Statistical Inference:

Hypothesis Testing &

estimation

Week 8 lectures
Melissa Wos-Oxley
Aims for Week 8: Statistical Inference
1. Key concepts in statistical inference: 2 most common types
• Hypothesis testing (tests of significance)
• Estimation (Confidence Intervals)
2. Introduce key terms like null hypothesis, test statistics, p-values,
sampling distributions, SRS
3. Statistical inference for proportions

Moore et al. 2017 – Chapters 6.1-6.2; (revision in chapters 1.4 & 5), for statistical
inference for proportions – chapter 8.1
Summary of topic to-date and where to next?
Statistics:
✓ Types of variables (& data), concept of sampling, and graphing data
✓ Descriptive statistics & summarises, shape of data distributions (3 S’s, and outliers)
✓ Relationships between variables (explanatory vs response variables) weeks 1-5
✓ Correlation
✓ Regression
✓ Producing data: types of studies, data collection, errors
✓ Probability:
✓ Marginal, joint, conditional, concept of independence (week 6)
✓ Probability models for discrete and continuous random variables (week 7) weeks 6-7
✓ Binomial probability distributions
✓ Normal probability distribution (use the standard Normal Table)
➢ Hypothesis testing and estimation:
➢ Statistical Inference for proportions (weeks 8 & 9)
➢ Statistical Inference for numerical data (weeks 10 & 11) weeks 8-11
➢ t-distributions (another Table to use)
Rather than collecting every possible datum in a population, we take random samples
to represent that population and make inferences from these samples back to the
whole population.

We can now combine this idea with our knowledge of probability and draw
conclusions about a population by applying statistical inference, powerful methods
that use probability to give confidence in our conclusions.

Two most common types of statistical inference: hypothesis testing (tests of

significance) and estimation (like confidence intervals).
 QUESTION
(HYPOTHESIS)

COLLECT DATA
in the form of POPULATION
n replicates
population parameters

SAMPLE STATISTICS µ, σ, N, 𝑝
𝑥,ҧ s, 𝑝ො
with an assumed
“Normal sampling distribution” MAKE CONCLUSIONS
ABOUT POPULATION
ESTIMATION
point estimation &
confidence intervals HYPOTHESIS
TESTING
test-statistic &
p-value
Hypothesis testing:
• “tests of significance” used to assess the evidence provided by the data in favour of
some claim about the population parameters.
• A 6-step process: setting a hypothesis, applying a test and generating & interpreting
both a test statistic and p-value.

Estimation using Confidence Intervals (CI):

• equations used to estimate population parameters (like the population mean, µ, or
population proportion, p).
• e.g. determining where the “true” population mean likely sits with % confidence.
Why conduct a statistical test & point estimation for assessing differences between
sample means or proportions?

To seek for significant differences between samples (i.e. groups of replicate

datapoints), thus determining an effect/response by a treatment/factor. Then, to
determine where the true population mean or proportion likely sits.
➢Compare a measurement of a response variable from one sample to an expected value
➢Compare a measurement from one sample to a second sample taken from the same
subject or at the same location at two points in time. So that samples are related or
dependent
➢Compare measurements between two samples that are independent from each other
You are interested in water quality of the Torrens River; so you collect water from the upper reaches
which represent rural land-use and also from the lower reaches which represent urban land-use to
look for differences in nutrients (P & N).

The factor is land-use; where there are 2 samples [rural vs urban]; each nutrient is a dependent
variable; population is the units of water
You are interested in testing whether a new antibiotic will kill a certain bacterial strain; you grow
bacteria on agar plates and then add the antibiotic (at 1% concentration) to half of the plates, leaving
the other half of the plates free from antibiotic.

The factor is antibiotic use; where there are 2 samples [1% antibiotic use vs control]; counts of
bacterial colonies or zone of inhibition is the dependent variable; population is bacterial colonies
You are interested in knowing whether cigarette smoking changes the bacterial assemblages and
diversity of your nasal passages; you sample people that currently smoke and those that have never.

