3. You are a policy analyst in a nonprofit organization: \"My Best Friend is My

Question

3. You are a policy analyst in a nonprofit organization: "My Best Friend is My DOGma" who advocate and fund dog rescue and adoption program. You write a report based on an analysis of the effective ness of three different programs: A, B and C, which was implemented in 300 animal shelters across the countries. Your dependent variable is an adoption rate in percentage point unit. All independent variables presented below are in thousand-dollar unit. Here's the result. Coefficients 20 0.3 0.05 0.17 IVs Standard Error 0.1 0.9 0.02 0.08 Constant Adjusted R-square 0.99 N 300 c) (Bonus point: 20 points) You present the analysis to your colleague at "Are You Kitten me?" organization. Your colleague says to you "It is highly pawssible (pun intended) that your analysis has a serious problem." Your colleague is generous enough to detect and alert the problem to you, but not enough to give you a detail about the problem in this analysis. Based on the above table, what is a sign or information that you as a practitioner in statistics and econometrics should be worried about and what could be a potential problem or the cause of problem?

Explanation / Answer

Adjusted R2 tells you “how much information the model is explaining, relative to all possible information”, and things that test for whether the independent variables are significant, usually in the form of a p-value.

Degree of freedom = N - number of coefficients = 300 -4 = 296

For 5% significance level and degree of freedom = 296, t value is 1.97

For variable A, t = Coefficients / Standard error = 0.3 / 0.9 = 0.33

As, t statistic for variable A is less than the critical value of 1.97, variable A is insignificant in the model.

The model has high adjusted R-square but one of the variables in the model is insignificant. This can be sign of a bad model. A high adjusted R-squared values are a symptom of overfit models. Overfitting is problem, where the model is trained enough on the given data such that it accurately works on the trained data but the accuracy is low on new observations. When choosing a regression model, our goal is to approximate the true model for the whole population. If we accomplish this goal, our model should fit most random samples drawn from that population.

Related Questions

Navigate

• Browse (All)
• Browse 3
• Subjects

---

---

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