GE 2113 | BIO STATISTICS AND EPIDEMIOLOGY - LECTURE
INFIRENTIAL STATISTICS
TRANSCRIBED BY: ELLEMA, ARIANNE KENDRA P.
Inferential statistics involves drawing
conclusions about populations by examining
samples. It allows us to make inferences about
the entire set, including specific examples
within it, based on information obtained from a
subset of examples. These inferences rely on the
principles of evidence and utilize sample
statistics as a basis for drawing broader
conclusions.
The accuracy of inferential statistics depends
largely on the accuracy of sample data and
how it represents the larger population. This can
be effectively done by obtaining a random
sample. Results that are based on non-random
samples are usually discarded. Random
sampling - though not very straightforward
always – is extremely important for carrying out
inferential techniques.
A. Hypothesis Testing
B. Level of Significance
C. Calculated and Critical Value
D. Statistical Tools
HYPOTHESIS TESTING
a. Null
b. Alternative
Hypothesis testing, sometimes called
significance testing, is an act in statistics
whereby an analyst tests an assumption
regarding a population parameter.
AE | 2D
| BMLS – 2D | RTRMF | Date Transcribed:
Hypothesis testing is used to assess the
plausibility of a hypothesis by using sample
data.
Such data may come from a larger population,
or from a data-generating process. The word
"population" will be used for both of these cases
in the following descriptions.
Hypothesis testing is a decision-making process
for evaluating claims about a population.
Moreover, it is basically testing an assumption
that we can make about a population.
Examples
1. Does the mean height of Second Year
Medtech students differ from 66 inches?
2. Is the proportion of Second Year Medtech
male student’s height significantly higher than
the Second Year Medtech female students?
Null Hypothesis
- It is denoted by H0.
- it is the initial claim.
- It shows no significant difference, no changes,
nothing happened, no relationship between
two parameters.
- The independent variable has no effect on
the dependent variable.
- H0 : Ц 1 = Ц 2
(Note: Ц – population parameter)
10/ 09/ 23
Null hypothesis used:
( = ) equal to, the same as, not changed from,
is.
The null hypothesis is usually a hypothesis of
equality between population parameters; e.g.,
a null hypothesis may state that the population
mean return is equal to zero.
The alternative hypothesis is effectively the
opposite of a null hypothesis (e.g., the
population mean return is not equal to zero).
Thus, they are mutually exclusive, and only one
can be true. However, one of the two
hypotheses will always be true.
Alternative Hypothesis
-It is denoted by Ha or H1.
-It is the contrary to the null hypothesis.
-It shows that there is significant difference, an
effect, change, relationship between a
parameter and specific value.
-The independent variable has an effect on the
dependent variable.
-Ha : Ц1 Ц2 , Ha : Ц1 > Ц2 , Ha : Ц1 < Ц2
Alternative Hypothesis used:
(≠) not equal, different from, changed from, not
the same as
(>) greater than, above, higher than, longer
than, bigger than, increased, at least
1
(<) less than, below, lower than, smaller than, Ha : The mean score in Ha : The current percentage of unmarried
shorter than, decreased or reduced from, at couple is different from 34%.
Math of the Second
most. or
Year Medtech Ha: Ц ≠ 8.1
H0: Ц = 34%
Stating/Formulating the Null and Alternative students is not 8.1. Ha: Ц ≠ 34%
Hypothesis for Each Statement
Example 1. The average of bus drivers in Example 6: A chemist invented an additive to
Tacloban City is 38.8 years. Example 4. A social worker wants to test (at α = increase the lifespan of rechargeable battery.
H0: Ц = 38.8 years 0.05 whether the average Body Mass Index The said additive will extend the battery’s life
Ha : Ц ≠ 38.8 years (BMI) of the students under feeding program is span to 48 months.
different from 18.2 kg.
Example 2: The average number of calories of H0 : The average lifespan extension of
a low-calorie meal is at most 300. Null and Alternative Null and Alternative rechargeable battery is 48 months.
H0: Ц = 300 calories Hypotheses in Words Hypotheses in Ha : The average lifespan extension of
Ha : Ц < 300 calories Symbols rechargeable battery is not 48 months.
H0 : The average BMI H0: Ц = 18.2 or
Example 3: The school has 99% confidence of the students under H0: Ц = 48 months
level cord claims that the mean score in Math feeding program is Ha: Ц ≠ 48 months or Ц < 48 months or Ц > 48
of the Second Year Medtech students is 8.1 The not different from months
teacher wishes to find out if the claim is true. 18.2 kg.
She tests if there is significant difference Ha : The average BMI Ha: Ц ≠ 18.2 Example 7: We want to test whether the mean
between the batch mean score and the mean of the students under height of Second Year Medtech students is 60
score of students in her class. feeding program is inches.
different from 18.2 kg. H0 : The mean height of Second Year Medtech
Null and Alternative Null and Alternative students is 60 inches.
Example 5: In 2015, it was recorded that around Ha : The mean height of Second Year Medtech
Hypotheses in Words Hypotheses in
34% of the population were not married. A students is not (or less than or higher than) 60
Symbols researcher surveyed a random sample of 500 inches.
H0 : The mean score in couples. He found out that 18% of them were or
living together but unmarried. Test at 5% H0: Ц = 60 inches
Math of the Second significance level if the current percentage of Ha: Ц ≠ 60 inches or Ц < 60 inches or Ц > 60
Year Medtech H0: Ц = 8.1 unmarried couple is different from 34%. inches
students is 8.1. H0 : The current percentage of unmarried BENEFITS OF HYPOTHESIS TESTING
couple is 34%. (or The current percentage of  Helps assess the accuracy of new ideas
unmarried couple is not different from 34%.) or theories by testing them against data.

