Homework
4: ECO220Y
Required Exercises: Chapter 6: 5, 17, 18, 20, 28, 31, 35
Required Problems:
(1) Answer and EXPLAIN your answer. How do the graphs below differ?
20 20
15 15
y14
y15
10 10
5 5
0 0
0 5 10 15 20 0 5 10 15 20
x14 x15
(A) the variance of y14 is much larger than the variance of y15
(B) the scatter plot for x15 and y15 shows no relationship whereas x14 and y14 are related
(C) the scatter plot for x14 and y14 shows a steeper relationship than the one between x15 and y15
(D) the scatter plot for x14 and y14 shows a strong relationship than the one between x15 and y15
(E) All of the above
(2) Are each of these observational or experimental data? Explain and specifically apply course concepts to each.
(a) Data on the interest rate and growth rate of GDP over time in Canada
(b) Obesity rate in mice fed low‐carb diet versus regular diet
(c) Prices and quantities sold of bath tissue for randomly selected retail outlets
(3) See Exercise 40 (in Chapter 6) in the textbook for background. Consider 2009 data (see StatLink button on page 15 of
http://www.oecd.org/pisa/pisaproducts/48852548.pdf.) Here is the variance‐covariance matrix for these data.
. correlate reading math science, covariance;
(obs=65)
| reading math science
-------------+---------------------------
reading | 2664.7
math | 2926.14 3576.25
science | 2841.29 3255.2 3143.86
(a) Create a correlation matrix. (Note: You need only the information in the matrix above.)
(b) See the following six scatter plots (which continue onto the next page). Are reading and math scores strongly
correlated or associated? Are there any significant concerns about outliers?
Students' Scores on PISA Students' Scores on PISA
65 countries in 2009 (r = 0.95) 62 countries in 2009 (r = 0.96)
Mean Reading Score
Mean Reading Score
550 550
500 500
450
450
400
350 400
300 350
300 400 500 600 350 400 450 500 550
Mean Math Score Mean Math Score
Page 1 of 5
Students' Scores on PISA Students' Scores on PISA
65 countries in 2009 (r = 0.98) 63 countries in 2009 (r = 0.98)
Mean Reading Score
Mean Reading Score
550 550
500 500
450
450
400
350 400
300 350
300 400 500 600 350 400 450 500 550
Mean Science Score Mean Science Score
Students' Scores on PISA Students' Scores on PISA
65 countries in 2009 (r = 0.97) 62 countries in 2009 (r = 0.98)
600 550
Mean Math Score
Mean Math Score
500 500
450
400
400
300 350
300 400 500 600 350 400 450 500 550
Mean Science Score Mean Science Score
(c) Which kind of data are these? How does that affect the interpretation of the correlations?
(d) Consider the 2012 PISA data (see StatLink button on page 19 of http://www.oecd.org/pisa/keyfindings/pisa‐
2012‐results‐volume‐I.pdf). Compare and contrast these results with those from 2009 discussed earlier.
. correlate reading math science, covariance;
(obs=65)
| reading math science
-------------+---------------------------
reading | 2216.65
math | 2505.97 3075.51
science | 2338.37 2736.36 2576.39
(4) Read SW11 (on Quercus). Assess your understanding of this reading with this quiz.
(4.1) What does the phrase “empirical evidence” mean? It refers to evidence based on ____.
(A) many experts’ intuitions
(B) observations recorded as data
(C) the outcome of deductive reasoning
(D) the results of simulations derived from theory
(E) rigorous mathematical modeling of a phenomenon
► Questions (4.2) – (4.4): Consider the example about applying fertilizer to some randomly selected plots of
farmland. Suppose there are 50 plots in the treatment group and 50 plots in the control group.
(4.2) This is an example of which kind of data?
(A) time series data
(B) experimental data
(C) observational data
(D) natural experiment data
(E) longitutudinal (panel) data
Page 2 of 5
(4.3) What happens to the 50 plots in the control group?
(A) all get fertilizer
(B) none get fertilizer
(C) some selected plots get fertilizer
(D) some randomly selected plots get fertilizer
(E) all are subject to careful control such that they each receive the same water, sunlight, weeding,
seeds, wind, slope, etc.
(4.4) What is the distinguishing feature of a randomized controlled experiment in the farming example?
(A) that the 100 plots are randomly divided into the two groups
(B) that the 50 plots of land in each group are perfectly identical in every respect
(C) that the plots in the treatment group are carefully matched to plots in the control group
(D) that the randomization process has been controlled to ensure that the plots are comparable
(E) that the researchers have verified that all other variables are held constant across these plots
► Questions (4.5) – (4.9): Data on the unemployment rate, inflation rate, and growth rate, in each province for
each of the last twelve months would be an example of _____________ data.
