STT100

Hypothesis Testing
Assignment Help Lecture
WEEK 7
Trimester 2 2020
0nline
 Hypothesis Testing
Data Bases
And
Data collection
To be used to
CONDUCT HYPOTHESIS TESTING

IS THE SAMPLE PROPORTION OR MEAN

SIGNIFICANTLY DIFFERENT THAN THE POPULATION
 Let us talk about 2. Conduct a Hypothesis test following the STEPS:

assignments
•State the Null hypothesis (Ho )and the Alternative
hypothesis (HA ).
•Choose the level of significance. All of you will choose
95%.
•Find the critical value. Since you are using a 95% level of
Hypothesis significance, for a proportion this will be using ‘z‘ =1.96
(see the z table) and for a sample mean this will be using ‘t‘
testing and a sample of 10 t= 1.833 (see the t table) .
•Find the test statistic.
You do this using the appropriate formulae
1. Select one of the areas
of investigation below. •Draw a conclusion. If your test statistic is greater than the
Or choose one of your critical value reject the null, your sample is different to the
population (recommended).
own
3. What does your result indicate?
STATE CLEARLY YOUR Analyse this in economic educational or social
HYPOTHESIS terms.
HYPOTHESIS TESTING
Details and
PROCEEDURES
1. The null hypothesis Ho is what you believe to be true and is always Ho =
Eg equal to the population mean that you have found from the ABS or Other
Statistics.
The alternative hypothesis is what you claim to be or believe to be, Ha or H1.
The null hypothesis can be:
Ha ≠ 0…......this is a 2 tail test

If Ha < 0 or

Ha > 0… this is a one tail test

WHEN TO USE WHICH FORMULAE TO FIND THE TEST STATISTIC

1.
Use when:
Use when:
• A percentage is
• Standard deviation of
given
the population is given
• There are only 2
• “z” when n> or = 30 and
the population standard 2. possible
outcomes eg Yes
3. deviation is known
or No
Or “t”
“t” when n < 30 and population standard
deviation is not known.
So you use the sample standard deviation,
 Terms
• One tail test
• Two tail test
• Null Hypothesis • Type I Error
• Alternative Hypothesis • Type II Error
• Critical Value • Test Statistic
• Level of significance • Rejection zone
• Point estimate
• interval estimate
Alternative Hypothesis two tail test rejection region

type 11 Error confidence interval

type 1 Error
one tail test Critical value
level of significance

Point estimate null hypothesis

Test Statistic
 Critical value or the level of significance DEFINITIONS
 • The cut off point between the null and alternative hypothesis in
either a one or two tail test. Its value indicates the area under the
normal curve in the tails (0.05) of the distribution or the area under
the curve excluding the tails (0.95). Also known as two tail test
ONE tail test  • A statistical test using the normal curve that is
interested in finding if the sample deviates from the
• A statistical test using the normal curve that population in both directions, if the sample is either =
is interested in finding if the sample or the population when testing a hypothesis. Ho = 0 or
deviates from the population in one Ha 0
direction, if the sample is either > or < than This TWO tail tests if your answer to the
the population when testing a hypothesis.
test statistic is either < OR > the
Ho = 0 or Ha > 0
normal. Generally is means that it is not
This ONE tail tests if your answer the same as the normal data

to the test statistic is < the normal

This ONE tail tests if your answer

to the test statistic is > the normal
 DEFINITIONS…cont
Critical value
• It is found by looking up a z or t table based on the level of significance and in the case of t the
sample size. It divides the rejection zone from the non rejection zone when testing a hypothesis.

p value
• The area in the tail beyond the test statistic. The area represents the probability of finding
another test statistic, using a different sample. For the model to be reliable we want the p value
to be LOW.

type 1 Error type 11 Error

• When we erroneously(wrongly) reject the null and • When we erroneously (wrongly) accept the null
accept the alternative when in fact we should when in fact we should reject the null and accept
accept the null. Its different to normal but we say the alternative. It is not different to normal but we
that it is not different say that it is different.
Both these can be avoided using a larger sample “n” ie. increasing the evidence and by using a higher level of
significance (95 instead of 95%)
 DEFINITIONS…cont

Test Statistic
• This statistic is found by using the appropriate z or t formulae. Its result will be used to reject or
accept the null hypothesis because th answer is > or < the critical value which is the cut off point
between the reject or accept zone on the normal curve.

level of significance
• This determines if the sample or data that you are testing is significantly different or not from the
normal population data. This is either 95% or 99% which sets the area under the curve where we
accept the null and 1 – 0.95 = 0.05 or 5% or 1 – 0.99= 0.01 or 1% is the area under the curve
which sets the reject the null zone.