AE | 2D 2
  Allows researchers to determine whether
the evidence supports their hypothesis,
helping to avoid false claims and
conclusions.
 Provides a framework for decision-
making based on data rather than
personal or biases.
 Helps to reduce the effects of chance
and confounding variables, providing a
robust framework for making informed
conclusions.
LIMITATIONS OF HYPOTHESIS TESTING
 Relies exclusively on data and doesn’t
provide a comprehensive
understanding of the subject being
studied.
 The accuracy of the results depends on
the quality of the available data and
statistical methods used.
 Inaccurate data or inappropriate
hypothesis formulation may lead to
incorrect conclusions or failed tests.
 Can lead to errors, such as analyst either
accepting or rejecting a null hypothesis
when they shouldn’t have which can
results to false conclusion or missed
opportunities to identify significant
patterns or relationship in the data.
LEVEL OF SIGNIFICANCE
Level of significance means how sure a
researcher is that the results found are not
accidental (not by chance).
AE | 2D
A level of significance of p = 0.05 means that • It is the probability of making the wrong
there is a 95% probability that the results found decision when the null hypothesis is true.
in the study are the result of a true • In public health research, α is usually
relationship/difference between groups being 0.01 or 1%. In social science research, α
compared. It also means that there is a 5% is usually 0.05% or 5% and 0.10 or 10% in
chance that the results were found by chance other studies.
alone and no true relationship exists between • This implies that there is 1%, 5%, or 10%
groups. probability of rejecting a true null
hypothesis.
For example, a level of significance of 0.05
means that there is a 5% chance that the result Level of Significance in Symbol
is insignificant, or that it just happened by α = 0.01
chance alone. α = 0.05 or
α = 0.10
Significant difference means that the difference Note: If the alternative hypothesis used is ≠,
between the groups could not have happened then alpha level will be divided by 2.
by chance. It can happen that a difference is α/2 = 0.005
seen between two groups, however, if the α/2 = 0.025 or
difference is significant, then the researchers α/2 = 0.05
accept that it was caused by the specific
experimental intervention used during the Level of Test Types
experiment. Significance α One-Tailed Two-tailed
0.05 +1.645 or - ±1.96
Note: The lower the level of significance, the 1.645
less likely the result was found by chance; or, to 0.01 +2.33 or - ±2.58
state it another way, the more likely the 2.33
researchers are correct in stating that the given 0.001 +3.09 or - ±3.30
intervention is effective. 3.09
In summary, level of significance Example 1: Maria uses 5% level of significance
• It is denoted by alpha or α which refers in proving that there is no significant change in
to the degree of significance in which the average number of enrollees in the 10
we accept or reject the null hypothesis. sections for the last two years. Find the level of
• 100% accuracy is not possible in significance.
accepting or rejecting a hypothesis.
3
It means that the chance that the null
hypothesis (H0) would be rejected when it is
true is 5%. (or α = 0.05 is the area under the
normal curve within the rejection region.)
Example 2: In 2015, it was recorded that around
34% of the population were not married. A
researcher surveyed a random sample of 500
couples. He found out that 18% of them were
living together but unmarried. Test at 5%
significance level if the current percentage of
unmarried couples is different from 34%.
Answer: α = 0.10
Example 4: A sample of 100 private school
students were surveyed whether they planned
to transfer to public school in the incoming
school year. Results showed that 40% of these
students will transfer to public school. Test the
average number of private school students that
will transfer to public school is not 40% at 5%
level of significance.
Answer: α = 0.025
Example 5: The average number of years to
finish basic education is 14. A sample of 30
senior high school students were asked and
found out that the mean number of years to
finish their basic education is 12 with a standard
deviation of 2 years. Test the hypothesis at 93%
confidence interval that the average number
of years to finish basic education is less than 14
years.
Answer: α = 0.07
Interpretation:
▪ If the value of the test statistic is less
extreme than the critical value, then the
null hypothesis cannot be rejected.
▪ If the test statistic is more extreme than
the critical value, the null hypothesis is
rejected and the alternative hypothesis
is accepted.
▪ In other words, the critical value divides
the distribution graph into the
acceptance and the rejection region. If
the value of the test statistic falls in the
rejection region, then the null hypothesis
is rejected otherwise it cannot be
rejected.
Areas
Rejection Region: also known as critical region,
is the set of all values of the test statistic that
causes us to reject the null hypothesis.

Answer: α/2 = 0.025 CALCULATED AND CRITICL VALUE Non-rejection Region: also known as
acceptance region, is the set of all values of
Example 3: An average construction worker Critical Value the test statistic that causes us to fail to reject
hourly rate pay in the Philippines is Php 62.50  can be defined as a value that is the null hypothesis.
with a standard deviation of Php 6.01. A compared to a test statistic in hypothesis
random sample of 20 manufacturing workers testing to determine whether the null Level of Test Types
were asked on their hourly rate and found out hypothesis is to be rejected or not. Significance α One-Tailed Two-tailed
that they had an average pf Php 50.00 hourly  cut-off value that is used to mark the 0.05 +1.645 or - ±1.96
rate pay with a standard deviation of Php 5.00. start of a region where the test statistic, 1.645
Construct a 90% confidence interval that the obtained in hypothesis testing, is unlikely 0.01 +2.33 or - ±2.58
average hourly rate for construction workers is to fall in. In hypothesis testing, the critical 2.33
higher than the average hourly rate for value is compared with the obtained 0.001 +3.09 or - ±3.30
manufacturing workers. test statistic to determine whether the 3.09
null hypothesis has to be rejected or not.