(4.5) Could “observational” correctly fill in the blank? (A) correct (B) incorrect
(4.6) Could “experimental” correctly fill in the blank? (A) correct (B) incorrect
(4.7) Could “cross‐sectional” correctly fill in the blank? (A) correct (B) incorrect
(4.8) Could “time series” correctly fill in the blank? (A) correct (B) incorrect
(4.9) Could “longitudinal (panel)” correctly fill in the blank? (A) correct (B) incorrect
(4.10) With observational data that shows that in schools with smaller class sizes the learning outcomes are
better than in schools with larger classes, why is it difficult to answer “Does reducing class size improve
elementary school education?” It is difficult because ____.
(A) there are no data available that quantitatively measure outcomes for learning
(B) there is a lot of variability in outcomes across students: each student is different
(C) class sizes vary little across schools making it hard to separate the signal from the noise
(D) factors like neighborhood wealth vary across schools and affect class sizes and outcomes
(E) all of the above
► Questions (4.11) – (4.14): You wonder if the format of a questionnaire affects how students answer about
their undergraduate experience. You select a random sample of 30 students and randomly divide them into two
groups. One group answers the questionnaire online while the other uses pen and paper. Which are valid
criticisms of your study design? (A) valid; (B) not valid
(4.11) While the results will not by systematically wrong, there will be a fair bit of sampling noise and
this will limit your ability to answer your research question.
(4.12) You have made no attempt to ensure that the two groups are otherwise identical and this means
that your data should not be used to answer your research question.
(4.13) You have failed to ensure that other factors are held constant across the two groups and this will
lead to an overestimate of the causal effect of the questionnaire format.
(4.14) You should have conducted a randomized controlled experiment rather than relying on
observational data.
Page 3 of 5
(5) Get comfortable using the terms “endogenous,” “exogenous,” and “endogeneity bias.” For these three contexts – (A)
Canadian inflation and interest rates, (B) chocolate consumption and Nobel Laureate production, and (C) drug dosage
and hours of sleep – apply these terms appropriately. Answer with several sentences for each case.
(6) Consider the cross‐tabulation below of two dummy variables from a survey of 774 respondents like Carlin et al.
(2017). The variable male is 1 if the respondent is male and 0 otherwise. The variable chosedom is 1 if the respondent
chose the dominant credit card and 0 otherwise. What is the coefficient of correlation between these two variables?
| male
chosedom | 0 1 | Total
-----------+----------------------+----------
0 | 178 215 | 393
1 | 183 198 | 381
-----------+----------------------+----------
Total | 361 413 | 774
(7) See the data and scatter diagram below from the Council of Ontario Universities. (Data retrieved on September 22,
2017 from http://cou.on.ca/numbers/multi‐year‐data/enrolment/.) It records the number of full time equivalent (FTE)
students enrolled in undergraduate (UG) programs across all of Ontario’s universities annually since 2000.
(a) The coefficient of correlation between the variables year and UG_tot_FTEs is 0.9683. In light of the given
background information, what does that value of the correlation mean?
(b) How would the coefficient of correlation change if enrolments were measured in 1,000s of FTE students?
(c) How would the coefficient of correlation change if year were recorded as 0, 1, 2, …, 16 instead of 2000, 2001,
…, 2016?
(d) How would the coefficient of correlation change if the level of enrolment in the year 2000 were 420,000
instead of 244,945? Also, how would that affect its ability to summarize the strength of the relationship?
year UG_tot_FTEs
2000 244945
2001 257488
ON Universities: Rising Undergraduate Enrolments
2002 278765 450000
2003 311660
Total FTE UG Enrolment
2004 327371 400000
ON Universities:
2005 341882
2006 350030 350000
2007 348611
2008 352945
300000
2009 367901
2010 381583
250000
2011 391502
2000 2005 2010 2015
2012 400272 Year
2013 406407
2014 410086
2015 413206
2016 420687
Page 4 of 5
(8) Researchers often compute the covariance and correlation between pairs of dummy variables. This simple context is
a great opportunity to build conceptual understanding. Without actually computing the coefficient of correlation, use the
cross tabulations below to assess whether the correlation will be: exactly zero, very close to zero, positive, or negative.
In cases where it is not zero nor close to zero (no relationship), share an assessment of the strength of the relationship.
(a) First situation to assess:
| y
x | 0 1 | Total
-----------+----------------------+----------
0 | 949 233 | 1,182
1 | 2,268 550 | 2,818
-----------+----------------------+----------
Total | 3,217 783 | 4,000
(b) Second situation to assess:
| y
x | 0 1 | Total
-----------+----------------------+----------
0 | 28 386 | 414
1 | 562 24 | 586
-----------+----------------------+----------
Total | 590 410 | 1,000
(c) Third situation to assess:
| y
x | 0 1 | Total
-----------+----------------------+----------
0 | 382 659 | 1,041
1 | 152 807 | 959
-----------+----------------------+----------
Total | 534 1,466 | 2,000
(d) Fourth situation to assess:
| y
x | 0 1 | Total
-----------+----------------------+----------
0 | 1,638 702 | 2,340
1 | 462 198 | 660
-----------+----------------------+----------
Total | 2,100 900 | 3,000
Page 5 of 5