The factor is lifestyle habits; where there are 2 samples [smokers vs non-smokers]; diversity of
bacteria in the mouth; population is the human population
• When we “count” outcomes across categorical variables (like in binomial last week
classifying & counting “successes” and “failures”), where data is of counts or of
percents obtained from counts, then we are interested in making inferences about
population proportions (p).
➢ statistical inferences for proportions – z-test & CI for p

• When we “measure” variables and are interested in making inferences about

population means (µ).
➢ statistical inferences for numerical data – t-test & CI for µ
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
2. Choose an appropriate statistical test and calculate the test statistic
3. Determine the sampling distribution of the test statistic (from step 2)
4. Find the p-value associated with the test statistic
5. Make a decision to either “reject” or “accept” the null hypothesis (based on step 4)
6. State the conclusion in the context of the scientific problem -relate it back to the real
world (population!)
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
• Hypotheses are statements about the population parameter in question.
• A test of significance is intended to assess the evidence provided by data against a
NULL hypothesis (H0) in favour of an alternative hypothesis (H1 or HA).
• The “Null” is a statement that NO effect or NO difference is present.
• The “Alternative” is a statement that there is an effect or difference (two-sided
alternative) or an effect or difference in a specific direction (one-sided alternative).
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
EXAMPLE: You are interested in water quality of the Torrens River; so you collect water from the upper reaches which represent
rural land-use and also from the lower reaches which represent urban land-use to look for differences in nutrients (P & N).
Null = mean (population) of N in the upper reaches of the River Torrens does not differ from the mean
of N in the lower reaches of the River Torrens.
Alternative = mean of N in the upper reaches of the River Torrens does differ (is greater than or less
than) the mean of N in the lower reaches of the River Torrens
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
EXAMPLE: You are interested in water quality of the Torrens River; so you collect water from the upper reaches which represent
rural land-use and also from the lower reaches which represent urban land-use to look for differences in nutrients (P & N).
Null = mean (population) of N in the upper reaches of the River Torrens does not differ from the mean
of N in the lower reaches of the River Torrens.
Alternative = mean of N in the upper reaches of the River Torrens does differ (is greater than or less
than) the mean of N in the lower reaches of the River Torrens
H0: µRURAL = µURBAN
H1: µRURAL ≠ µURBAN
H1: µRURAL > µURBAN
H1: µRURAL < µURBAN
6 steps in carrying out a hypothesis test (generic recipe)
2. Choose an appropriate statistical test and calculate the test statistic
• In hypothesis tests, the test statistic summaries the differences between the observed
and expected data.
• It is an algorithm, that uses the sample statistics 𝑥,ҧ s, 𝑝,Ƹ p, n, SE to produce a single
value (the test statistic).
• The test statistic is a random variable with a distribution that we know.
• The choice of statistical test takes into account your experimental design, how data is
collected, the ‘type’ data you have, and some key assumptions
6 steps in carrying out a hypothesis test (generic recipe)
3. Determine the sampling distribution of the test statistic (from step 2)

• The sampling distribution of a statistic is the distribution of all possible values taken by
the test statistic when all possible samples of a fixed size n are taken from the
population.
• It is a theoretical concept – we do not actually build it.
• The sampling distribution of a statistic is the ‘probability distribution’ of the statistic.
6 steps in carrying out a hypothesis test (generic recipe)
3. Determine the sampling distribution of the test statistic (from step 2)

Sampling variability:
• Each time we take a random sample from a population, we are likely to get a different set
of individuals and thus calculate a different test statistic.
• If we take a lot of random samples of the same size from a given population, the variation
from sample to sample - the sampling distribution – will follow a predictable pattern.
• Sampling distributions are never exactly Normal, but as sample size increases, the
sampling distribution of the test statistic becomes approximately Normal.

"A simple random sample (SRS) of size n consists of n

individuals from the population chosen in such a way that
every set of n individuals has an equal chance to be the sample
actually selected."
6 steps in carrying out a hypothesis test (generic recipe)
4. Find the p-value associated with the test statistic based on knowledge of the
sampling distribution of the test statistic.
• We use the test statistic to find the p-value:
“the probability of getting the observed value of the test statistic or a more extreme
value if the null is true”.
• The p-value can be interpreted as the strength of the evidence provided by the
observed data against the null.
• Large test statistic values generally result in small p-values, indicating strong evidence
against the null.
• p-values must always fall between 0-1.
6 steps in carrying out a hypothesis test (generic recipe)
5. Make a decision to either “reject” or “accept” your null hypothesis (based on step 4)
• Set an alpha value (α) or significance level – a fixed value that the researcher sets as a
benchmark as decisive! For example, 0.05 is most commonly used across most scientific
disciplines.
• If the p-value is small – REJECT the NULL (H0). If using α = 0.05, then reject if the p-value
< 0.05.
• If the p-value is large – do NOT reject the NULL (H0). If using α = 0.05, then do not reject if
the p-value > 0.05.
• If the p-value is small and we reject the NULL, we state that “the data is statistically
significant”.
6 steps in carrying out a hypothesis test (generic recipe)
6. State your conclusion in the context of the scientific problem -relate it back to the real
world (population!)
• Make a conclusion, support your statements with the test statistic, p-values, n and df.
• Present the output in a table or imbed it onto a graph.
Estimation- confidence intervals

