model

#1
I have a query about instrumental variable (IV) and 2SLS (actually three stage) regression. I am trying to estimate effect of core team(a group of 5 students who work together in first semester of MBA) on the performance of MBA grads (performance, the dependent variable is salary after graduation).

In the first stage, I regresses Fall semester GPA (Y1) on team id, GMAT score , International student (considering that international students grapple with adjustment factors), subject of graduate degree and years of experience.

In the second stage, I regressed cumulative GPA (Y2)on predicted value of Fall semester GPA from previous regression and tracks (i.e. if student has taken marketing specialization or finance specialization).

Finally, I regress log of salary on predicted value of Y2 (Y2hat) and employment stream (Finance, software or consulting).

I have two questions:
a) Is it a correct way to approach?

b) In first regression, team id has P value of 0.084 and t stat of 1.73. Can it be considered significant? Further, R^2 value is .1846, though altogether Prob > F = 0.0000.

Thanks in advance.
 
#2
Well What i understand over here is

Y is your salary factor
X are your on team id, GMAT score (Continuous data), International student Discrete Data , subject of graduate degree Discrete Data and years of experience (Continuous data)

you can use chi sqr for discrete and regression for continuous

you can see the P value supports the alternate hypothesis (as per the convention i can say ). i.e. you must have taken
null hypo = no dependency on factors
alt hypo = dependency on factors

here your P value is greater than 0.05 that means relation ship is there

your R sqr value is ~ 18.5% that means the given factors are not significantly related to salary / offer (Y) factor

THAT MEANS
There needs to be some other facts which are influencing the results.
The discrete and continuous data needs to be studied separately.
 
#3
Thanks, Prince. Would you please clarify one more thing? As some data is in deciles (salary) and some in quintile (GMAT score), will there be any other way to approach?
 
#4
A MA of 7 is really high unless you have seasonality (and that would be an unusual lag to have seasonality on since it usually occurs at 12/3 or 4). Most ARMA models have 3 or less - in part because you can commonly model a high MA as a lower order AR process. Note that certain AR processes, negative autocorrelation I believe, show up as a sine wave.



-----------------------------
The most attractive feature of designer Lingerie Dress very cheap is its soft and delicate embroidery designs.