This data shows the relationship between spending per pupil and the
composite score on some standardized exam.
a)There is another variable which measures whether state adopted a new
statewide curriculum during the past 5 years.
b) Run the regression and interpret the parameter estimates and
interpret the dummy variable.
explain.

State	"Spending per Pupil($)"   "Composite Score"	  New Curriculum Used
============================================================================
Louisiana	4,049	                    581	              Yes
Mississippi	3,423                  	    582	              Yes
California	4,917	                    580	              Yes
Hawaii	        5,532	                    580          	No
South Carolina	4,304	                    603         	No
Alabama	        3,777	                    604         	No
Georgia 	4,663	                    611	                Yes
Florida 	4,934	                    611          	Yes
New Mexico	4,097	                    614          	Yes
Arkansas	4,060	                    615         	Yes
Delaware	6,208	                    615         	No
Tennessee	3,800	                    618         	No
Arizona	        4,041	                    618         	No
West Virginia	5,247	                    625	                Yes
Maryland	6,100                       625	                 Yes
Kentucky	5,020	                    626          	No
Texas	        4,520	                    627	                No
New York	8,162	                    628	               Yes
North Carolina	4,521	                    629	               No
Rhode Island	6,554	                    638	                Yes
Washington	5,338	                    639	                 Yes
Missouri	4,483	                    641          	No
Colorado	4,772	                     644	        Yes
Indiana	         5,128	                     649	        No
Utah	         3,280	                     650	        Yes
Wyoming	         5,515	                     657	        No
Connecticut	 7,629	                      657	        Yes
Massachusetts	 6,413	                     658	        Yes
Nebraska	5,410	                     660	        No
Minnesota	5,477	                     661	        Yes
Iowa	        5,060	                      665	        Yes
Montana	        4,985	                      667	         No
Wisconsin	6,055	                      667	         Yes
North Dakota	4,374	                       671	         No
Maine	        5,561	                       675	        Yes

---

Clarification of Question by k9queen-ga on 18 Nov 2003 22:30 PST

I am supposed to clarify these are for studying purposes

Hi k9queen, 

First off, let's start with some summary statistics.  I've named the
variables in the model STATE, SPENDING, SCORE, NEW_CURR.

                   Mean       Std Dev       Minimum       Maximum
 SPENDING    5068.82857    1085.69144    3280.00000    8162.00000
 SCORE        631.17143      27.57502     580.00000     675.00000
 NEW_CURR       0.57143       0.50210       0.00000       1.00000

The only thing in this table that might be useful is the mean of
NEW_CURR. Remember that, since NEW_CURR is a dummy variable that can
take the values of 0 and 1 only, its mean of 0.57143 is also the
percentage of the number of states that have adopted the new
curriculum (where NEW_CURR = 1).  So 57.143% have adopted this
curriculum.

Now we run a regression of the variables SPENDING and NEW_CURR on
SCORE (since, surely, the thing we're interested in finding is the
effect of spending and the curriculum choice on student performance).
The resulting output (from TSP, and  including a bunch of information
that will be of little or no value to you):

 Current sample:  1 to 35
 Number of observations:  35

        Mean of dep. var. = 631.171      LM het. test = 7.01475 [.008]
   Std. dev. of dep. var. = 27.5750     Durbin-Watson = 1.97525 [<.516]
 Sum of squared residuals = 537880.  Jarque-Bera test = 7.84107 [.020]
    Variance of residuals = 16299.4   Ramsey's RESET2 = 338.984 [.000]
 Std. error of regression = 127.669    Schwarz B.I.C. = 221.919
                R-squared = .116601    Log likelihood = -218.364
       Adjusted R-squared = .089831

            Estimated    Standard
 Variable  Coefficient     Error       t-statistic   P-value
 SPENDING  .119913       .655148E-02   18.3032       [.000]
 NEW_CURR  -2.98439      44.8987       -.066469      [.947]


I just copied this output straight from the TSP output file, so you
can ignore most of the numbers above.

The first thing to note is the R-squared value of 0.116601.  So only
11.66% of the variance in scores is 'predicted' by the variables in
the model, SPENDING, and NEW_CURR.  That's not a lot of predictive
power.

When we look at the estimated coefficients of the regression, the
coefficient on SPENDING is 0.119913 with a standard error of
0.655148x10^-2, which is small enough that the coefficient is
significant.  This says that for every extra dollar spent on students,
SAT scores rise by 0.119913.

If you check this yourself by just multiplying some numbers by
0.119913, you'll see that it's a very rough predictor of the data in
the table provided, but this explains why the R-squared is so low.

The second coefficient, that for the dummy variable NEW_CURR, is
-2.984139, but it has a standard error of 44.8987, which is much
higher than the coefficient itself, and so this coefficient can not be
considered significant.

What this tells us is that spending does have some (though quite
limited) value as a predictor for SAT scores, but the adoption of the
new curriculum does not have any value as a score predictor.

I hope this was clear enough for you.  If you have any problems,
please ask for a clarification.

Best of luck with your studies.

Hibiscus