The confidence interval (CI) is a range of values that’s likely to include a population value with a
certain degree of confidence.
It is often expressed a % whereby a population means lies between an upper and lower interval.

CI are powerful and informative,

represent on a graph as error bars
instead of SD or range.
Next 7 lectures STATISTICAL INFERENCES

hypothesis testing (significance tests)

& estimation (confidence intervals, CI)

population proportions (p) A population mean (µ) Comparing 2 population

means (µ)
z-tests One-sample t-test Paired t-test & two-sample t-
(Wk 8, 9) (Wk 10) test
(Wk 11)
CI for proportions CI for a population mean CI for the difference in 2
(Wk 9) (Wk 10) population means
(Wk11)
Hypothesis
Next 7 lectures testing of population proportions
We frequently collect “count” data on categorical variables
• Whether or not a person is employed
• What telecommunications company a person uses
• What university a person has studied at
We were first introduced to “discrete random variables” in Week 7 (Monday lecture), when we
learned about binomial probability models – interested in solving “probability” of an event
occurring (HINT: probability of x successes out of n trials).
Now, we are interested in significance testing! i.e. asking specific questions regarding whether
there is evidence to support a particular claim about a population proportion. Is the difference
between the OBSERVED data and what we would EXPECT to see if H0 was true, large enough
to raise doubts about H0.
Next 7 lectures
EXAMPLE 1: Your company produces a sunblock lotion designed to protect the skin from UVA and UVB exposure to the
sun. You hire a testing company to compare your product with your major competitor’s product. For 13 of 20 randomly
chosen subjects, your product provides better protection; for the remaining seven, the competitor’s product does.
Do you have evidence to claim that your product provides better protection?
That is, does the population proportion significantly differ from 50%?

EXAMPLE 2: A Google research study asked 5013 smartphones users about how they used their phones. In response to
a question about purchases, 3563 reported that they purchased an item after using their smartphones to search for
information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use their phones to research an
item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

EXAMPLE 3: A sample of university student mobile phone users were asked about their service provider, where 18 of the
33 students surveyed were Optus customers. Optus claims that the proportion of Optus customers among university
student mobile phone users is 50%.
Does the sample provide evidence to support Optus' claim?
That is, does the population proportion significantly differ from 50%?
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
2. Choose an appropriate statistical test and calculate the test statistic
3. Determine the sampling distribution of the test statistic (from step 2)
4. Find the p-value associated with the test statistic
5. Make a decision to either “reject” or “accept” the null hypothesis (based on step 4)
6. State the conclusion in the context of the scientific problem -relate it back to the real
world (population!)
Next 7 lectures
Considerations for hypothesis testing of population proportions:
• Tests for the null hypothesis of a single population proportion (H0: p = p0) are based on the z
statistic from the one-sample z-test; where the test statistic summarises the differences
between the observed and expected data.
• The test statistic is a random variable with a distribution that we know
• Assumptions:
✓Random: a random sample of subjects for testing
✓Normal: Assuming a null (H0: p = p0) is true, we can use Normal distribution to calculate
p-values (elaborate on next slide)
• The p-value can be interpreted as the strength of the evidence provided by the observed
data against H0.
Normal Approximation for binomial distributions
• As n gets larger, something interesting happens to the shape of a binomial
distribution, that is, it becomes approximately Normal. In this case, the sampling
distribution of 𝑝Ƹ becomes approximately Normal.
• As a rule of thumb, we will use the Normal approximation when n is large; that
• n x p ≥ 10, AND
• n(1 – p) ≥ 10
- page 324 of Moore et 2017.
Next 7 lectures
EXAMPLE 1: Your company produces a sunblock lotion designed to protect the skin
from UVA and UVB exposure to the sun. You hire a testing company to compare your
product with your major competitor’s product. For 13 of 20 randomly chosen subjects,
your product provides better protection; for the remaining seven, the competitor’s
product does.

