Econ113 – Econometrics - Dobkin
USCS – Spring 2011
Mid-Term Exam Topics Basics
Given data – find mean, median, variance, covariance and correlation. Give interpretation of each. Strengths of weaknesses of each. For mean & median, how does median correct for possible issues with outliers? For Cov vs Cor, how does correlation correct for unit of measure issues with covariance? Normal Distribution – finding probabilities of a range of possible outcomes.
Given mean and variance (or given mean & standarddeviation) and assuming a normal distribution, be able to use your z-tables to find the probability of some range of events happening. See week 3 section notes for examples with solutions. See lecture March 31. Be careful, √ ଶ √
OLS Regressions (every exam has asked to churn out beta from some data sample)
Find regression of some model (may involve taking log). Given some data, be ready and able to find ଵ & (nearly guaranteed question) Plot your data (get x and y axis right) and interpret your regression line.
Be able to interpret your regression coefficients in words (remember this is different for level-level, log-level and lob-log regressions). Given your values for & ଵ, be ready to be given some variable and estimate the dependent variable of your model. Calculate r-squared of your estimate, (will include churning out SSR & SST). Interpret R-sqrd correctly.
See week five mid-term review section notes – and week four section notes. OLS Assumptions.
What do assumptions 1-4 mean about where our estimate, ଵ is centered around?
Assumption Three – practically every exam has some question about this assumption? Zero conditional mean, or | 0. If the tricky assumption that’s typically violated. Be ready explain what this means, and given some model from a previous question, be ready to produce examples where this assumption is violated. Violation of assumption three implies some kind of bias, be ready to state which direction the bias is (positive or negative). Lastly, be ready to state under what conditions this assumption holds
See April 7th notes & April 19th, omitted variable bias notes.
https://sites.google.com/site/curtiskephart/ta/econ113
TA Curtis Kephart
Omitted Variable Bias. Going from a two-variable regression to a one variable regression, what is the bias of your ଵ equal to? What do the terms that define that bias mean? What does the ‘fifth assumption’ tell you about the variance of our estimate? Assumption 6.
Hypothesis Testing (although uncommon on previous 113 mid-terms, Dobkin has resorted the class for Spring 2011, making a question on this area more likely)
Given a ଵ estimate & ଵ , be ready to find the tstatistic.
Given a t-stat, be able to perform a hypothesis test – 5% significance. 1% significance.
Given the above, be ready to calculate the one-sided or two-tailed p-value. And ready to interpret a given p-value (example, given a p-value of 0.015 means that there’s a 1.5% (very-low) chance our study was a fluke such that the significance of ଵ is aberrant. Also, this would pass at the 5% sig level (since 0.015 ! 0.05), but not at the 1% significance level, since 0.015 " 0.01) Be ready to list three’ish reasons why you choose a twosided hypothesis test.
Given some scenario, be able to state the correct null and alternative hypothesis. For what values of do you use a student dist z-table?
Be ready with interpretations for your p-value. Be ready to explain what rejecting your null hypothesis means for how good your estimate is.
See past final exams for good p-value & hypoth-test questions. Dobkin moved lecture s around this year.
Extra Credit Questions – when a test is curved the way this one is, there are technically no extra credit questions. These questions tend to be really tricky question that the math majors in the class have something to be entertained by. They are usually straight out of people’s notes. Additional Topics – this to-do list is based on a review of all past exams. You can be asked really anything emphasized in lecture. (that is, don’t hate me if I left off a topic covered on the exam) Advice from Class Grader - I asked the class grader for areas where people had the most issues with the HW (which, given this test is curved, is probably a good area to focus additional study)
She says: paraphrased: issues with interpreting what R^2 is and understanding the different assumptions. Exam grading may be very picky with the exact interpretation of B1, B0, and R^2, so I would make sure they know exactly what to write for that.
But other than that, I think it would be good to go over the assumptions and from HW 2, the probability problem (normal distribution problem) was a common problem they got wrong.
