Quantitative Methods 2: ECON 20003

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Quantitative Methods 2

ECON 20003

WEEK 1
INTRODUCTION AND GENERAL INFORMATION
ABOUT QUANTITATIVE METHODS 2 (QM2)
ESTIMATION AND HYPOTHESIS TESTING OF A
POPULATION MEAN
Reference:
SSK: § 9.3-9.4, 10.1-10.3, 10.5, 12.1-12.4
WWL: Ch 1-2, § 3.1-3.6, 3.8-3.9
Dr László Kónya
January 2020
INTRODUCTION AND GENERAL INFORMATION

Subject Coordinator
and Lecturer: Dr László Kónya
Office: 460, 4th floor, FBE Building
Phone: 8344 0355; Email: [email protected]
Consultation hours: WED 1:00-2:00, FRI 11:00-12:00
(or by appointment)

Lectures: There are two one-hour lectures a week and


two equivalent lecture streams.
(1) WED 11:00 – 12:00 (The Spot-B01, Copland Theatre)
FRI 9:00 – 10:00 (Law-GM15, D.P. Derham Theatre)
(2) WED 3:15 – 4:15 (Medical-C216, Sunderland Theatre)
FRI 2:15 – 3:15 (Law-GM15, D.P. Derham Theatre)

The lectures will be recorded and made available on LMS.


L. Kónya, 2020 UoM, ECON 20003, Week 1 2
Tutorials: Start in week 1.
Sign up for one and only one tutorial class by the end of
week 1 via the Student Portal.

Subject website: LMS


The subject guide, some review resources, the lecture
notes and the tutorial materials can be downloaded
from this website in due time.
Also, time to time, important messages might be
uploaded onto the subject website, so visit it regularly.

Texts: In this subject you will learn topics from intermediate


statistics and introductory econometrics alike.
For this reason, I cannot endorse a single textbook that
covers all topics. Instead, we have a prescribed text
and a recommended text.
Although the lecture notes are fairly detailed, they are
not meant to substitute for these books.
L. Kónya, 2020 UoM, ECON 20003, Week 1 3
Prescribed text (WWL): Recommended text (SSK):
Wooldridge, J.M. et al. (2017): Selvanathan, E.A. et al. (2017):
Introductory Econometrics Asia- Business Statistics, 7th edition,
Pacific Edition, 1st edition, Cengage Learning Australia.
Cengage Learning Australia.

These books can be purchased online. See the links in the Subject Guide.
L. Kónya, 2020 UoM, ECON 20003, Week 1 4
Software: R programming language and software environment for
statistical computing and graphics. It is not a menu
driven program, you need to learn to write R codes to
operate it.

R is a fully functional standalone program, but RStudio,


an integrated development environment for R, can
assist in writing, compiling, debugging and executing R
codes.

Both programs are free. R can be downloaded from the


Comprehensive R Archive Network (CRAN) website
(https://cran.r-project.org/), while RStudio is available at
https://www.rstudio.com/products/rstudio/download/.

You will need to use these programs to complete the


assignments.

Your knowledge on how to operate these programs will


not be tested on the exam, but some exam questions
will be based on R printouts.
L. Kónya, 2020 UoM, ECON 20003, Week 1 5
Prerequisites:
The main subject pre-requisite for Quantitative Methods 2 is Quantitative
Methods 1 (ECON10005), the first year quantitative subject taught in the
Faculty of Business and Economics, or an equivalent subject taught at
other faculties of the university or at another tertiary institution.
QM2 students are supposed to master the topics discussed in Chapters
1-13, 17-18 of the SSK book. Namely,
Graphical descriptive techniques and numerical descriptive statistics;
Joint, marginal and conditional probability;
Random variables and probability distributions (binomial, uniform,
normal, t);
Sampling distribution, estimation and hypothesis testing of one
population mean and the difference between two population means;
Simple linear regression and correlation.

Some, but not all, of these topics are reviewed in the Revision slides,
which will be uploaded to LMS in due time. You might also wish to look at
https://creativemaths.net/blog/videos-for-teaching-and-learning-statistics/.
L. Kónya, 2020 UoM, ECON 20003, Week 1 6
In QM2 we have four aims:
• To learn quantitative skills that are essential in the majority of jobs that
business, commerce and economics graduates obtain.
• To implement these skills using examples from accounting, management,
marketing, economics and finance.
• To gain experience implementing the techniques using the R statistical
software and RStudio.
• The focus is on choosing the appropriate technique for each problem,
implementing it correctly, and interpreting the results.