Do you have evidence to claim that your product provides better protection?

That is, does the population proportion significantly differ from 50%?
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses
2. Choose an appropriate statistical test and calculate the test statistic
3. Determine the sampling distribution of the test statistic (from step 2)
4. Find the p-value associated with the test statistic
5. Make a decision to either “reject” or “accept” the null hypothesis (based on step 4)
6. State the conclusion in the context of the scientific problem -relate it back to the real
world (population!)
6 steps in carrying out a hypothesis test (generic recipe)
1. Pose the null and alternative hypotheses

For significance testing of a single population proportion, p:

H 0 : p = p0 H0: p = 0.5 (null hypothesis)

H1: p ≠ p0 H1: p ≠ 0.5 (two-sided alternative hypothesis)
H1: p > p0 H1: p > 0.5 (one-sided alternative hypothesis)
H1: p < p0 H1: p < 0.5 (one-sided alternative hypothesis)

Find the P-value by calculating the

probability of getting a z statistic this
large or larger in the direction
specified by the alternative
hypothesis Ha:
6 steps in carrying out a hypothesis test (generic recipe)
2. Choose an appropriate statistical test and calculate the test statistic

Tests for the null hypothesis of a single population (H0: p = p0) are based on the z statistic
from the one-sample z-test:
Inference about a population proportion p from an SRS of
size n is based on the sample proportion, 𝑝:Ƹ

count of successes in sample 𝑋

𝑝Ƹ = =
size of sample 𝑛
6 steps in carrying out a hypothesis test (generic recipe)
2. Choose an appropriate statistical test and calculate the test statistic

Tests for the null hypothesis of a single population (H0: p = p0) are based on the z statistic
from the one-sample z-test:
Inference about a population proportion p from an SRS of
size n is based on the sample proportion, 𝑝:Ƹ

count of successes in sample 𝑋

𝑝Ƹ = =
size of sample 𝑛

13
𝑝Ƹ= 20 = 0.65
0.65 −0.5 0.15 0.15 0.15
n=20 Then, z = 0.5(1−0.5) = 0.5(0.5) = = = 1.34
√ √ √0.0125 0.112
20 20
p0 = 0.5
6 steps in carrying out a hypothesis test (generic recipe)
3. Determine the sampling distribution of the test statistic (from step 2)

p-values are then calculated from the sampling distribution, Z ~N(0,1) distribution, (when
the expected value of successes, n × p0, and the expected number of failures, n(1 - p0)
are both greater than 10.
Check the conditions for using a Normal approximation:
n × p0 = 20 x 0.5 = 10
n(1 - p0) = 20(1 – 0.5) = 10
As both are at least 10, we may use the Normal approximation.
6 steps in carrying out a hypothesis test (generic recipe)
4. Find the p-value associated with the test statistic (in this case z-statistic)
The z statistic has approximately the standard Normal distribution when H0 is true. P-values
therefore come from the standard Normal distribution. Also, upon testing the conditions, both n×p0
and n(1 - p0) were greater than 10, so we can obtain p-values from the Z~N(0,1) distribution.
From Table A or the function in excel “=NORM.S.DIST(z,TRUE)” (from Week 7 Friday’s lecture),
we can find P(z < 1.34). It is 0.9099. So the probability in the upper tail is 1-0.9099 = 0.0901.

The p-value is the

Find the P-value by calculating the probability of getting a z statistic this large or larger
area in both tails, p = in the direction specified by the alternative hypothesis Ha:
2 x 0.0901 = 0.1802
(two-sided p-value).
One-sided? or two-sided?
It depends upon your H1
(alternative hypothesis)
6 steps in carrying out a hypothesis test (generic recipe)
5. Make a decision to either “reject” or “not reject” the null hypothesis (based on step 4)

P-value (two-sided) = 0.1802

If we set our alpha value (α) at 0.05: to reject the null if a p-value is less than 0.05 and
accept the null if a p-value is greater then 0.05.

Then our NULL HYPOTHESIS of H0: p = 0.5 is not rejected.

6 steps in carrying out a hypothesis test (generic recipe)
6. State the conclusion in the context of the scientific problem -relate it back to the real
world (population!)

