In: Statistics and Probability
For a regression of test score (T) on the endogenous variable student-teacher ratio (R), Hoxby (2000) suggests using as an instrument the deviation of potential enrollment from its long-term trend (P), where "potential enrollment" means how many children of kindergarten age there are (whether or not they attend public school). Which of the following arguments would NOT support P as an instrument for R?
a. | Due to high adjustment costs of buildings and teachers and the small/discrete number of classrooms per school, schools cannot perfectly adjust each year to maintain a target student-teacher ratio |
b. | Parents with young children are more likely to move into good school districts with low student-teacher ratio |
c. | There are fluctuations in birth rate due to sheer random chance |
d. | Changes in school district quality are slow and contribute to the long-term enrollment trend, but not deviations from the trend |
PLEASE PROVIDE EXPLANATION IN ANSWER
note that
regression analysis tries to capture the pattern in the variation of the independent variables to account for the variation in the dependent variable
Here dependent variable is T
and independent variables are R , P etc.
so something that is purely random cant be captured by the model , any pattern in the data can be captured
a. Due to high adjustment costs of buildings and teachers and the small/discrete number of classrooms per school, schools cannot perfectly adjust each year to maintain a target student-teacher ratio
clear pattern given , hence would support
b. Parents with young children are more likely to move into good school districts with low student-teacher ratio
clear pattern given , hence would support
c. There are fluctuations in birth rate due to sheer random chance
only randomness , hence would not support
d. Changes in school district quality are slow and contribute to the long-term enrollment trend, but not deviations from the trend
clear pattern given , hence would support