ECO 231W

Undergraduate Econometrics

Some clarifications about the Midterm 2

Here are two clarifications about two exam questions:

2.c) The “college major control” to which that question refers is the mathematical content of each college major. From the paper:
"Two different college major controls were used. The first is the standard set of dummy variables for individual college majors. The second college major control is a single variable, the mathematical content of each college major.”

2.k) Although the question asks you to interpret beta_8, we are not looking for the formal interpretation. Rather we are looking for an explanation about which is the likely sign of this coefficient and why.

Midterm 1 rephrasing

I noticed that some students were confused about a few items because of the phrasing, so I rephrased some of the questions. Check the exam questions 2.j, 2.k and 2.n and see if it makes things clearer.

The Final

The Final exam is, as you probably know, on Wednesday 12/16. To find the place and time, look in the calendar page.

The final has one material question, and one essay question. The material question is in the same style as in the midterms. In fact, most items will be the same or similar to the ones in the previous midterms. I may come up with a new question or two, but the previous midterms are really the best guide. I have no intention of surprising you. The material question is worth 70% of the exam grade.

The essay question gives you either a topic, or sometimes a specific research question. You are supposed to write how you would go about doing research on that topic. You can find examples of essay questions in the previous finals I published in the
downloads page. The essay question is worth 30% of the exam grade. Here I give you more details and guidance which may be useful in preparing for the essay question in the exam.

Content: you must explain things in a fair amount of detail. For example, what would you be looking for on a data sets, and which problems do you expect to encounter? You must be specific. “I will check for measurement error in the most important variables” is better than not saying anything, but I’m really looking for “I am expecting a possible measurement error in ***, because of ***. I will try to establish if this is indeed the case by *** (describe action you will take, which can involve looking in the data, in the questionnaire, in the codebook, in another data set [what kind], in the literature [which literature]. You must think about it in the specific case). How will you try to solve it? Can you? How? Do you anticipate it will be basically impossible? What will you do in this case? Which bias do you expect?

Structure: I suggest that you prepare a clear essay structure in advance, where you are sure you will cover every aspect you must discuss in an organized manner. Give the matters their due importance. You are not repeating the last class, you are adapting it to an actual situation. For example, I told you there that you should check for sample selection at a certain point. It was just quick reminder. However, you must discuss this in some detail in the situation, give it a separate paragraph. Which type of sample selection? How do you suppose you will find that out in this specific case? What will you do, etc.

Style: there is no restriction to the size of the essay. We can’t avoid this: flow, style and grammar matter. We don’t particularly reward those, but indirectly we do. We do specifically reward the organization and clarity of your arguments.


I know that you must feel overwhelmed, and also perhaps scared at not knowing what will come. Don’t think this way, because it is not true. The final is not that different from the midterms, and you can prepare very well using those. The essay is in the same style: you know what to expect, you just don’t know the situation. Take a moment to appreciate the fact that although this course was indeed a lot of work, you can answer a question about how you would go about doing a research project on your own. One semester ago you would not be able to even start answering this question, and now you can go all the way in detail. In spite of all the worry about grades that fill your mind most of the time, try to find some satisfaction in what you actually learned: no grade, no bad day, no bad luck is going to take this away from you. Good luck, though luck is hardly necessary.

Midtem 2 answers published

I just published the exam answers. You can find them in the downloads page.

Midterm 2 preparation

Midterm 2 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them.

If you want to know more about the intentions behind the midterm and how to prepare, read
this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Midterm 1 answers published

I just published the exam answers. You can find them in the downloads page.

Midterm 1 preparation

Midterm 1 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them.

The paper question refers to the project paper, which can be found in the
Downloads page. The paper will not be given to you during the exam, nor are you allowed to bring it with you. However, if the question refers to a table in the paper, I will include a copy of the table in the exam.