Subject overview:
Building on QM1 or some equivalent prerequisite unit, QM2 is structured
as follows:
1) Statistical inference about one, two or more populations with
parametric and nonparametric techniques (6 weeks)
2) Regression analysis with cross-sectional data (4 weeks)
3) Regression analysis with time series data (2 weeks)
L. Kónya, 2020 UoM, ECON 20003, Week 1 7
Assessment: Tutorial attendance and participation 10%
Three assignments 15%
Mid-semester test 5%
Final exam (2-hour) 70%

Tutorial attendance and participation:


You can gain 10% credit for
i. attending at least 10 tutorials from week 2 onwards,
ii. actively attempting the tutorial exercises on the tutorials, and
iii. presenting evidence of completing the previous week ‘homework’
exercises to your tutor on request.
During each tutorial class students work through the tutorial handout
individually in their own pace. The tutorial handouts are self-explanatory
and sufficiently detailed and the tutors provide assistance if needed.

Three assignments:
There will be three assignments for 5% credit each.
i. Online submission via LMS.
ii. Students can work alone or in a group of two (not three or four …).
L. Kónya, 2020 UoM, ECON 20003, Week 1 8
iii. Students in a group must submit a single copy of their assignment
and will get the same assignment marks.
iv. No late assignments are accepted and no extensions will be given.

Mid-semester test:
There will be an online mid-semester test for 5%.
i. The test will be held during week 7 of the semester.
ii. Students can undertake the test online via LMS at any time of their
choosing between 10 a.m. on Tuesday 21 April and 4 p.m. on
Thursday 23 April.
iii. The test will cover the material presented during lectures up to the
end of week 5 and in the tutorials up to the end of week 6.
iv. The test will consist of 10 multiple choice questions and 5 true or
false questions.
v. There is a time limit of 30 minutes to complete the test.
vi. To complete the test successfully students will need to have critical
value tables for each of the distributions covered during the lectures
and a calculator.
L. Kónya, 2020 UoM, ECON 20003, Week 1 9
Note: Students who loose some internal (i.e. tutorial attendance, assignment
or mid-semester test) marks for valid reasons can apply for special
consideration to get the lost marks transferred to the final exam.

Final exam at the end of the semester:


It is worth 70% of the final grade for this subject.
i. It will be a 2-hour exam during the University's normal end of
semester assessment period, with the time, date and location
provided by the University's administration later in the semester.
ii. The exam will cover all the material covered during lectures and
tutorials throughout the semester. There will be no surprises. The
questions and tasks on the final exam will be similar in terms of
style and difficulty to those in the tutorial problem sets.
iii. A formula sheet and critical value tables for each distribution
required to complete the exam will be provided to students in the
exam package.
iv. On the exam students will neither be asked nor tested on how to
use R / RStudio, but they will have to interpret R printouts.
L. Kónya, 2020 UoM, ECON 20003, Week 1 10
v. Students will need to bring an approved calculator (Casio FX82
with any suffix) into the exam. Make sure that you can use it
efficiently (you can download the manual from LMS).
vi. You must pass the exam, i.e. to achieve 50% of the total exam
mark, to successfully complete the subject.
vii. Supplementary exam will not be provided in case of absence during
the examination period, unless it is due to serious illness or some
other legitimate reason. In those exceptional cases you might wish
to apply for special consideration.
(https://students.unimelb.edu.au/admin/special-consideration).

Note: Students affected by the novel coronavirus (COVID-19) outbreak


a) should apply for special consideration;
b) will be exempt from attending the tutorials, but they still need to sign up for
a tutorial class, to do the tutorial exercises and email their solutions with the
relevant R / RStudio program and printout to their tutors every week;
c) need to complete the assignments and the online mid-semester test;
d) can find additional relevant information at:
https://students.unimelb.edu.au/student-support/coronavirus.
L. Kónya, 2020 UoM, ECON 20003, Week 1 11
Educational research by Rose, C. and Goll, L. (“Accelerate your
learning”,1992) reveals, that
after 48 hours students on average retain:
• 10% of what they read,
• 20% of what they hear,
• 30% of what they see,
• 50% of what they see and hear,
• 70% of what they talk over with others,
• 80% of what they use and do in real life,
• 95% of what they teach someone else.

L. Kónya, 2020 UoM, ECON 20003, Week 1 12


So, if you wish to complete this unit successfully,

1) Attend the lectures and the tutorials regularly.


