Interlude: A brief primer on empirical analysis Let\'s say you want to conduct a

Question

Interlude: A brief primer on empirical analysis Let's say you want to conduct an empirical study to compare the annual salaries of high school teachers in the state of New York versus those in the state Rational crime basics 9 of New Jersey. You collect data on 100 teachers in each state, and calculate a simple average. You find that the average salary in New York is $2,000 higher than the average salary in New Jersey. You conclude that New York high school teachers are better paid than their New Jersey peers. Based on the limited information you ave, this is a sound conclusion. But what other information may be useful to your study? As of now, the only information you have to explain the difference in salaries is the state in which each teacher is employed. But a teacher's salary may depend on many other factors, such as: number of years of experience; number of years of education; public or private school teacher; inner-city or suburb teacher; grade level taught; subject taught; age; gender; race; union status; and, possibly, several others. Perhaps your finding is largely explained by the fact that, in your sample, the New York teachers have more years of experience than their New Jersey counterparts. This may be important information to consider Now assume that you can collect all the relevant data discussed above. With these data, you can use regression analysis to estimate your model. At its most basic, when you estimate a simple regression equation you are able to isolate the effect you are interested in studying. With the salary example, you can now distinguish between the salaries of New York and New Jersey teachers by controlling for al other variables. In other words, this statistical technique, in a sense, forces all the other variables to be identical for the teachers, and can just focus on the state in which the teacher is employed. If you find that New York teachers still have a larger annual salary than New Jersey teachers, you can say that their salary is larger with all else equal. That is, the difference in salary cannot be explained by differences in any of the other variables that are being controlled for in the regression equation Furthermore, with a single regression equation, you can also isolate any other effect you are interested in studying. For example, you can determine the effect of

Explanation / Answer

With the help of statistical and economic tools like regression analysis, we can measure the effect of one or more variable on other variable or variables keeping the influence of other factors constant. One such relationship is comparing annual incomes of new york high school teachers and new Jersey high school teachers keeping other things like age, experience etc constant. We can then say whether there salary is lower or higher than new Jersey peers

It is common to discuss the sign, magnitude and statically significance of test e.g we can say the income of Newyork teachers is positive, is greater by 1000. But 1000 is average amount. We may have to qualify the statement by saying that we are 95% (or 90% etc) confident that average salary in Newyork high schools is between 500 to 1500 dollars greater in new york than in new Jersey.

But there is scope for legitimate debate regarding which data, method etc should be used. The best study should openly sat what it's limitations are. Further no empirical study can be truly definitive in the sense that empirical techniques tend to have shortcomings and empirical work is contentious, but it does not reflect poorly on economic reasoning.

Related Questions

Navigate

• Browse (All)
• Browse I
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.