If you want to know more about the intentions behind the midterm and how to prepare, read
this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Introducing Fall 2015's Paper

The paper we will be using this semester is

Mathematical College Majors and the Gender Gap in Wages

by Catherine J. Weinberger. It was published in 1999 by the journal Industrial Relations.

We will be using this paper in the midterm questions, in the replication, and in the project.

Exams 3:25-4:40 group

By mistake I brought the leftover exams for the 3:25-4:40 group home with me. I will not be coming to the department until Monday. If you want to know your grade, please email Alon. If you can wait, I will put the exams in the box at Harkness around 2 pm on Monday.

Midterm 2 answers published

I just published the exam answers. You can find them in the downloads page.

Typo on the exam preparation

There was a typo on Question 1, item h. The beta_5 should be beta_1, of course. I corrected it. There was also a typo on the model in item a before. I had beta_2 twice there, but I fixed that one as well a few days ago.

Midterm 2 preparation

Midterm 2 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them. If you want to know more about the intentions behind the midterm and how to prepare, read this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Midterm 1 answers published

I just published the exam answers. You can find them in the downloads page.

About the paper part of the Midterm

I received a question by email which made me a bit concerned. Lest there be any confusion, let me be clear: the paper part of the midterm refers to the project paper, which can be found in the Downloads page. The paper will not be given to you during the exam, nor are you allowed to bring it with you. However, if the question refers to a table in the paper, I will include a copy of the table in the exam.

About the Adjusted R-squared

The paper uses a modified form of the R2, the “adjusted R2.” Here is a small explanation about what it means.

Remember that the R
2 never decreases when we add more explanatory variables, and in fact often increases, even if the variable should not be included at all. For example, consider the final grade problem we discussed in class. Say that we included the explanatory variables # classes attended, # office hours attended, and # sections attended. Now, if we add the variable “number of times the student said hello to the instructor,” we can all agree that it is a very silly thing to add to the model, right? Formally this variable is probably not a confounder. However, the R2 may actually increase anyway.

The adjusted R
2 penalizes the inclusion of new variables in such a way that it only increases if the new variable improves the fit more than what would be expected to improve anyway by chance.

Note that the interpretation of the adjusted R
2 differs from the interpretation of the R2. The R2 is a measure of how well the regression line fits the data. The adjusted R2 helps to compare the fit quality of two regressions. In the example above, we can use adjusted R2 to compare the regression that included hours spent studying to the regression that did not include this variable.
Thus, adjusted R
2 is useful in selecting which regression is best. It is worth mentioning that in a univariate regression the R2 will be equivalent to the adjusted R2. In a multivariable regression, these two measures of the model's performance will generally differ.

Midterm 1 preparation

Midterm 1 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them. If you want to know more about the intentions behind the midterm and how to prepare, read this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Introducing this semester's paper!

The paper we will be using this semester is

Gender Differences in the Effect of Education on the Slope of Experience-Earnings Profiles

by Kevin C. Duncan. It was published in 1996 by the American Journal of Economics and Socioloy.

This is a classic topic in Economics. There exist a large literature on earning gaps due to gender (and also racial, height, beauty, etc.) differences. The issue is whether this is due to discrimination or actual productivity differences.

We will be using this paper in the midterm questions, in the replication, and in the project.

The Final

The Final exam is, as you probably know, on the coming Monday and Tuesday. To find the place and time, look in the calendar page.

The final has one material question, and one essay question. The material question is in the same style as in the midterms. In fact, most items will be the same or similar to the ones in the previous midterms. I may come up with a new question or two, but the previous midterms are really the best guide. I have no intention of surprising you. The material question is worth 70% of the exam grade.

The essay question gives you either a topic, or sometimes a specific research question. You are supposed to write how you would go about doing research on that topic. You can find examples of essay questions in the previous finals I published in the
downloads page. The essay question is worth 30% of the exam grade. Here I give you more details and guidance which may be useful in preparing for the essay question in the exam.

