Assignment #1
ECON 323: Econometric Analysis 2 - Winter 2024
Hard copy due February 5, 2024, 11:30am in class- output and
interpretations due
Intermediate Deadlines (all due online - only hand in the code and re-
sults, not interpretations):
(a): Due January 18, 11:59 pm
(b)-(i) incl: Due January 25, 11:59 pm
(j)-(l) incl: Due February 1, 11:59 pm
While cooperating on the assignment is encouraged, plagiarism is not.
I will only accept hand written assignment submitted in person. DO NOT
SUBMIT YOUR ASSIGNMENT ELECTRONICALLY. Late assignments
will receive a 10% penalty per day that it is late, up to the time that it
is corrected in class, after which they will receive a mark of zero. Show
your work as no marks will be allocated for the final answer alone. For the
intermediate deadlines, the weight will be transferred to the final submission
upon the upload of a vif to VIF.uwaterloo.ca or the self-declaration of an
absence. No late assignments will be accepted for the intermediate deadlines.
Use Stata or R to do these. If you choose to use another software, please
get my approval by January 17.
Question 1
Using the Canadian Community Health Survey (CCHS) (2017-18) from
Odesi, answer the following questions.
(a) In preparation for the rest of the assignment, download the data
and transform it so that you can use it. Report summary statistics (mean,
standard deviation, minima and maxima) for the variables listed in parts
(b) and (k).
(b) Regress using ordinary least squares how many “active days” re-
1
spondents have in a week (PAADVDYS) on gender (DHH SEX: male is the
baseline), age (DHHGAGE: recode it to a linear variable. using the mid
points of the category and 85 for the oldest group), province of residence
(GEO PRV: Ontario is the baseline), educational attainment (EHG2DVR3:
recode it so that you can control for whether the individual has not finished
high school, is a high school graduate, or attended post-secondary. High
school degree is the baseline) and health status (GEN 005: please code it
so that better health is a higher number). Report your results in equation
form.
(c) If an observation in the sample is such that the individual is a 48
years old female with some post-secondary education, lives in Saskatchewan,
how many times does the model predict that they will be active in a week?
(d) This observation is such that they were active for 3 days. Calculate
the residual.
(e) Perform a t-test of the equality of the coefficient of age to -0.05 at
5% of statistical significance (in the regression you ran in (b)).
(f) Should you control for education in your regression? Discuss.
(g) Is heteroskedasticity present in this model?
(h) Now adjust your method of estimation so that you treat the depen-
dent variable as the type of variable it should be and go back to assuming
homoskedasticity, regardless of your results in (g). Repeat questions (b) and
(f) using the correct estimation method and report the “new answers”. Has
anything changed? Why?
(i) What is the average marginal effect for age in this regression? Why
should you consider the AME rather than the coefficient in this situation?
(j) One could argue that there is endogeneity between health and not
participating in physical activity. What do we mean by this?
(k) Going back to the specification of your model in (b), assuming that
you could use perceived mental health (GEN 015) as an instrument, how
would you do so? Estimate your model.
(l) What conditions should GEN 015 fulfil for it to be considered a good
instrument? Show whether GEN 015 fulfils them (you can assume that all
variables are normally distributed for this sub-question).