Does the population proportion significantly differ from 50%?

Answer: No, it does not!

Do you have evidence to claim that the new sunscreen provides better protection?
Answer: There is no statistical evidence to support the claim that the sunscreen provides
better protection.
Summary of the assumptions/conditions for inference on p
1) The data used for the estimate are an SRS from the population studied.
2) The population size is at least 10 times as large as the sample used for inference.
This ensures that the standard deviation of 𝑝Ƹ is close to p(1 − p)/n.
3) The sample size, n, is large enough that the sampling distribution can be
approximated with a Normal distribution. How large a sample size is required
depends in part on the value of p and the test conducted. Otherwise, rely on the
binomial distribution (will not show you this further in this topic).
 Next 7 lectures
Suppose that the sunscreen provided better UVA and UVB protection for 15 of the 20
subjects. Perform the significance test and summarise the results.

That is, does the population proportion significantly differ from 50%?

STEP 1: H0: p = 0.50 H1: p ≠ 0.50

STEP 2: z = 2.23

15
𝑝Ƹ
= = 0.75
20 0.75 −0.5 0.25 0.25 0.25
n=20 Then, z = 0.5(1−0.5) = 0.5(0.5) = √0.0125 = 0.112 = 2.23
√ √
20 20
p0 = 0.5
Next 7 lectures
Suppose that the sunscreen provided better UVA and UVB protection for 15 of the 20
subjects. Perform the significance test and summarise the results.

That is, does the population proportion significantly differ from 50%?

STEP 1: H0: p = 0.50 H1: p ≠ 0.50 If conditions are met, we should do a one-sample z test for the
STEP 2: z = 2.23 population proportion p.

STEP 3: YES, conditions are met ✓Random: We assume the company chose a random
sample of subjects for testing.
✓Normal: Assuming H0: p = 0.50 is true, the expected
Normal approximation to Binomial: numbers of successes and failures are np0 = 20(0.50) = 10
Z ~N(0,1) and n(1 – p0) = 20(1 - 0.50) = 10, respectively. Because
both of these values are at least 10, we can use the Normal
approximation.
✓The sampling distribution: z ~ N(0,1)
Next 7 lectures
Suppose that the sunscreen provided better UVA and UVB protection for 15 of the 20
subjects. Perform the significance test and summarise the results.

That is, does the population proportion significantly differ from 50%?

STEP 1: H0: p = 0.50 H1: p ≠ 0.50

STEP 2: z = 2.23 Finding the probability of z = 2.23 in Table A or using
the function in excel “=NORM.S.DIST(2.23,TRUE)”.
STEP 3: YES, conditions are met
STEP 4: p-value = 0.026 We find that P(z < 2.23) is 0.987.

So the probability in the upper tail is 1-0.987 = 0.013

(one-sided p-value), or the probability for both tails is
2 x 0.013 = 0.026 (two-sided p-value).
Next 7 lectures
Suppose that the sunscreen provided better UVA and UVB protection for 15 of the 20
subjects. Perform the significance test and summarise the results.

That is, does the population proportion significantly differ from 50%?

STEP 1: H0: p = 0.50 H1: p ≠ 0.50

STEP 2: z = 2.23
If we set alpha (α) at 0.05: to reject the null
STEP 3: YES, conditions are met if a p-value is less than 0.05 and accept
STEP 4: p-value = 0.026 the null if a p-value is greater then 0.05.
STEP 5: reject the NULL
Then our NULL HYPOTHESIS of H0: p =
0.5 is REJECTED!
Next 7 lectures
Suppose that the sunscreen provided better UVA and UVB protection for 15 of the 20
subjects. Perform the significance test and summarise the results.

That is, does the population proportion significantly differ from 50%?

