Question

1. In class I frequently mention that we are less concerned with the statistical significance of regression coefficients than a typical econometric exercise. In fact I might make mention of this once a week or more. (a) (5 points) In your own words explain why this is the case. (b) (5 points) Is there a substitute for this concept of significance that takes the place of significance of regression coefficients? Explain.

Answer 1

1.) As some coefficients may come out to be statistically insignificant due to various factors like problem of small sample size or data may still ne important for our model. Also, it doesn't always indicates a practical significance and can be misinterpreted if the language in reporting result is not carefully chosen.

2.) We can look at correlation coefficients as a tool to measure the relation or association between two arrays of data. The correlation coefficient intervals range between -1 to 1. Where , -1 = perfect negative correlation

0 = no correlation,

1 = perfect positive correlation.