Content: you must explain things in a fair amount of detail. For example, what would you be looking for on a data sets, and which problems do you expect to encounter? You must be specific. “I will check for measurement error in the most important variables” is better than not saying anything, but I’m really looking for “I am expecting a possible measurement error in ***, because of ***. I will try to establish if this is indeed the case by *** (describe action you will take, which can involve looking in the data, in the questionnaire, in the codebook, in another data set [what kind], in the literature [which literature]. You must think about it in the specific case). How will you try to solve it? Can you? How? Do you anticipate it will be basically impossible? What will you do in this case? Which bias do you expect?

Structure: I suggest that you prepare a clear essay structure in advance, where you are sure you will cover every aspect you must discuss in an organized manner. Give the matters their due importance. You are not repeating the last class, you are adapting it to an actual situation. For example, I told you there that you should check for sample selection at a certain point. It was just quick reminder. However, you must discuss this in some detail in the situation, give it a separate paragraph. Which type of sample selection? How do you suppose you will find that out in this specific case? What will you do, etc.

Style: there is no restriction to the size of the essay. We can’t avoid this: flow, style and grammar matter. We don’t particularly reward those, but indirectly we do. We do specifically reward the organization and clarity of your arguments.


I know that you must feel overwhelmed, and also perhaps scared at not knowing what will come. Don’t think this way, because it is not true. The final is not that different from the midterms, and you can prepare very well using those. The essay is in the same style: you know what to expect, you just don’t know the situation. Take a moment to appreciate the fact that although this course was indeed a lot of work, you can answer a question about how you would go about doing a research project on your own. One semester ago you would not be able to even start answering this question, and now you can go all the way in detail. In spite of all the worry about grades that fill your mind most of the time, try to find some satisfaction in what you actually learned: no grade, no bad day, no bad luck is going to take this away from you. Good luck, though luck is hardly necessary.


Correction on Midterm 3

In question 2.f, it says: “…once we control for the height and weight variables…” Instead, it should say: “…once we control for the height variables…”

Midterm 4 answers published

Find the answers to Midterm 4 in the download page.

Midterm 1 Preparation

Midterm 1 preparation is published in the download page.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

In this course, as in life, one can always do better. There are always ways to go from a good answer to a brilliant answer. However, as in life, time is scarce. Hence you must learn to prioritize. First think about what are the most important things you should say in your answer. Think how you would say them in a few words. Only then bother with flourishes.

I will give you a booklet with the questions, and another with the space for answering each of the questions. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra pearls of wisdom.

Why do I write the exams so that they are long? Because I am measuring more than your knowledge. I am measuring your maturity with the material. Part of what one gains from studying and reflecting about a topic is the ability to know what really matters, a sense of priority, and a way of expressing your ideas concisely and with the right language. Why do I care to know your maturity? Because I want you to be able to do the same kind of reasoning on the spot in the middle of a company meeting. You need to practice in order to get there.

Presenting this semester's paper

The paper we will be using this semester is

Beauty and the Labor Market

by Hamermesh and Biddle. It was published in 1994 by the American Economic Review, which is the journal with the highest impact in the economics profession.

The topic is a lot of fun. It speaks about an unusual form of discrimination in the labor market: that in favor of beautiful people. Did you ever consider that? Has it ever crossed your mind that more beautiful people can be earning more money just because of their looks? We are not talking about fashion models, we are talking about desk jobs too! Well, this is what this paper claims.

We will be using this paper in the midterm questions, in the replication, and in the project.

Final preparation

Please check the Final entry in the Assignments page for details about the Final. Also, observe that there are several hours of Live Help events the day before the exam.

1. The final will have the same style as the midterm, and several questions in the same style. It will have questions about whether a variable should be included or not, omitted variable bias, and coefficient interpretation.

2. There will be questions about hypothesis testing. You will be asked to “answer a question,” which means that you will have to set up a test, from the null hypothesis to the rejection rule.