AE | 2D 4
STEPS TO DETERMINE CRITICAL VALUE
• Step 1: Subtract the confidence level
from 100%. 100% - 95% = 5%.
• Step 2: Convert this value to decimals to
get α. Thus, α = 5%.
• Step 3: If it is a one-tailed test then the
alpha level will be the same value in
step 2. However, if it is a two-tailed test,
the alpha level will be divided by 2.
• Step 4: Depending on the type of test
conducted the critical value can be
looked up from the corresponding
distribution table using the alpha value.
T-CRITICAL VALUE
A t-test is used when the population standard
deviation is not known and the sample size is
lesser than 30. A t-test is conducted when the
population data follows a Student t distribution.
STEPS TO CALCULATE T-VALUE
• Determine the alpha level.
• Subtract 1 from the sample size. This
gives the degrees of freedom (df).
• If the hypothesis test is one-tailed
then use the one-tailed t distribution
table. Otherwise, use the two-tailed t
distribution table for a two-tailed test.
• Match the corresponding df value
(left side) and the alpha value (top
row) of the table. Find the
intersection of this row and column
to give the t critical value.
AE | 2D
Test Statistics for 1 Sample t-test: - Reject the null hypothesis if the test
Formula: statistic does not lie in the
acceptance region (two-tailed
hypothesis test).
Where:
Ẍ is the sample mean
μ. is the population mean
s is the sample standard deviation
n is the size of the sample.
Test Statistic for 2 Sample t-test >:
 For one-tailed test or directional test:
- Reject the null hypothesis if test
statistics > t-critical value (right-tailed
hypothesis test).
 For one-tailed test or directional test:
- Reject the null hypothesis if test
statistic < t critical value (left tailed
hypothesis test).
The decision criterion is used for all tests. Only
the test statistics and critical value change.
Z-CRITICAL VALUE
A z test is conducted on a normal distribution
 For two-tailed test or directional test <: when the population standard deviation is
5
known and the sample size is greater than or
equal to 30.
STEPS ON CALCULATING Z-CRITICAL VALUE
• Find the alpha level.
• Subtract the alpha level from 1 for a
two-tailed test. For a one-tailed test
subtract the alpha level from 0.5.
• Look up the area from the z distribution
table to obtain the z critical value. For a
left-tailed test, a negative sign needs to
be added to the critical value at the
end of the calculation.
Test Statistics for one sample z-test:
, ð is the population standard
deviation.
Test Statistic for two sample Z-test:
F-CRITICAL VALUE
The F test is largely used to compare the
variances of two samples. The test statistic so
obtained is also used for regression analysis.
AE | 2D
STEPS ON CALCULATING F CRITICAL VALUE
• Find the alpha level.
• Subtract 1 from the size of the first
sample. This gives the first degree of
freedom. Say, x
• Similarly, subtract 1 from the second
sample size to get the second df. Say, y.
• Using the f distribution table, the
intersection of the x column and y row
will give the f critical value.
Test Statistic for large samples: , conducted. The critical value needs to be
variance of the first sample and variance
of the second sample.
Test Statistic for small samples: , area closest to 0.4921. The closest area is
variance of the first sample and variance
of the second sample.
CHI-SQUARE CRITICAL VALUE
The chi-square test is used to check if the
sample data matches the population data. It
can also be used to compare two variables to
see if they are related.
THE CHI-SQUARE CRITICAL VALUE IS GIVEN AS
FOLLOWS:
• Identify the alpha level.
• Subtract 1 from the sample size to
determine the degrees of freedom (df).
• Using the chi-square distribution table,
the intersection of the row of the df and
the column of the alpha value yields the
chi-square critical value.
Test statistic for chi-squared test statistic:
CRITICAL VALUE CALCULATION
Example1: Suppose a right-tailed z test is being
calculated for a 0.0079 alpha level. Then the
steps are as follows:
• Subtract the alpha level from 0.5. Thus,
0.5 – 0.0079 = 0.4921
• Using the z distribution table find the
0.4922. As this value is at the intersection
of 2.4 and 0.02 thus, the z critical value =
2.42.
ILLUSTRATING THE REJECTION REGION
Example 1: According to a survey, the average
daily usage of social media worldwide of
global internet users amounts to 142 minutes
per day. Sofia conducts her own survey among
her 10 friends to find if their time spent on social
media is significantly higher than the global
survey.
6
 Friend’s Name Minutes per The computed t-value = 1.593 the global computed t-
Day Spent on The critical t-value = 1.833 usage. value is at the
Social Media non-rejection
Illustration: Ha: The region.
Allen 132
average
Bryan 148
daily usage
Ellen 165
of social
Jake 157
media of
Mindie 120
her friends is
Shamsie 144 higher than
Candice 136 the global
Dory 160 usage.
Mitch 185
Mila 173 Example 2: A medical trial is conducted to test
Important Note:
- If the computed value is less than the whether a certain drug reduces cholesterol
Formulating Hypotheses: level. Upon trial, the computed z-value of 2.715
critical value, the computed value is
Null Hypothesis (H0): The average daily usage lies in the rejection area.
at the non-rejection area (or the
of social media of her friends is the same as the
computed value is at the
global usage. Null Hypothesis (H0): The certain drug is
acceptance region.), thus, we fail to
Alternative Hypothesis (Ha): The average daily effective in reducing cholesterol level by 60%.
reject the null hypothesis (or we
usage of social media of her friends is higher Alternative Hypothesis (Ha): The certain drug is
accept the null hypothesis.)
than the global usage. not effective in reducing cholesterol level by
- If the computed value is greater
Given: 60%.
than the critical value, the
Ц = 142 n = 10 Ẍ = 152 s = 19.855
computed value is at the rejection
Asked: t = ? Illustration:
area, thus, we reject the null
Solution:
hypothesis.