2) Take notes on the lectures and tutorials.
3) Read the relevant chapter(s) from the prescribed and recommended
textbooks every week.

4) Work on the tutorial problems every week.


5) Cooperate with your fellow students.
6) If you do not understand something, get help as soon as possible
before you miss too much – ask your tutor, your lecturer, or visit the
Discussion Board on LMS.

7) Repeatedly review what you learnt previously in order to reinforce


everything several times.

L. Kónya, 2020 UoM, ECON 20003, Week 1 13


DESCRIBING A SINGLE POPULATION:
ESTIMATION AND HYPOTHESIS TESTING OF A
POPULATION MEAN

• In QM1 you already learnt about how to describe a single population by


estimating its mean and by performing a hypothesis test on it.
You are supposed to be familiar with these topics, so in QM2 we just
briefly review them. If you need more assistance, study the Review 1
and Review 2 slides and the recommended texts.

• In case of random sampling, any statistic is a random variable because


it is a function of some randomly selected sample items.

The probability distribution of all possible values of a statistic generated


by random samples of the same size is called the sampling distribution
of the given statistic.
Like probability distributions in general, sampling distributions can be
characterized by their mean, variance (standard deviation), and shape.
L. Kónya, 2020 UoM, ECON 20003, Week 1 14
SAMPLING DISTRIBUTION OF THE SAMPLE MEAN

• Consider a random sample drawn from population X : ( ; ).

(identically and independently distributed)

The point estimator of the population mean () is the sample mean:

It has the following properties.

i)

The expected value of the sample mean is equal to the


population mean.
L. Kónya, 2020 UoM, ECON 20003, Week 1 15
ii)

The variance of the sample mean is equal to the


population variance divided by the sample size
(assuming, that the sampled population is infinitely large,
or is finite but sampling is with replacement).

The standard deviation of a statistic is called its standard error.

The standard error of the sample mean (just


iii) like its variance) is a decreasing function of the
sample size.

iv) Shape / form of the sampling distribution:


a) If the sampled population is normally distributed, X : N( ; ),
then the sample mean is also normally distributed.
b) If the sampled population is not normal but n  30, then
according to the Central Limit Theorem (CLT), the sample
mean is still approximately normally distributed.
L. Kónya, 2020 UoM, ECON 20003, Week 1 16
ESTIMATING THE POPULATION MEAN

• Assume that we repeatedly draw random samples of the same size from
a normally distributed population, or that the sampled population is not
normal but we draw reasonably large samples and thus the CLT holds.

If z/2 denotes the (1- /2)100% percentile of the standard normal


distribution, that is P(Z > z/2) = /2, then

Ex ante (before drawing a sample) Ex post (after drawing a sample)

Unknown X-bar is an estimator, while x-bar


constant is an estimate, i.e. a particular
number, so this is not a random
Random interval limits interval and it either contains  or
(even if n,  and  are fixed) not.
L. Kónya, 2020 UoM, ECON 20003, Week 1 17
When  is unknown but the sample size is large enough to estimate 
satisfactorily, we can replace  with its estimate s to obtain an estimate
of the standard error of the sample mean,

When  is known and the sample mean is at least approximately


normally distributed, or  is unknown and the sampled population is at
least approximately normal, the (1-)100% confidence interval
estimate of the population mean is given by

if  is known and X-bar ~ N


if  is unknown and X ~ N

Note: Commenting on these intervals we cannot talk about ‘probability’.


Instead we use the term ‘confidence’ as we have (1-) degree of
confidence that the single interval obtained from the sample at hand is
not extreme but indeed contains .
L. Kónya, 2020 UoM, ECON 20003, Week 1 18
Ex 1: (Selvanathan et al., p. 397. ex. 10.50)
A NSW Department of Consumer Affairs officer responsible for enforcing laws
concerning weights and measures routinely inspects containers to determine if
the contents of 10kg bags of potatoes weigh at least 10kg as advertised on the
container. A random sample of 25 bags whose container claims that the net
weight is 10kg yielded the following statistics: x-bar =10.52, s2 =1.43. Estimate
with 95% confidence the mean weight of a bag of potatoes. Assume that the
weights of 10kg bags of potatoes are normally distributed.
Let X denote the weight of a bag of potatoes. We do not know its population
mean and standard deviation, but we are told that it is normally distributed.
Hence, we can develop the 95% confidence interval using

From the sample

From Selvanathan et al. Appendix B, p.1076, tα/2,df=n-1 = t0.025,24 = 2.064.