3. There may be a question asking the assumptions required for one of the theorems covered in class, or asking you to enunciate an entire theorem, including the assumptions. Always describe the assumptions in detail, like I did the first time I presented each one of them in class. I may ask you to interpret or explain some assumptions, or the meaning of a theorem and why it’s important.

4. There will be several questions examining different failures of MLRs, including one or more of the following: misspecification of the functional form, sample selection, missing data, measurement error, multicollinearity, and heteroskedasticity.

5. There will be a question about proxy variables.

6. There will be questions about instrumental variables, and you will have to argue whether a certain variable is an instrument.

7. There may be a math question. As usual, the point will not be to show math prowess, but rather to measure if you understand and can use the basic math principles required in this course, such as conditional expectations, and summation operations.

8. My questions are rarely abstract. You will be asked all of the above within the context of a specific model, and you should answer with respect to said model.

9. As before, you can always improve on your answers, so you must keep the time carefully. This exam tries to measure your ability to conduct applied research on your own, so you must answer the questions showing your independence. Maturity and clarity of thought show up in the answers, so try to be precise and prioritize. If you have extra time, then come back and consider details, mention puzzling aspects, etc.

10. There will be an essay question, which is worth between 20 and 30% of the exam grade. There I will give you a research question, which may not even be completely defined, and ask you to describe how you would carry out a research project about it. You must base your answer on the last class (which is of course dependent on the entire course). You must consider the following:

a) The question has to be precise. It has to involve the relation between a clear dependent variable and a clear explanatory variable. I may give you an imprecise question, you must refine it to a tractable degree.

b) You must understand what is hard about the question.

c) You must write a model that tries to tackle the question difficulties.

d) You must speak about the data. Can you find the variables you need in a ready survey? What are you looking for in the data set? What do you realistically expect to find in a data set, and what may be difficult? Are you concerned about the way the data was collected, meaning, are there specific data collection strategies that could occur in your problem to which you would give preference, or which you would like to avoid? Is there any modification in your data you expect to do, such as eliminating variables, clumping variables, eliminating observations, etc?

e) You must consider how you would come up with proxies for possible unobservable variables. Which variables that may exist in a real-world data set could be used as proxies for what you need? Include them in the requirements for your data set.

f) You must allow that you will not find proxies for everything. What do you think will still remain in the error term? How will this impact your estimates?

g) How will you tackle general possible failures in your assumptions, such as misspecification of the functional form, heteroskedasticity, or non-normality?

h) Which hypothesis tests, if any, would you perform? Describe the test.

i) What is the result you expect to find, and why?

j) What are the recommendations for policy that your results would imply? Consider the implications for all possible
results, not only for the result you expect.

k) What do you expect to tell your readers with respect to your results? In light of the major issues you expect to find in your research project, what do you expect to have to say as words of caution to your readers?


Best of luck on Sunday, though luck is hardly necessary.

Happy Holidays!

Midterm preparation

Please check the Midterm entry in the Assignments page for details about the Midterm. Also, observe that there is a special Live Help Event that will be held by Linda on Sunday from 3 to 5 pm. The following are recommendations that might help you study.

1. The midterm will surely have questions about whether a variable should be included or not, and why. It will also have questions about omitted variable bias. Most likely I will ask you to compare models with and without certain variables, and ask what happens to the OLS estimators, both with respect to the bias and the variance.

2. There will be a question that asks you to interpret coefficients. Learn how to interpret B0 and the slope coefficients, both with regular variables as well as dummies.

3. There may be a question designed to test your familiarity with the manipulations of conditional expectations. I may test your knowledge of summations indirectly, but this will not be the focus.

4. There may be a question asking the assumptions required for one of the theorems covered in class, or asking you to enunciate an entire theorem, including the assumptions. Always describe the assumptions in detail, like I did the first time I presented each one of them in class. I may ask you to interpret or explain some assumptions, or the meaning of a theorem and why it’s important.