STEP 1: H0: p = 0.50 H1: p ≠ 0.50

STEP 2: z = 2.23 Conclusion: There is
STEP 3: YES, conditions are met statistical evidence to
STEP 4: p-value = 0.026 support the claim that this
STEP 5: reject the NULL new sunscreen offers
STEP 6: Yes, the population proportion does differ better protection than its
significantly from 0.5. competitor’s.
Next
used their 7 lectures
EXAMPLE 2: A Google research study asked 5013 smartphones users about how they
phones. In response to a question about purchases, 3563 reported that they
purchased an item after using their smartphones to search for information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use
their phones to research an item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

STEP 1: H0: p = 0.70 H1: p ≠ 0.70 or H1: p > 0.70

STEP 2: z = 1.54

3563
𝑝Ƹ = 5013 = 0.71
n=5013
p0 = 0.7
0.71 −0.7 0.01
0.01 0.01
Then, z = 0.7(1−0.7) = 0.7(0.3) = = = 1.54
√ √ √0.00004189 0.0065
5013 5013
Next
used their 7 lectures
EXAMPLE 2: A Google research study asked 5013 smartphones users about how they
phones. In response to a question about purchases, 3563 reported that they
purchased an item after using their smartphones to search for information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use
their phones to research an item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

STEP 1: H0: p = 0.70 H1: p ≠ 0.70 or H1: p > 0.70

STEP 2: z = 1.54
STEP 3: YES, conditions are met. If conditions are met, we should do a one-sample z test for the
population proportion p.

Normal approximation to Binomial: ✓Random: We assume the company chose a random

Z ~N(0,1) sample of subjects for testing.
✓Normal: Assuming H0: p = 0.70 is true, the expected
numbers of successes and failures are np0 = 5013(0.70) =
3509 and n(1 – p0) = 5013(1 - 0.70) = 1504, respectively.
Because both of these values are at least 10, we can use
the Normal approximation.
✓The sampling distribution: z ~ N(0,1)
Next
used their 7 lectures
EXAMPLE 2: A Google research study asked 5013 smartphones users about how they
phones. In response to a question about purchases, 3563 reported that they
purchased an item after using their smartphones to search for information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use
their phones to research an item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

STEP 1: H0: p = 0.70 H1: p ≠ 0.70 or H1: p > 0.70

STEP 2: z = 1.54
STEP 3: YES, conditions are met
STEP 4: p-value = 0.124
Finding the probability of z = 1.54 in Table A or using
the function in excel “=NORM.S.DIST(1.54,TRUE)”.

We find that P(z < 1.54) is 0.938.

So the probability in the upper tail is 1-0.938 = 0.062

(one-sided p-value), or the probability for both tails is
2 x 0.062 = 0.124 (two-sided p-value).
Next
used their 7 lectures
EXAMPLE 2: A Google research study asked 5013 smartphones users about how they
phones. In response to a question about purchases, 3563 reported that they
purchased an item after using their smartphones to search for information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use
their phones to research an item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

STEP 1: H0: p = 0.70 H1: p ≠ 0.70 or H1: p > 0.70

STEP 2: z = 1.54
STEP 3: YES, conditions are met
STEP 4: p-value = 0.124 If we set alpha (α) at 0.05: to reject the null if a
STEP 5: fail to reject the NULL p-value is less than 0.05 and accept the null if
a p-value is greater then 0.05.

Then our NULL HYPOTHESIS of H0: p = 0.7 is

NOT rejected!
Next
used their 7 lectures
EXAMPLE 2: A Google research study asked 5013 smartphones users about how they
phones. In response to a question about purchases, 3563 reported that they
purchased an item after using their smartphones to search for information about the item.
Is there evidence to support the statement that greater than 70% of smartphone users use
their phones to research an item prior to its purchase.
That is, does the population proportion significantly differ from 70%?

STEP 1: H0: p = 0.70 H1: p ≠ 0.70 or H1: p > 0.70

STEP 2: z = 1.54
STEP 3: YES, conditions are met.
STEP 4: p-value = 0.124
Conclusion: There is no
STEP 5: fail to reject the NULL
statistical evidence to support
STEP 6: No, the population proportion does not differ
the claim that greater than 70%
significantly from 0.7 (either side of 0.7).
of smartphone users use their
phones to research an item
prior to its purchase.
Next 7 lectures
Take home problem!
EXAMPLE 3: A sample of university student mobile phone users were asked about their
service provider, where 18 of the 33 students surveyed were Optus customers. Optus
claims that the proportion of Optus customers among university student mobile phone
users is 50%.
Does the sample provide evidence to support Optus' claim?
That is, does the population proportion significantly differ from 50%?

STEP 1: H0: H1 :
STEP 2: z =
STEP 3: Are conditions met? Normal approximation to Binomial: Z ~N(0,1)
STEP 4: p-value =
STEP 5: accept or reject the NULL?
STEP 6: Make a statement…………