• This Value is measured in standard deviations which

is found after setting the level of significance. Its
Critical value value is found by looking up its value in a z table or
t table.
 DEFINITIONS…cont

• Used to estimate a confidence interval. A formulae Alternative Hypothesis

is used which gives two values (+ and - between
which we are confident that our answer is between This is the hypothesis that the
on the normal curve. If it is between < + 5% and > sample ,or country or person IS
-5% on the normal curve then we are 95%
significantly different to the
confident if between < + 1% and > -1% on the
normal curve then we are 99% confident average. IT IS UNUSUAL OR
ABNORMAL

NULL Hypothesis
Test Statistic
This is the hypothesis
Found by using a test statistic formulae which gives that the sample or
ONE point measured as a z score in standard
deviations. if this point > critical value reject the null country or person IS NOT
if < the critical value then accept the null significantly different to
the average. IT IS NOT
UNUSUAL or ABNORMAL
 Level of significance
95% significantly different (reject
Accept null zone null) or not significantly Reject null zone accept the
If the answer to the test different (accept null. alternative
statistic formulae is in this If the answer to the test
region then accept the null 5% statistic formulae is in this
hypothesis. The sample is region then reject the null
not significantly different hypothesis. The sample is
significantly different
Test statistic Critical Value
Found by looking up z table Test statistic
Found by using the test statistic Found by using the test statistic
formulae if the answer is < = 1.645 = Z cut off point
between accept null and reject formulae if the answer is >
critical value then accept null critical value then reject the null
null
 OPTION 1
Examples of possible
sample PROPORTIONS
Use when:
• A percentage is
given
• There are only 2
2. possible
outcomes eg Yes
or No
See other possible examples at the
end of this section
 STEPS TO FOLLOW TO EXAMPLE CALCULATION: Assume the sample proportion of brown
COMPLETE AN HYOTHESIS eyes is 90% or 0.9 and the population proportion of brown eyes is
TEST 55% 0r 0.55
 Z =
Use the following to calculate the test
STEPS: statistic
•State the Null hypothesis (Ho )and the Alternative
hypothesis (HA ). Population
Sample  Z =
•Choose the level of significance. All of you will
proportion proportion
choose 95%.
•Find the critical value. Since you are using a 95%
level of significance, for a proportion this will be using
‘z‘ =1.96 (see the z table) and for a sample mean this
 Z =
will be using ‘t‘ and a sample of 10 t= 1.833 (see the t
table) .
•Find the test statistic. Population 1 -Population proportion
 Z =
You do this using the formulae= proportion Sample size

•Draw a conclusion. If your test statistic is greater

Z= 2.2247
than the critical value reject the null, your sample is Z= 2.2247
different to the population (recommended). So reject the null and conclude
5. Draw conclusions that the sample is significantly Alternative
different to the population at Hypothesis ZONE
Null Hypothesis ZONE Accept H0 the 95% significance level
Reject H0
If test statistic in this zone
If test statistic in this zone
That is your answer to the equation is LESS than 1.645

That is your answer to the

1.645
OPTION 2

Examples of possible
Sample MEAN and sample
Standard Deviation
“t” when n < 30 and population standard
deviation is not known.
So you use the sample standard deviation,
STEPS:
EXAMPLE CALCULATION: Assume the sample mean average Australian man
•State the Null hypothesis (Ho )and the Alternative (18 years and over) was 175.6 cm tall and from your sample of 10 PIA
hypothesis (HA ). students you find the mean sample height to be 160cm with a SD of 10cm
•Choose the level of significance. All of you will choose Use the following to calculate the test
95%. statistic  t =
•Find the critical value. Since you are using a 95% level
of significance, for a proportion this will be using ‘z‘ Sample Population
=1.96 (see the z table) and for a sample mean this will mean mean
be using ‘t‘ and a sample of 10 t= 1.833 (see the t table)  t =
.
•Find the test statistic.
You do this using the formulae=   t=
Sample
Note it is minus
standard and the curve is to
deviation Sample size T = -5.55 the left because
•Draw a conclusion. If your test statistic is greater than Z= -5.55 the sample is less
So reject the null and conclude than the
the critical value reject the null, your sample is different that the sample is significantly
to the population (recommended). population
different to the population at the
5. Draw conclusions 95% significance level and below Alternative
-1.833 the critical value Hypothesis ZONE
Null Hypothesis ZONE Accept H0 Reject H0
If test statistic in this zone If test statistic in this zone

That is your answer to the equation is LESS than- 1.833

That is your answer to the

OPTION 3
Examples of possible
POPULATION MEAN and
Population Standard
Deviation • Standard deviation of
Use when:

the population is given

• “z” when n> or = 30 and
the population standard
deviation is known
 Two Possible Types of Error

• Decision making is never perfect and mistakes can

be made

– Type I error: rejecting the null when it is true

– Type II error: accepting the null when it is false
Choice of Significance Level

• Why 5%?
• Like its complement, the 95% confidence level, it
is a convention. A different value can be chosen
• If the cost of making a Type I error is especially
high, then set a lower significance level, e.g. 1%.
The significance level is the probability of making a
Type I error