5. There will be a question on hypothesis testing. It will require that you set up a hypothesis test completely from the statement of the null hypothesis to the rejection rule. You will not be required to know specific distributions, not critical values other than the 5% one we always mention (1.96). I may ask you to relate the p-value to the test you set up.

6. Usually my questions will not ask you to do all those things abstractly. Rather, I am more likely to pose a specific model, or a situation, and ask you to reason about those things inside the particular example. It is no use to memorize certain principles and rules if you cannot apply them to real life research questions.

7. Most questions in the exam can be answered in many different ways. It is possible to be wrong, but there can be many different right answers. It is ok if you believe that the answer to a question is “I am not sure,” as long as you can defend what is puzzling about the situation, and your answer makes sense. That said, there are better answers than others, so it is not only about being right, but rather about how completely you can analyze the problem. Remember that in this course there is always a way to improve your answers, so you must also choose when to stop and concentrate your efforts on a different question. Don’t get greedy and waste time.

8. You will be pressed for time. This is partly due to the extremely subjective nature of many of the questions. However, the exam is highly predictable. You can prepare in advance for most questions, and only adapt what you prepared for the specific example in the exam. A student that prepared in advance and dominates the material will tend to be faster than the rest, so I understand that the time is acting here as a tool to separate different levels of maturity about the subject taught in this class. There will be no catch nor any question that requires you to be smart or creative under pressure if I can avoid it. Therefore you will not do worse in this test because you can’t have ideas on the spot.

9. Always remember that this course is graded on a curve, and that the curve in this course is higher than in most other courses. This means that the proportions of A, A-, B+ etc. is much larger than in other courses. Hence, judge the exam and your performance in comparison with others. People seem to forget this and are often positively surprised at the grade they get in the end of the course. Though I try, I am not writing an exam to get a perfect 60% average, or something of the sort. Instead, I am thinking about what you should know, and asking you exactly that. This is why I tell you what will be in the exam. Remember that I could always have made it easier, but I could also have made it harder. I choose exactly the questions that give me a sense of your level of absorption, so I can make corrections to the way I teach, not the way I test you.

Best of luck on Wednesday, though luck is hardly necessary. Make good use of the resources until then.




Midterm 4 Preparation

Midterm 4 preparation is published in the download page.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

By now you are used to the system. This exam is one question shorter than the last, but the questions are more open ended. Plan well, so that you don’t run out of time.

Can’t wait to see how you will do!

Midterm 1 answers published

Find the answers to Midterm 1 in the download page.

Midterm 2 Preparation

Midterm 2 preparation is published in the download page.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

In this course, as in life, one can always do better. There are always ways to go from a good answer to a brilliant answer. However, as in life, time is scarce. Hence you must learn to prioritize. First think about what are the most important things you should say in your answer. Think how you would say them in a few words. Only then bother with flourishes.

I will give you a booklet with the questions, and another with the space for answering each of the questions. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra pearls of wisdom.

Why do I write the exams so that they are long? Because I am measuring more than your knowledge. I am measuring your maturity with the material. Part of what one gains from studying and reflecting about a topic is the ability to know what really matters, a sense of priority, and a way of expressing your ideas concisely and with the right language. Why do I care to know your maturity? Because I want you to be able to do the same kind of reasoning on the spot in the middle of a company meeting. You need to practice in order to get there.

Midterm 2 answers published

Find the answers to Midterm 2 in the download page.

Midterm 3 Preparation

Midterm 3 preparation is published in the download page.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

By now you are used to the system. This exam is shorter than the last, but the topic is historically harder. You will still be pressed for time though.

Long exams have many advantages. The first one is that you will never be stumped. Have you ever stared at an exam question for a while, not knowing the answer? Did you feel like it was bad luck that they asked exactly that part of the material for which you didn’t prepare so well? This will not happen here. If you don’t know the answer, you can still try another question, and thus keep increasing your grade. You may even be able to make up for what you don’t know. Remember: everybody has the same time. Don’t feel frustrated because you knew the answers, but could not write them down as well as you prepared them. This is just your ego, I know very well that you would do much better if you had more time. The point of the exam is to give you the incentive to study, and order students for the grade. All of you have the same time, just show me you can do better than others. You will be able to show me all you know in the Final exam, when you will have 3 hours to answer. Usually the Final is a bit bigger than the midterms, but not so much more.

