Introduction to Hypothesis
testing Statistical Inference
about Means and Proportions
Senior Lectures by:
RESEARCH
METHOD Formulating the HYPOTHESIS
Test of the Hypothesis Statement of the PROBLEM
1. Recognition of the FACTS 2. Discovery of the Problem
3. Problem Formulation
1. Design Test
What is a Hypothesis?
Hypothesis Testing for One Population Value:
–
Population mean
(
a. (population standard deviation) is given (known):
Use z/standard normal/bell shaped distribution
b. (pop std dev) is not given but s (sample std dev) is given
Use student’s t distribution
–
Population proportion (
)
–
Population Variance (
Use 2 (Chi-Square) distribution. Population Standard Deviation =
Example: The mean monthly cell phone bill the student of AUCA is = 10000 Rwf
Example: The proportion of the students of AUCA with cell phones is p = .80 Use z/standard normal/bell shaped distribution
Hypothesis
Hypothesis: A statistical hypothesis is a statement on a
probabilistic model and a hypothesis test is a method to determine
the possibility of that statement based on a sample.
Presumptions thus often provide the occasion for an
investigation. For this reason it is called research hypothesis.
I a
ssume
the m
ean
M
onthly
cell ph
one bill
the stu
dents o
f
AUCA
Purpose of hypothesis testing
•
The purpose of hypothesis testing is to
determine whether there is enough statistical
evidence in favor of a certain belief about a
parameter.
Example
: Is there statistical
evidence in a random
sample of potential
customers, that support the
hypothesis that more than
10% of the potential
• States the assumption (numerical) to be tested. This
hypothesis is assumed to be true, and the collected data
will be analyzed to see
if it is contradictory to the null
hypothesis.
Research Hypothesis
“
The mean monthly cell phone bill the student of AUCA is
less than 10000 Rwf”
Example of the Null Hypothesis:
The mean monthly cell phone bill the student of AUCA is at least ten thousand Rwf.
H0: 10000
• Always contains “=” , “≤” or “” sig • May or may not be rejected
The Alternative Hypothesis, H
a
or
H
1
• Is the opposite of the null hypothesis
Example
– The mean monthly cell phone bill the student of
AUCA is less than 10000 Rwf
( H
a:
< 10000)
• Never contains the “=” , “≤” or “
” sign
• May or may not be accepted
• Is generally the hypothesis that is believed (or needs
to be supported) by the researcher.
This is
what
you w
ant to
Examples: Give the null hypothesis and the alternative
hypothesis
• Is there statistical evidence in a random sample of potential
customers, that support the hypothesis that more than 10% of the
potential customers will purchase a new products?
• You want to show that people find the new design for a recliner
chair more comfortable than the old design.
• You are trying to show that cigarette smoke has an effect on the
quality of a person’s life.
• The mean age of the students enrolled in evening classes at a
certain college is greater than 36 years.
• The mean weight of packages shipped on Air Express during the
past month was less than 36.7 lb.
The critical concepts are these:
1. There are two hypotheses: the null and the alternative hypotheses.
2. The procedure begins with the assumption that the null hypothesis
is true.
3. The goal is to determine whether there is enough evidence to infer
that the alternative hypothesis is true, or the null is not likely to be
true.
4. There are two possible decisions:
Reject the null: To conclude that there is enough evidence to
infer that the alternative hypothesis is true.
Fail to reject the null: To conclude that there is insufficient
evidence to support the alternative hypothesis.
Claim:
the mean life
expectancy in Africa is
over than 50.1 years
is 60: x = 60 years
Is X=60 likely if Ho:
≤ 50.1
REJECT
Null Hypothesis
If not likely,
Hypothesis Testing Process
Suppose the
sample mean of the Life expectancy H0: ≤ 50.1
Ha: > 50.1
Sample
Sample
Population
Sampling Distribution of x
≤ 50.1
If H0 is true
If it is unlikely that we
would get a sample
mean of this value ...
... then we reject the null
hypothesis that
≤
50.1
... if in fact this were the population mean…
x=60
Reject H0 Do not reject H0
Level of Significance,
a
• Defines unlikely values of sample statistic if null hypothesis is true
– Defines rejection region of the sampling distribution. • Is designated by a , (level of significance)
– Typical values are .01, .05, or .10
• Is selected by the researcher at the beginning. • Provides the critical value(s) of the test .
a
Normal
distribution
If Alpha equals:
0.1
0.05
0.01
One tail
Critical region
2.33
1.64
1.28
Two-tailed
H
0: μ ≥ 50.1
H
a: μ < 50.1
0
H
0:
μ≤ 50.1
H
a:
μ> 50.1
H
0:
μ= 50.1
H
a:
μ≠ 50.1
a
a
/2
Represents critical value Lower tail test
Level of significance =
a
0
0
a
/2
a
Upper tail testTwo-tailed test
Rejection
region is
shaded
Interpreting the p-value…
The smaller the p-value, the more statistical evidence exists
to support the alternative hypothesis.
•
If the p-value is
less than 1%,
there is
overwhelming
evidence
that supports the alternative hypothesis.