Interpretation of Data:
t= = 1.593 H0: The
average Since the We fail to
Determine the critical value. Use df = n – 1 = 9,
daily usage computed reject the null
one-tailed test at 5% level of significance.
of social value is less hypothesis, H0.
media of than the (or We accept
To find the critical values, use the t – distribution
her friends is critical value, the null
Table: (not attached)
the same as the hypothesis.) The computed z-value is greater than the
critical value.

AE | 2D 7
 hypothesis, H0 (or we accept the null Answer: Rejection
hypothesis, H0)
 If the computed value is at the rejection Example 6: Given
Interpretation of Data: area, we accept the null hypothesis, H0. Critical t-value of ± 1.311
H0: The Computed t-value of -1.134
certain drug The certain We reject the Interpretation of Data: The computed t-value is at the _______________
is effective in drug is not null The computed t-value is at the non- region.
reducing effective in hypothesis, H0 rejection area. Therefore, we fail to reject the
cholesterol reducing in favor of null hypothesis, H0 (or we accept the null
level by 60%. cholesterol alternative hypothesis, H0.)
Ha: The level by 60%. hypothesis,
certain drug Ha. (or We Example 4: Given:
is not accept the Critical t-value of 1.318
effective in alternative Computed t-value of 1.1 Answer: Acceptance
reducing hypothesis.) The computed t-value is at the _______________
cholesterol region.
STATISTICAL TOOLS
level by 60%.
Correlation
Example 3: Sketch the rejection region of the A correlation is a statistical measure of the
test hypothesis with critical values of ± 1.753 relationship between two variables. The
and determine if the computed t-value of – measure is best used in variables that
1.52 lies in that region. demonstrate a linear relationship between
Illustration: Answer: Acceptance each other. The fit of the data can be visually
represented in a scatterplot. Using a
Example 5: Given scatterplot, we can generally assess the
Critical t-value of -1.671 relationship between the variables and
Computes t-value of -2.45 determine whether they are correlated or not.
The computed t-value is at the _______________
region. Formula:

Important Note:
Where:
 If the computed value is at the non-
rejection area, we fail to reject the null

AE | 2D 8
 • rxy – the correlation coefficient of the
linear relationship between the variables
x and y
• xi – the values of the x-variable in a
sample
• x̅ – the mean of the values of the x-
variable
• yi – the values of the y-variable in a
sample
• ȳ – the mean of the values of the y-
variable
STEPS TO CALCULATE THE CORRELATION
COEFFICIENT
1. Obtain a data sample with the values of
x-variable and y-variable.
2. Calculate the means (averages) x̅ for
the x-variable and ȳ for the y-variable.
3. For the x-variable, subtract the mean
from each value of the x-variable (let’s
call this new variable “a”). Do the same
for the y-variable (let’s call this variable
“b”).
4. Multiply each a-value by the
corresponding b-value and find the sum
of these multiplications (the final value is
the numerator in the formula).
5. Square each a-value and calculate the
sum of the result
6. Find the square root of the value
obtained in the previous step (this is the
denominator in the formula).
7. Divide the value obtained in step 4 by
the value obtained in step 7.
AE | 2D
Example: John is an investor. His portfolio
primarily tracks the performance of the S&P 500
and John wants to add the stock of Apple Inc.
Before adding Apple to his portfolio, he wants
to assess the correlation between the stock and
the S&P 500 to ensure that adding the stock
won’t increase the systematic risk of his
portfolio.
Step 1: Gather the prices for the last five years.
Step 2: Find the mean od S&P 500 and Apple.
Step 3:
Step 4: Multiple values from a and b
Step 5: Square root of both a and b
Calculate the coefficient:
Interpretation: The coefficient indicates that the
prices of the S&P 500 and Apple Inc. have a
high positive correlation. This means that their
respective prices tend to move in the same
direction. Therefore, adding Apple to his
portfolio would, in fact, increase the level of
systematic risk.
Pearson R Correlation
The Pearson correlation coefficient (r) is the
most common way of measuring a linear
correlation. It is a number between –1 and 1
that measures the strength and direction of the
relationship between two variables.
What is the Pearson correlation coefficient?
The Pearson correlation coefficient (r) is the
most widely used correlation coefficient and is
known by many names:
• Pearson’s r
• Bivariate correlation
• Pearson product-moment correlation
coefficient (PPMCC)
• The correlation coefficient
The Pearson correlation coefficient is a
descriptive statistic, meaning that it summarizes
the characteristics of a dataset. Specifically, it
describes the strength and direction of the
linear relationship between two quantitative
variables.
9
Although interpretations of the relationship
strength (also known as effect size) vary
between disciplines, the table below gives
general rules of thumb:
The Pearson correlation coefficient is also an
inferential statistic, meaning that it can be used
to test statistical hypotheses. Specifically, we
can test whether there is a significant
relationship between two variables.
Visualizing the Pearson correlation coefficient
Another way to think of the Pearson correlation
coefficient (r) is as a measure of how close the
observations are to a line of best fit.
The Pearson correlation coefficient also tells you
whether the slope of the line of best fit is
negative or positive. When the slope is
negative, r is negative. When the slope is
positive, r is positive.
AE | 2D
When r is 1 or –1, all the points fall exactly on
the line of best fit:
When r is greater than .5 or less than –.5, the
points are close to the line of best fit:
When r is between 0 and .3 or between 0 and –
.3, the points are far from the line of best fit:
When r is 0, a line of best fit is not helpful in
describing the relationship between the
variables:
When to use the Pearson correlation coefficient
Pearson correlation coefficient is a good
choice when all of the following are true:
• Both variables are quantitative: You will
need to use a different method if either
of the variables is qualitative.
• The variables are normally distributed:
You can create a histogram of each
variable to verify whether the
distributions are approximately normal.
It’s not a problem if the variables are a
little non-normal.
• The data have no outliers: Outliers are
observations that don’t follow the same
patterns as the rest of the data. A
scatterplot is one way to check for
outliers—look for points that are far
away from the others.
• The relationship is linear: “Linear” means
that the relationship between the two
variables can be described reasonably
well by a straight line. You can use a
scatterplot to check whether the
relationship between two variables is
linear.
10
Calculating the Pearson correlation coefficient
Below is a formula for calculating the Pearson
correlation coefficient (r)
Example: Dataset
Imagine that you’re studying the relationship
between newborns’ weight and length. You
have the weights and lengths of the 10 babies
born last month at your local hospital. After you
convert the imperial measurements to metric,
you enter the data in a table:
Step 1: Calculate the sums of x and y
- Start by renaming the variables to
“x” and “y.” It doesn’t matter which
variable is called x and which is
AE | 2D
called y—the formula will give the
same answer either way.
- Next, add up the values of x and y.
(In the formula, this step is indicated
by the Σ symbol, which means “take
the sum of”.)
Step 3: Calculate the cross product and its sum
Calculating the sums of x and y
In a final column, multiply together x and y (this
Weight = x
is called the cross product). Take the sum of the
Length = y
new column.
Calculating the cross product and its sum
Σx = 3.63 + 3.02 + 3.82 + 3.42 + 3.59 + 2.87 + 3.03
+ 3.46 + 3.36 + 3.30
Σx = 33.5
Σy = 53.1 + 49.7 + 48.4 + 54.2 + 54.9 + 43.7 + 47.2
+ 45.2 + 54.4 + 50.4
Σy = 501.2
Step 2: Calculate x2 and y2 and their sums
Create two new columns that contain the
squares of x and y. Take the sums of the new
columns.
Calculating x2 and y2 and their sums
Step 4: Calculate r
Use the formula and the numbers you
calculated in the previous steps to find r.
Calculating r.
n = 10
11