With 95% confidence the mean weight of a bag of potatoes is


somewhere between 10.02kg and 11.02kg.
L. Kónya, 2020 UoM, ECON 20003, Week 1 19
TESTING THE POPULATION MEAN

• There are two types of statistical inference.

Estimation Hypothesis testing

Point estimation Interval estimation

• Hypothesis testing, in general, is a six-step procedure:

1) Set up the null and alternative hypotheses;


2) Determine the test statistic and its sampling distribution;
3) Specify the significance level;
4) Define the decision rule;
5) Take a sample and calculate the value of the test statistic;
6) Make a statistical decision and draw the conclusion.
L. Kónya, 2020 UoM, ECON 20003, Week 1 20
The details are discussed on the Review 2 slides, here we just consider
an illustrative example.

Ex 2: (Selvanathan et al., p. 501, ex. 12.53)


A diet doctor claims that the average Australian is more than 10kg overweight.
To test his claim, a random sample of 100 Australians were weighed, and the
difference between their actual weight and their ideal weight was calculated and
recorded.
a) Do the data allow us to infer at the 5% significance level that the doctor’s
claim is true?

i. Let diff denote the difference between actual weight and ideal weight (kg).

HA :  > 10 and H0 :  = 10

ii. Since n is 100, we can rely on CLT. However,  is unknown, so let’s


assume that the population of diff is not extremely non-normal.
The test statistic is t.

L. Kónya, 2020 UoM, ECON 20003, Week 1 21


iii. The significance level is given,  = 0.05.

iv. This is a right-tail test, so the entire rejection region is located under the
right tail of the sampling distribution.
Reject H0 if the value of the test statistic calculated from the sample
is greater than tα,df=n-1 = t0.05,99  t0.05,100 = 1.660.

v. The sample mean and standard deviation are 12.175 and 7.898,
respectively.

vi. Since tobs = 2.7539 > 1.660 = tα, we reject H0. Hence, at  = 0.05 there is
enough evidence to conclude that the diet doctor is right, the average
Australian is more than 10kg overweight.

L. Kónya, 2020 UoM, ECON 20003, Week 1 22


b) Find the p-value of the test. What does it suggest?

The t-table does not show the exact p-value. However, it is certainly smaller
than 0.005, so H0 can be rejected even at the 0.5% significance level.

c) Perform the test in part (a) with R this time.


You will learn in the tutorials how to import the data from an Excel file and how
to run the t-test with R/RViews. The t.test(diff, mu = 10, alternative = "greater")
command returns the following printout:

R reports the test statistic (t), the


degrees of freedom, the p-value,
the alternative hypothesis, the
95% ‘one-sided’ confidence
interval (do not worry about it)
and the sample mean.

Check whether R performed the required test (i.e. a right-tail t-test this time)
and whether the p-value <  = 0.05. Since p-value  0.0035, we reject H0.
L. Kónya, 2020 UoM, ECON 20003, Week 1 23
• The (single-sample) Z / t test and the corresponding confidence interval
for a population mean are based on the following assumptions:

i. The variable of interest is quantitative measured on an interval or


ratio scale (see Review 1, slides #7, 8);
ii. The sample is a random sample (see Review 1, slide #3);
iii. Either (Z test) the population standard deviation, , is known and
the sample mean is at least approximately normally distributed
(because the sampled population itself is normally distributed or
the sample size is large and thus CLT holds),
or (t-test)  is unknown but the sampled population is normally
distributed (at least approximately).

L. Kónya, 2020 UoM, ECON 20003, Week 1 24


WHAT SHOULD YOU KNOW?
• Important definitions and concepts, like
population, sample, parameter, statistic, descriptive statistics,
inferential statistics, sampling error, non-sampling error, types of
data/variable, measurement scales, estimator, estimate, etc.
• To compute normal probabilities. To use the standard normal and t
tables (Tables 3 and 4 in Appendix B of Selvanathan et al., 2017).
• The sampling distribution of the sample mean.
• To estimate a population mean with a single value and with a confidence
interval.
• To carry out the six steps of hypothesis testing and to apply this
procedure to testing a population mean.
• To estimate a population mean and to run a test for a population mean
with R/RStudio.
• To be able to use the Casio FX82 calculator efficiently.
L. Kónya, 2020 UoM, ECON 20003, Week 1 25

You might also like