•
If the p-value is
between 1% and 5%,
there is a
strong
evidence
that supports the alternative hypothesis.
•
If the p-value is
between 5% and 10%
there is a
weak
evidence
that supports the alternative hypothesis.
•
If the p-value
exceeds 10%,
there is
no evidence
that
Interpreting the p-value…
Overwhelming
Evidence
(Highly
Significant)
Strong Evidence
(Significant)
Weak Evidence
(Not Significant)
No Evidence
(Not Significant)
Actual
situation
Our
decision
Null (Ho)
hypothesis is
false
Null (Ho)
hypothesis is
true
Reject the null
(Ho) hypothesis
Correct
decision
Type I
error (α)
Called Level of Significance
Do not reject the
null (Ho)
hypothesis
Type II
error (β)
Correct
decision
(1-β)
Conclusions of a Test of Hypothesis…
If we reject the null hypothesis, we conclude
that there is enough evidence to infer that the
alternative hypothesis is true.
If we fail to reject the null hypothesis, we
conclude that there is not enough statistical
evidence to infer that the alternative
hypothesis is true.
This does not mean that we
n
s
μ
x
t
n1
The test statistic is:
Using Small Samples (The
population must be approximately
normal)
Review: Steps in Hypothesis Testing
• Specify the population value of interest.
•
Assumptions
: Randomization, quantitative variable,
normal population distribution (robustness?)
• Formulate the appropriate null and
alternative
hypotheses.
• Specify the desired level of significance.
• Determine the rejection region or p_value
Example: Lower Tail z Test for Mean
Test the claim that the true mean monthly cellphone bill the
student of AUCA is less than ten thousand Rwf.
Assume σ = 800
1. Specify the population value of interest
Mean monthly cell phone bill the student of AUCA
2. Formulate the appropriate null and alternative hypotheses
Ho: μ 10000 Ha: μ < 10000 (This is a lower tail test)
3. Specify the desired level of significance
Reject H0 Do not reject H0
4. Determine the rejection region
a
= .05
-zα= -1.64 0
This is a one-tailed test with a = .05
Since σ is known, the cut off value is a z value:
Reject H0 if z < za = -1.64 ; otherwise do not reject H0
En Excel: insert function(fx)<select category Statistical<select function (NORMSINV)<PROBABILITY Write the value (0.05) =1.64
5. Obtain sample evidence and compute the test statistic
Suppose a sample is taken with the following results:
n = 20, = 9500
(
= 800 is assumed known)
Then the test statistic is:
X
Example: Lower Tail z Test for Mean
80
.
2
795
.
2
178.89
500
20
800
10000
9500
n
σ
μ
x
Reject H0 Do not reject H0
a
= .05
0
Z=-2.80
The Z value (Z = -2.80)
is in the rejection region.
We can reject the null
hypothesis in favor of
Ha.
z =-1.64
p-value =.00255
En Excel: insert function(fx)<select category Statistical<select function (NORMSDISTR) Write the value (-2.80) =0.00255
a
We can conclude that the p-value (0.00255) < a = .05
There is evidence that supports the alternative hypothesis, ie the monthly cellphone bill is less than ten thousand Rwf
6. Reach a decision and interpret the result
•
Compare the p-value with
a
–
If p-value <
a
, reject H
0–
If p-value
a
, do not reject H
0Here: p-value = .00255 a = .05
Since .00255 < .05, we reject the null hypothesis
p-value =.00255
a
= .05
-2.80
-1.64
p-value example
Obtain the p-value from a table or computer
Compare the p-value with a
Reject H0
Example: Upper Tail z Test for Mean (
Known)
H
0: μ ≤ 55 the average is not over S/55
per month
H
a: μ > 55 the average
is
greater than
S/55 per month.
Form hypothesis test:
A manager of mobile telephony in Peru states that the
monthly phone bill has increased more than 55 soles per
month. The company wishes to test this claim.
(Assume
= 25 is known)
Reject H0 Do not reject H0
•
Suppose that
a
= .10 is chosen for this test
Find the rejection region:
a= .10
z
α=1.28
0
Reject H
0Reject H0 if z > 1.28
En Excel: insert function(fx)<select category Statistical<select function (NORMSINV) <PROBABILITY. Write the value (0.10) =1.28
Obtain sample evidence and compute the test
statistic
Suppose a sample is taken with the following
results:
n = 80, x = 59
(
=25 was assumed known)
–
Then the test statistic is:
43
.
1
80
25
55
59
n
σ
μ
x
z
Reject H0 Do not reject H0
Example: Decision
a= .10
1.28
0
Reject H
0Reject H0 since z = 1.43 1.28
i.e.: there is enough evidence to say that the monthly phone bill has increased in Peru
z = 1.43
Reject H0
a= .10
Do not reject H0
1.28
0
Reject H0
z = 1.43
Calculate the p-value and compare to
a
p
-Value Solution
Reject H
0since p-value = .0764<
a
= .10
En Excel: insert function(fx)<select category Statistical<select function
Example: Two-Tail Test (
Unknown)
The average cost per room per night for hotels Serena in Africa
is said to be $200 per night. A random sample of 20 hotels
resulted in x= $250 and s = $115 Test at the
a
= 0.05 level.