Testing for the significance of the Pearson
correlation coefficient
The Pearson correlation coefficient can also be
used to test whether the relationship between
two variables is significant.
The Pearson correlation of the sample is r. It is
an estimate of rho (ρ), the Pearson correlation
of the population. Knowing r and n (the sample
size), we can infer whether ρ is significantly
different from 0.
• Null hypothesis (H0): ρ = 0
• Alternative hypothesis (Ha): ρ
≠0
To test the hypotheses, you can either use
software like R or Stata or you can follow the
three steps below.
AE | 2D
Step 1: Calculate the t value
Calculate the t value (a test statistic) using this
formula:
Step 2: Find the critical value of t
You can find the critical value of t (t*) in a t
table. To use the table, you need to know three
things:
• The degrees of freedom (df): For Pearson
correlation tests, the formula is df = n – 2.
• Significance level (α): By convention,
the significance level is usually .05.
• One-tailed or two-tailed: Most often,
two-tailed is an appropriate choice for
correlations.
Example: Finding the critical value of t
For a two-tailed test of significance at α = .05
and df = 8, the critical value of t (t*) is 1.86.
Step 3: Compare the t value to the critical value
Determine if the absolute t value is greater than
the critical value of t. “Absolute” means that if
the t value is negative you should ignore the
minus sign.
Example: Comparing the t value to the critical
value of t (t*)
t = 1.506
t* = 1.86
The t value is less than the critical value of t.
Step 4: Decide whether to reject the null
hypothesis
• If the t value is greater than the critical
value, then the relationship is statistically
significant (p < α). The data allows you
to reject the null hypothesis and
provides support for the alternative
hypothesis.
• If the t value is less than the critical
value, then the relationship is not
statistically significant (p > α). The data
doesn’t allow you to reject the null
hypothesis and doesn’t provide support
for the alternative hypothesis.
Example: Deciding whether to reject the null
hypothesis. For the correlation between weight
and height in a sample of 10 newborns, the t
value is less than the critical value of t.
Therefore, we don’t reject the null hypothesis
that the Pearson correlation coefficient of the
population (ρ) is 0. There is no significant
relationship between weight and height (p >
.05).
(Note that a sample size of 10 is very small. It’s
possible that you would find a significant
relationship if you increased the sample size.)
12
Reporting the Pearson correlation coefficient
If you decide to include a Pearson correlation
(r) in your paper or thesis, you should report it in
your results section.
You can follow these rules if you want to report
statistics in APA Style:
• You don’t need to provide a reference
or formula since the Pearson correlation
coefficient is a commonly used statistic.
• You should italicize r when reporting its
value.
• You shouldn’t include a leading zero (a
zero before the decimal point) since the
Pearson correlation coefficient can’t be
greater than one or less than negative
one.
• You should provide two significant digits
after the decimal point.
When Pearson’s correlation coefficient is used
as an inferential statistic (to test whether the
relationship is significant), r is reported alongside
its degrees of freedom and p value. The
degrees of freedom are reported in
parentheses beside r.
Example: Reporting the Pearson correlation
coefficient in APA Style
Newborns’ weight and length were moderately
correlated, although the relationship was not
statistically significant, r(8) = .47, p > .17.
Spearman Rank order
AE | 2D
Spearman rank order correlation coefficient, in
short Spearman’s rank correlation, measures
the strength and direction of association
between two ranked variables. It basically gives
the measure of monotonicity of the relation
between two variables i.e. how well the
relationship between two variables could be
represented using a monotonic function.
The formula for Spearman’s rank coefficient is:
Step 1: Create a table for the given data.
Step 2: Rank both the data in descending
order. The highest marks will get a rank of 1 and
where the lowest marks will get a rank of 5.
��� = Spearman’s rank correlation Step 3: Calculate the difference between the
coefficient ranks (d) and the square value of d.
di = Difference between the two ranks of each Step 4: Add all your d square values.
observation
n = Number of observations
The Spearman Rank Correlation can take a
value from +1 to -1 where,
• A value of +1 means a perfect
association of rank
• A value of 0 means that there is no
association between ranks Step 5: Insert these values into the formula.
• A value of -1 means a perfect negative
association of rank
Example = 1 - (6 * 14) / 5(25 - 1)
Consider the score of 5 students in Math and = 0.3
Science that are mentioned in the table.
Interpretation: The Spearman’s Rank Correlation
for the given data is 0.3. The value is near 0,
which means that there is a weak correlation
between the two ranks.
13