The second advantage is that this style of evaluation trains you to think fast, and to prioritize. How long do you think you will have in an interview? Somebody will fire questions at you. They will give you a situation, and ask what would you do. So, what would you do? Can you organize your answer in your head in 10 seconds? Can you answer in 3 minutes? Will you be able to pinpoint exactly what matters first?

So, don’t complain about lack of time in the exam. It’s part of the training, as everything else in this course. What is the worse that can happen? You will not be able to answer something you knew, and not win some points you think you deserved. How many? 10? 20? How much is that in the whole grade? Probably, almost nothing. Remember also that if you had more time, everybody else would too, so it wouldn’t make much of a difference in comparison to your colleagues. Now think of your interviews, or presentations, or meetings. There, this skill could mean getting job offers, or promotions. Realize that instead of facing the exams as one more unpleasant form of evaluation, you could be using them as an integral part of your training.

Can’t wait to see how you will do!

Midterm 3 answers published

Find the answers to Midterm 3 in the download page.

The Final

The Final exam is, as you probably know, on Monday. To find the place and time, look in the calendar page.

The final has one material question, and one essay question. The material question is in the same style as in the midterms. In fact, most items will be the same or similar to the ones in the previous midterms. I may come up with a new question or two, but the previous midterms are really the best guide. I have no intention of surprising you. The question will be longer than in the midterms, but not much longer, so I think you will be less pressed for time.

The essay question gives you either a topic, or sometimes a specific research question. You are supposed to write how you would go about doing research on that topic. You can find examples of essay questions in the previous years’ finals.

Content: you must explain things in a fair amount of detail. For example, what would you be looking for on a data sets, and which problems do you expect to encounter? You must be specific. “I will check for measurement error in the most important variables” is better than not saying anything, but I’m really looking for “I am expecting a possible measurement error in ***, because of ***. I will try to establish if this is indeed the case by *** (describe action you will take, which can involve looking in the data, in the questionnaire, in the codebook, in another data set [what kind], in the literature [which literature]. You must think about it in the specific case). How will you try to solve it? Can you? How? Do you anticipate it will be basically impossible? What will you do in this case? Which bias do you expect?

Structure: I suggest that you prepare a clear essay structure in advance, where you are sure you will cover every aspect you must discuss in an organized manner. Give the matters their due importance. You are not repeating the last class, you are adapting it to an actual situation. For example, I told you there that you should check for sample selection at a certain point. It was just quick reminder. However, you must discuss this in some detail in the situation, give it a separate paragraph. Which type of sample selection? How do you suppose you will find that out in this specific case? What will you do, etc.

Style: there is no restriction in the size of the essay. We can’t avoid this: flow, style and grammar matter. We don’t particularly reward those, but indirectly we do. We do specifically reward the organization and clarity of your arguments.


I know you must feel overwhelmed, and also perhaps scared at not knowing what will come. Don’t think this way, because it is not true. The final is not that different from the midterms, and you can prepare very well using those. The essay is in the same style: you know what to expect, you just don’t know the situation. Take a moment to appreciate the fact that although this course was indeed a lot of work, you can answer a question about how you would go about doing a research project on your own. One semester ago you would not be able to even start answering this question, and now you can go all the way in detail. In spite of all the worry about grades that fill your mind most of the time, try to find some satisfaction in what you actually learned: no grade, no bad day, no bad luck is going to take this away from you. Good luck, though luck is hardly necessary.


Introducing Spring 2017's Paper

The paper we will be using this semester is

Occupational differences in the wage penalty for obese women

by Ronald DeBeaumont. It was published in 2009 by the Journal of Socio-Economics.