(Assume the population distribution is normal)H
0:
μ
= 200
Since t = 1.94 is not greater than 2.093, and nor
less than -2.093, we cannot the null hypothesis in
favor of Ha. That is there is insufficient evidence
that true mean cost is different than $200
Reject H0 Reject H0
a
/2=.025
-t
α/2=Do not reject H0
0
t
α/2=a
/2=.025
-2.093 2.093 1.94 20 115 200 250 n s μ xt201
1.94
H
0:
μ
= 200
H
a:
μ
200
Example Solution: Two-Tail Test
En Excel: insert function(fx)<select category Statistical<select function (TINV) > Ok Write in probability (0.05) and in Deg_freedom (19) > OK. Then returns the inverse of
a= 0.05
n = 20
is unknown, so use a t statistic
Critical Value:
t19 = ± 2.093
P_value = 0.0500
Example:
A pharmaceutical company conducts research on the efficacy of a vaccine against measles. The variable considered is the antibody titers produced by the vaccine. The vaccine
produced by another laboratory reports an average titer of antibodies 1.9. To test whether the new vaccine is more effective than the older vaccine, the shot was given to 16 volunteers and obtained the following results:
Average titer of antibody 3 2.5 2.4 1.9 1.8 1.5 2.6 2.7 3.1 1.7 2.3 2.2
Steps using SPSS
To do this, click on Analyze, and then
Compare means
followed by One-Sample T test
SPSS Report (recommendation: work made by your hand and
checks with the same results).
One-Sample Statistics
N Mean Std. Deviation Std. Error Mean Average titer of
antibody 16 2.225 0.5183 0.1296
One-Sample Test
Test Value = 1.9
t df
Sig.
(2-tailed) Mean Difference
95% Confidence Interval of the Difference
Lower Upper Average titer
of antibody 2.508 15 0.024 0.325 0.049 0.601
Note 1: The value of p, or Sig gives us the SPSS default is bilateral, unilateral if we value: Sig /2 (.024/2 = 0.012)
Interpretation: This result indicates that the data are consistent with an average value greater than 1.9, because the difference found is highly significant
Hypothesis
Tests for Proportions
• Involves categorical values
• Two possible outcomes
– “Success” (possesses a certain characteristic)
– “Failure” (does not possesses that characteristic)
• Fraction or proportion of population in the “success” category is
denoted by p
• Sample proportion in the success category is denoted by
• When both np and n(1-p) are at least 5, can be approximated
by a normal distribution with mean and standard deviation.
pˆ
size
sample
sample
in
successes
of
number
ˆ
n
x
p
p
μ
pˆ
n
p)
p(1
σ
pˆ
Example: z Test for Proportion
Check:
n p = (144)(.80) = 115.2
n(1-p) = (144)(.20) = 28.8
A year ago the proportion of
students who had Cellular
was 80%. It is believed that
this
proportion
has
increased this year to check
this hypothesis, take a
random sample of 144
students and it was found
that 89% had cellular
Test at the
a
= .05
Z
Test for Proportion: Solution
a
= .05
n = 144, = =.89
Reject H0 at a = .05
H
0: p .80
H
a: p
>
.80
Critical Values: ± 1.64
Decision:
Conclusion:
There enough evidence to reject Ho, ie the number of students who have mobile phones has increased.
2.70
144
.80)
.80(1
.80
.89
n
p)
p(1
p
ˆ
z
p
pˆ
Reject H0a= .05
Do not reject
H0
1.64
0
Reject H0
p
-Value Solution
a= .05
Reject H0 Do not reject H0 1.64
0
Reject H
0z = 2.70
Reject H0 since p-value = .0035 < a = .05
En Excel: insert function(fx)<select category Statistical<select function
A marketing company claims that it receives =
4%
responses from its Mailing. To test this claim, We take a random sample of n = 500 were surveyed with x = 25 responses.Test at the a = .05 significance
Test for Proportion: Example
H
0: p = .04
H
a: p
.04
a
= .05
n
= 500, x = 25
= x/n = 25/500 = 0.05
pˆTest Statistics:
14
.
1
500
.04)
.04(1
.04
.05
n
p)
p(1
p
ˆ
z
p
Reject Reject .025 .025Critical Values: 1.96
Decision: Do no reject Ho at a = .05
Summary: Steps for a t test for a single sample
Restate the question as a research hypothesis and a null hypothesis about the populations.
Collect data.
Determine the characteristics of the comparison distribution. Choose Method (Z test or t test).
The mean or proportion is known of the population. Compute the standard deviation or proportion by: *Previous studies or research or
*Take a pilot sample or
*Calculate the variance of the distribution of means (S2 /n)
Take the square root, to get SE.
Note, we’re calculating t with N-1 df.
Determine the cut off sample score on the comparison distribution at which the null hypothesis should be rejected.
Decide on an alpha and one-tailed vs. two-tailed Look up the critical value in the table.
Determine your sample’s t score or Normal value distribution.
Decide whether to reject or not reject the null hypothesis. (If the observed value of “t” or “Z” exceeds the table value, reject.)