PHI COEFFICIENT
A Phi Coefficient (sometimes called a Mean
Square Contingency Coefficient) is a measure
of the association between two binary
variables.
For a given 2×2 table for two random variables
x and y:
The Phi Coefficient can be calculated as:
Φ = (AD-BC) / √(A+B)(C+D)(A+C)(B+D)
Example.
Suppose we want to know whether or not
gender is associated with political party
preference. We take a simple random sample
of 25 voters and survey them on their political
party preference. The following table shows the
results of the survey:
We can calculate the Phi Coefficient between
the two variables as:
Φ = (4*4-9*8) / √(4+9)(8+4)(4+8)(9+4) = (16-72) /
√24336 = -0.3589
AE | 2D
How to Interpret a Phi Coefficient
Similar to a Pearson Correlation Coefficient, a
Phi Coefficient takes on values between -1 and
1 where:
• -1 indicates a perfectly negative
relationship between the two variables.
• 0 indicates no association between the
two variables.
• 1 indicates a perfectly positive
relationship between the two variables.
In general, the further away a Phi Coefficient is
from zero, the stronger the relationship between
the two variables.
In other words, the further away a Phi
Coefficient is from zero, the more evidence
there is for some type of systematic pattern
between the two variables.
KENDALL RANK CORRELATION COEFFICIENT
Kendal Rank Correlation Coefficient, also
known as Kendall’s Tau, is a non-parametric
measure of relationships between columns of
ranked data. The Tau correlation coefficient
returns a value of 0 to 1, where:
• 0 is no relationship,
• 1 is a perfect relationship.
A quirk of this test is that it can also produce
negative values (i.e. from -1 to 0). Unlike a linear
graph, a negative relationship doesn’t mean
much with ranked columns (other than you
perhaps switched the columns around), so just
remove the negative sign when you’re
interpreting Tau.
Several version’s of Tau exist.
• Tau-A and Tau-B are usually used for
square tables (with equal columns and
rows). Tau-B will adjust for tied ranks.
• Tau-C is usually used for rectangular
tables. For square tables, Tau-B and Tau-
C are essentially the same.
Formula:
Kendall’s Tau = (C – D / C + D)
Where: C is the number of concordant pairs
and D is the number of discordant pairs.
Example.
Two interviewers ranked 12 candidates (A
through L) for a position. The results from most
preferred to least preferred are:
• Interviewer 1: ABCDEFGHIJKL.
• Interviewer 2: ABDCFEHGJILK.
Calculate the Kendall Tau Correlation.
Step 1: Make a table of rankings. The first
column, “Candidate” is optional and for
reference only. The rankings for Interviewer 1
should be in ascending order (from least to
greatest).
14
 Step 4: Sum the values in the two columns:

When all concordant pairs have been

counted, it looks like this: Step 5: Insert the totals into the formula:
Kendall’s Tau = (C – D / C + D)
Step 2: Count the number of concordant pairs, = (61 – 5) / (61 + 5) = 56 / 66 = .85.
using the second column. Concordant pairs are
how many larger ranks are below a certain Interpretation: The Tau coefficient is .85,
rank. For example, the first rank in the second suggesting a strong relationship between the
interviewer’s column is a “1”, so all 11 ranks rankings.
below it are larger.
Perfect Correlation
Counting how many values are below the
second column seems very odd when you first
Step 3: Count the number of discordant pairs do it. But it does work. Just as a thought
and insert them into the next column. The experiment, here’s what the spreadsheet looks
number of discordant pairs is similar to Step 2, like if both interviewers were in perfect
only you’re looking for smaller ranks, not larger agreement:
ones.

However, going down the list to the third row (a

rank of 4), the rank immediately below (3) is
smaller, so it doesn’t count for a concordant
pair.

