ECON 560 – Data Informed Decision Making

Problem Set 4
10 Points
All answers to problem set questions must be typed so they can be reviewed by Turnitin. Please
submit your completed problem set through Canvas using the instructions found on the syllabus.
Problem 1 (2 points)
There has been an increasing amount of attention on supplementing America’s power grids with
solar panels and wind turbines to provide sources of renewable energy. However, unlike
traditional sources that are weather-resistant, solar panels can only generate power while the sun
is out, and wind turbines require enough windy weather. Hence, energy experts have called on
companies to create many solar panels and wind turbines scattered across various parts of the
power grid to counter this issue.
In about 4-6 sentences, explain how the law of large numbers can be applied to here to explain
why it is beneficial from an energy generation standpoint to hook up thousands of solar panels
and wind turbines scatter across different places to the electric power grid vs. only a few of these
in limited locations. (Hint: think about the reasoning behind the casino and insurance examples
to get started).
The law of large numbers says as the number of trials increases the result will become more
predictable and accurate. For companies to create many solar panels and wind turbines will help
smooth out the bad weather conditions. An example would be if an area is windy, another area
sunny this means at least an area will be creating some energy. This geographic diversity helps
average out differences and helps increase accuracy of renewable energy for the power grid. This
leads the power supply becoming more predictable and accurate and less dependent on certain
weather conditions. This then makes the power grid better equipped to meet the energy demands
it needs.
Problem 2

This problem uses the Job training.xlsx data posted on Canvas and requires Excel’s Data
Analysis Toolpak. Note that you must copy the output when asked for, but if it’s not asked for
then no output is required and you can just write the answer.

A local job training program has been funded by the state of Virginia. You are contracted to
evaluate the effectiveness of this program and determine if it has improved the hourly wages of
workers.
You have information on 600 people who applied to be part of the 12-week job training program.
Due to capacity constraints, 200 of these applicants were accepted into the program while 400
were denied acceptance.
The data includes hourly wages of all 600 individuals 1 year after the job training program
ended, including those who were accepted into the job training program (Accepted) and those
who were denied entry (Denied). For clarity, this is a cross-section of data after any job training
has taken place.
You want to compare the mean hourly wage of those who completed the job training vs. those
who applied but were denied.

A. (2 points) Construct the 90% confidence interval and the 99% confidence interval for the
mean hourly wage of those accepted into the program. Explain what these confidence intervals
tell us and why there is a difference between these intervals.
90% confidence level of hourly wage lies between $12.02 to $13.72 while the 99% confidence
level of hourly wage lies between $11.53 to $14.21. The 99% confidence level has a wider range
because a higher confidence level requires a higher certainty.
B. (1 point) Assuming a two-tailed test, what are the null hypothesis and alternative hypothesis
of this study?
The null hypothesis would be the hourly wage of all 600 individuals one year after the job
training program. The alternative hypothesis would be the mean hourly wage of those who
completed the job training vs. those who applied but were denied.
C. (1 point) Use Excel’s Data Analysis Toolpak to conduct the appropriate t-test for this study
and copy and paste the output below. Hint: The unequal size of the samples is not a problem
when conducting t-tests.
t-Test: Two-Sample Assuming Unequal
Variances

Variab Variab
le 1 le 2
12.869 11.531
Mean 6 48
53.296 27.115
Variance 3 5
Observations 200 400
Hypothesized Mean
Difference 0
df 303
2.3144
t Stat 5
0.0106
P(T<=t) one-tail 56
1.6498
t Critical one-tail 98
0.0213
P(T<=t) two-tail 11
1.9678
t Critical two-tail 24
D. (3 points) Based on a two-tailed test, determine if your results are statistically significant at
the α = 0.05 level and whether we reject the null hypothesis. Then do the same at the α = 0.01
level. Explain how you arrived at these conclusions and be clear with your language.
At the 0.05 level we can reject the null hypothesis because the difference in means is statistically
significant. At the 0.0l level we can fail to reject the null hypothesis because the difference is not
statistically significant.
E. (1 point) T-tests assume that we have drawn the sample at random to make claims about the
population. Based on the details of the job training program in the setup of the problem, describe
whether we are working with random samples in this situation and why this is or is not a problem
for drawing conclusion about the effectiveness of the job training program. About 3-5 sentences
is fine here.
In this problem we are not working with random samples. These applicants were chosen in the
program because they had what it took and the capacity of the program was only 200 people so
the people were not chosen through random selection. If there were any observed differences in
the wages it is not just the training program it could also be due to the factors the applicants had
before the program like their previous experience. With that it is harder to draw conclusions
about the effectiveness of the job training program.