We will be using this paper in the midterm questions, in the replication, and in the project.

Midterm 1 preparation

Midterm 1 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them.

The paper question refers to the project paper, which can be found in the
Downloads page. The paper will not be given to you during the exam, nor are you allowed to bring it with you. However, if the question refers to a table in the paper, I will include a copy of the table in the exam.

If you want to know more about the intentions behind the midterm and how to prepare, read
this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Midterm 1 answers published

I just published the exam answers. You can find them in the downloads page.

Midterm 2 preparation

Midterm 2 preparation is published in the download page.

This document has all the questions that will be in the actual midterm, with 2 differences: (1) the Material Question will have an actual situation, with real variables like class attendance and smoking. The preparation has generic variables like y and x
1. (2) In the midterm you will only receive a subset of these questions, so that you have time to answer them.

If you want to know more about the intentions behind the midterm and how to prepare, read
this.

The midterm will take place in the regular classroom in the regular class time. You should plan around 1 to 1:05 hour to write your answers.

I will give you one booklet with the questions, and two other booklets, one for the Material Question, and one for the Paper Question. These booklets have space for answering each of the items. Since the answer space of each question is set, you don’t need to write the answers in a row. You can write the bulk of the answers to all questions, and if you have extra time you may go back and complete them with extra flourishes.

Midtem 2 answers published

I just published the exam answers. You can find them in the downloads page.

The Final

To find the place and time of the Final, look in the calendar page.

The final has one material question, and one essay question. The material question is in the same style as in the midterms. In fact, most items will be the same or similar to the ones in the previous midterms. I may come up with a new question or two, but the previous midterms are really the best guide. I have no intention of surprising you. The material question is worth 70% of the exam grade.

The essay question gives you either a topic, or sometimes a specific research question. You are supposed to write how you would go about doing research on that topic. You can find examples of essay questions in the previous finals I published in the
downloads page. The essay question is worth 30% of the exam grade. Here I give you more details and guidance which may be useful in preparing for the essay question in the exam.

Content: you must explain things in a fair amount of detail. For example, what would you be looking for on a data sets, and which problems do you expect to encounter? You must be specific. “I will check for measurement error in the most important variables” is better than not saying anything, but I’m really looking for “I am expecting a possible measurement error in ***, because of ***. I will try to establish if this is indeed the case by *** (describe action you will take, which can involve looking in the data, in the questionnaire, in the codebook, in another data set [what kind], in the literature [which literature]. You must think about it in the specific case). How will you try to solve it? Can you? How? Do you anticipate it will be basically impossible? What will you do in this case? Which bias do you expect?

Structure: I suggest that you prepare a clear essay structure in advance, where you are sure you will cover every aspect you must discuss in an organized manner. Give the matters their due importance. You are not repeating the last class, you are adapting it to an actual situation. For example, I told you there that you should check for sample selection at a certain point. It was just quick reminder. However, you must discuss this in some detail in the situation, give it a separate paragraph. Which type of sample selection? How do you suppose you will find that out in this specific case? What will you do, etc.

Style: there is no restriction to the size of the essay. We can’t avoid this: flow, style and grammar matter. We don’t particularly reward those, but indirectly we do. We do specifically reward the organization and clarity of your arguments.


I know that you must feel overwhelmed, and also perhaps scared at not knowing what will come. Don’t think this way, because it is not true. The final is not that different from the midterms, and you can prepare very well using the prep questions. The essay is in the same style: you know what to expect, you just don’t know the situation. Take a moment to appreciate the fact that although this course was indeed a lot of work, you can answer a question about how you would go about doing a research project on your own. One semester ago you would not be able to even start answering this question, and now you can go all the way in detail. In spite of all the worry about grades that fill your mind most of the time, try to find some satisfaction in what you actually learned: no grade, no bad day, no bad luck is going to take this away from you. Good luck, though luck is hardly necessary.