AE | 2D 15

And, inserting the totals into the formula we
get:
Tau = (66 – 0) / (66 + 0) = 1, which is (as we
expect) perfect agreement.
Calculating Statistical Significance
If you want to calculate statistical significance
for your result, use this formula to get a z-value:
Inserting the values from our results:
Interpretation: Finding the area for a z-score of
3.85 on a z-table gives an area of .0001 — a tiny
probability value which tells you this result is
statistically significant.
T-Test
A t test is a statistical test that is used to
compare the means of two groups. It is often
AE | 2D
used in hypothesis testing to determine whether
a process or treatment actually has an effect
on the population of interest, or whether two
groups are different from one another.
The t test tells you how significant the
differences between group means are. It lets
you know if those differences in means could
have happened by chance. The t test is usually
used when data sets follow a normal
distribution but you don’t know the population
variance.
For example, you might flip a coin 1,000 times
and find the number of heads follows a normal
distribution for all trials. So you can calculate
the sample variance from this data, but the
population variance is unknown. Or, a drug
company may want to test a new cancer drug
to find out if it improves life expectancy. In an
experiment, there’s always a control group (a
group who are given a placebo, or “sugar
pill”). So while the control group may show an
average life expectancy of +5 years, the group
taking the new drug might have a life
expectancy of +6 years. It would seem that the
drug might work. But it could be due to a fluke.
To test this, researchers would use a Student’s t-
test to find out if the results are repeatable for
an entire population.
In addition, a t test uses a t-statistic and
compares this to t-distribution values to
determine if the results are statistically
significant.
However, note that you can only uses a t test to
compare two means. If you want to compare
three or more means, use an ANOVA instead.
The T Score
The t score is a ratio between the difference
between two groups and the difference within
the groups.
- Larger t scores = more difference
between groups.
- Smaller t score = more similarity
between groups.
A t score of 3 tells you that the groups are three
times as different from each other as they are
within each other. So when you run a t test,
bigger t-values equal a greater probability that
the results are repeatable.
T-Values and P-values
How big is “big enough”? Every t-value has a p-
value to go with it. A p-value from a t test is the
probability that the results from your sample
data occurred by chance. P-values are from
0% to 100% and are usually written as a decimal
(for example, a p value of 5% is 0.05). Low p-
values indicate your data did not occur by
chance. For example, a p-value of .01 means
there is only a 1% probability that the results
from an experiment happened by
Calculating the Statistic / Test Types
There are three main types of t-test:
• An Independent Samples t-test
compares the means for two groups.
16
 • A Paired sample t-test compares means Although the manufacturers are different, you
from the same group at different times might be subjecting them to the same
(say, one year apart). conditions.
• A One sample t-test tests the mean of a
single group against a known mean. Example.
Calculate a paired t test by hand for the
Paired T Test (Paired Samples T Test or following data:
Dependent Samples T Test)
A paired t test (also called a correlated pairs t-
test, a paired samples t test or dependent
samples t test) is where you run a t test on
dependent samples. Dependent samples are
essentially connected — they are tests on the
Step 3: Square the differences from Step 1.
same person or thing. For example:
• Knee MRI costs at two different hospitals,
• Two tests on the same person before
and after training,
• Two blood pressure measurements on
the same person using different
equipment.
Step 1: Subtract each Y score from each X
When to Choose a Paired T Test / Paired score.
Samples T Test / Dependent Samples T Test
Choose the paired t-test if you have two
measurements on the same item, person or
thing.

But you should also choose this test if you have Step 4: Add up all of the squared differences
two items that are being measured with a from Step 3.
unique condition. For example, you might be
measuring car safety performance in vehicle
research and testing and subject the cars to a
series of crash tests. Step 2: Add up all of the values from Step 1
then set this number aside for a moment.

