You are on page 1of 3

Question 1

The human resources manager of a company wants to find out whether the training programs they
offer their employees increases productivity. The employee database has detailed records on 445
employees, including their productivity and how many hours of training they have attended. The
dataset also includes information on education level and years of experience.
The manager starts her analysis with the finding that the correlation coefficient between
productivity and training is 0.38.

a) Interpret the computed correlation coefficient. Can the manager conclude that the training
causes productivity to increase? [2 marks]

The manager extends her analysis by estimating the following regression model:
𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦! = 𝛽" + 𝛽# 𝑡𝑟𝑎𝑖𝑛𝑖𝑛𝑔! + 𝛽$ 𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛! + 𝛽% 𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒! + 𝑢!
Where training is the number of hours of training attended, education is the years of education of
the worker and experience is the years of work experience that the worker has.
b) Explain two advantages that regression analysis has over computing the correlation
coefficient. [4 marks]
c) Why does the manager include education and experience as explanatory variables even
though she only wants to find out whether the training has worked or not? [4 marks]
The results from estimating the regression model are given below:
Table Q1
Coefficients Std Error t Stat P-value
Intercept -0.541 2.091 -0.259 0.796
train 1.659 0.632 2.624 0.009
experience 0.045 0.044 1.027 0.305
educ 0.393 0.174 2.262 0.024

d) Interpret the estimated slope coefficients on train and educ. [6 marks]


e) Based on the regression results, would you conclude that the training has a significant
impact on employee productivity (at 5 percent significance)? [4 marks]

[20 marks in total]


Question 2
The Board of the company Alpha-Beta asks the HR team whether is true that female employees
are more likely to leave the company than male employees. The analyst draws up the following
summary table using data from the last five year:
Table Q2a
Employment status Gender Total
Male Female
Currently employed 300 500 800
Not employed 75 125 200
Total 375 625 1000

a) Discuss how the results in this table can be used to answer the Board’s question. [4 marks]
b) Provide a rough sketch of the visualization you would suggest for this information and
justify your choice. [4 marks]

To study the issue further, the analyst decides to compare the average job satisfaction ratings
between 300 male and 500 female employees currently working in the company to check whether
the mean job satisfaction score is significantly lower for women than for men. He finds that the
average score for women is 73 (out of 100) and the average score for men is 78.

c) Write down the null and alternative hypothesis for the test that the analyst plans to carry
out. [3 marks]
d) Explain why the analyst cannot make a decision simply by looking at the two average
values he has computed. [3 marks]

The analyst isn’t sure of what test to use so he performs all of the two-sample tests available in the
Data Analysis Toolpak add-in on MS Excel. The p-values he obtains for each test are given below:
Table Q2b
Name of test p-value
Paired t-test Error
Two-sample t-test with unequal variances 0.02
Two-sample t-test with equal variances 0.06
F-test for equal variances 0.13
e) Explain how you would interpret the results above to answer the question of whether
women have lower average job satisfaction than men. [6 marks]
[20 marks in total]
Question 3
The CEO of a software development firm wants to find out how employees feel about working
from home (WFH), which they have been doing since March 2020. From informal conversations
she has had with several employees from across the three departments in the firm (Admin, Sales
and Product Development), she has made the following observations:

Obs1. Preferences for working from home differ drastically between the three departments
Obs2. Due to WFH, employees are spending less time in meetings than they did before
Obs3. Female employees find it harder than the male employees to focus on their work when
they are WFH
Obs4. Employees do not seem to be working longer than the 40-hour work week norm

To try and verify these observations, the CEO decides to carry out a survey that will be sent to a
sample of the firm’s 350 employees.

a) Provide two reasons why the CEO plans to carry out the survey on just a sample of the
firm’s employees. [2 marks]
b) Based on the information given above (and the observations which need to be verified),
briefly describe a sampling strategy that you think would be most appropriate for the firm’s
CEO to follow. [6 marks]
c) For each of the four observations that the CEO wants to verify, write down the null and
alternative hypotheses to be tested and explain which type of test should be used to verify
these claims. [6 marks]

The p-values from testing each of the four claims using the correct testing method are given in the
table below.
Table Q3a
CEO observation p-value
Obs1 0.00
Obs2 0.08
Obs3 0.03
Obs4 0.18
d) Based on the information in Table Q3a, which of the four observations made by the CEO
can be confirmed at 5% significance level? [6 marks]
[20 marks in total]

You might also like