AE | 2D 17

Step 5: Use the following formula to calculate
the t-score:
1. The “ΣD” is the sum of
X-Y from Step 2.
2. ΣD2: Sum of the
squared differences
(from Step 4).
3. (ΣD)2: Sum of the
differences (from Step
2), squared.
If you’re unfamiliar with the Σ notation used in
the t test, it basically means to “add everything
up”. You may find this article useful: summation
notation.
AE | 2D
Step 6: Subtract 1 from the sample size to get
the degrees of freedom. We have 11 items. So
11 – 1 = 10.
Step 7: Find the p-value in the t-table, using the
degrees of freedom (df) in Step 6. But if you
don’t have a specified alpha level, use 0.05
(5%).
So, for this example t test problem, with df = 10,
the t-value is 2.228.
Step 8: In conclusion
Compare your t-table value from Step 7 (2.228)
to your calculated t-value (-2.74).
The calculated t-value is greater than the table
value at an alpha level of .05. In addition, note
that the p-value is less than the alpha level: p
<.05. So we can reject the null hypothesis that
there is no difference between means.
However, note that you can ignore the minus
sign when comparing the two t-values as ±
indicates the direction; the p-value remains the
same for both directions.
Independent T Test
The independent t-test, also called the two
sample t-test, independent-samples t-test or
student's t-test, is an inferential statistical test
that determines whether there is a statistically
significant difference between the means in
two unrelated groups.
Independent sample t-test is a statistical
technique that is used to analyze the mean
comparison of two independent groups. In
independent samples t-test, when we take two
samples from the same population, then the
mean of the two samples may be identical. But
when samples are taken from two different
populations, then the mean of the sample may
differ. In this case, it is used to draw conclusions
about the means of two populations, and used
to tell whether or not they are similar.
Assumptions in independent samples t-test:
• Assumes that the dependent variable is
normally distributed.
• Assumes that the variance of the two
groups are the same as the dependent
variable.
• Assumes that the two samples are
independent of each other.
• Samples are drawn from the population
at random.
18
 • In independent sample t-test, all
observations must be independent of
each other.
• In independent sample t-test,
dependent variables must be measured
on an interval or ratio scale.
Procedures for independent sample t-test:
1. Set up the hypothesis.
a. Null Hypothesis: It is assumed when the
means of the two groups are not significantly
different.
b. Alternative Hypothesis: Assumes that the
means of the two groups are significantly
different.
2. Calculate the standard deviation for the
independent sample t-test by using this formula:
3. Calculate the value of the independent
sample t-test by using this formula:
4. Degree of freedom for independent sample
t-test:
where
V= degree of freedom
N1+N2= number of observations in both
samples of the independent sample t-test.
5. Hypothesis testing: In hypothesis testing for
the independent sample t-test, statistical
decisions are made about whether or not the
AE | 2D
two population means are identical. Compare
the calculated value of the independent
sample t-test with the table value of the sample
t-test. If the calculated value of the
independent sample t-test is greater than the
table value of the predetermined significance
level, we will reject the null hypothesis and say
that the means of the two groups are different.
If the calculated value of the independent
sample t-test is less than the table value, then
we will say that the means of the two groups
are the same.
b.1 Mann-Whitney U Test
The Mann-Whitney U Test, also known as the
Wilcoxon Rank Sum Test, is a non-parametric
statistical test used to compare two samples or
groups.
The Mann-Whitney U Test assesses whether two
sampled groups are likely to derive from the
same population, and essentially asks; do these
two populations have the same shape with
regards to their data? In other words, we want
evidence as to whether the groups are drawn
from populations with different levels of a
variable of interest. It follows that the
hypotheses in a Mann-Whitney U Test are:
• The null hypothesis (H0) is that
the two populations are
equal.
• The alternative hypothesis
(H1) is that the two
populations are not equal.
Some researchers interpret this as comparing
the medians between the two populations (in
contrast, parametric tests compare the means
between two independent groups). In certain
situations, where the data are similarly shaped
(see assumptions), this is valid – but it should be
noted that the medians are not actually
involved in calculation of the Mann-Whitney U
test statistic. Two groups could have the same
median and be significantly different
according to the Mann-Whitney U test.
When to use the Mann-Whitney U Test
Non-parametric tests (sometimes referred to as
‘distribution-free tests’) are used when you
assume the data in your populations of interest
do not have a Normal distribution. You can
think of the Mann Whitney U-test as analogous
to the unpaired Student’s t-test, which you
would use when assuming your two populations
are normally distributed, as defined by their
means and standard deviation (the parameters
of the distributions).
The Mann-Whitney U Test is a common
statistical test that is used in many fields
including economics, biological sciences and
19
epidemiology. It is particularly useful when you
are assessing the difference between two
independent groups with low numbers of
individuals in each group (usually less than 30),
which are not normally distributed, and where
the data are continuous. If you are interested in
comparing more than two groups which have
skewed data, a Kruskal-Wallis One-Way analysis
of variance (ANOVA) should be used.
Mann-Whitney U Test Assumptions
Some key assumptions for Mann-Whitney U Test
are detailed below:
• The variable being compared between
the two groups must be continuous
(able to take any number in a range –
for example age, weight, height or heart
rate). This is because the test is based on
ranking the observations in each group.
• The data are assumed to take a non-
Normal, or skewed, distribution. If your
data are normally distributed, the
unpaired Student’s t-test should be used
to compare the two groups instead.
• While the data in both groups are not
assumed to be Normal, the data are
assumed to be similar in shape across
the two groups.
• The data should be two randomly
selected independent samples,
meaning the groups have no
relationship to each other. If samples are
paired (for example, two measurements
from the same group of participants),
then a paired samples t-test should be
used instead.
AE | 2D
• Sufficient sample size is needed for a
valid test, usually more than 5
observations in each group.
Example.
Consider a randomized controlled trial
evaluating a new anti-retroviral therapy for HIV.
A pilot trial randomly assigned participants to
either the treated or untreated groups (N=14).
We want to assess the viral load (quantity of
virus per milliliter of blood) in the treated versus
the untreated groups.
The data are shown below:
These data are both skewed with a sample size
of n=7 in each treatment arm, and so a non- After summing the ranks for each group, the
parametric test is appropriate. Before we Mann-Whitney U test statistic is selected as the
calculate the test, we choose a significance smallest of the two following calculated U
level (usually α=0.05). The first step is to assign values:
ranks to the values from the full sample (both
treatment groups pooled together) in order
from smallest to largest. We can then generate
a test statistic based on the ranks.
The table below shows the viral load values in
the treated and untreated groups ranked
smallest to largest, along with the summed
ranks of each group:
20

Where we let 1 denote the treated group and 2
denote the untreated group (denotation of
groups is arbitrary), where n1 and n2 are the
number of participants and where R1 and R2
are the sums of the ranks in the treated and
untreated groups, respectively. In this example,
U1=41 and U2=8. We therefore select U=8 as the
test statistic.
Important Note on Normal Approximation
There are situations where the sample size may
be too large for the reference table to be used
to calculate the exact probability distribution –
in which case we can use a Normal
approximation instead. Since U is found by
adding together independent, similarly
distributed random samples, the central limit
theorem applies when the sample is large
(usually >20 in each group). The standard
deviation of the sum of the ranks can be used
AE | 2D
to generate a z-statistic and a significance
value generated this way. If the null hypothesis
is true, the distribution of U approximates to a
Normal distribution.
Next we determine a ‘critical value’ of U with
which to compare our calculated test statistic,
which we can do using a reference table of
critical values and using our sample sizes (n=7 in
both groups) and two-sided level of
significance (α=0.05).
In our current example, the critical value can
be determined from the reference table as 8.
Finally, we can use this to accept or reject the
null hypothesis using the following decision rule:
Reject H0 if U ≤ 8.
Given that our U statistic is equal to the
critical value, we can reject the null hypothesis
that the two groups are equal and accept the
alternative hypothesis that there is evidence of
a difference in viral load between the groups
treated with the new therapy versus untreated.
Wilcoxon Signed Rank
The Wilcoxon signed-rank test is a non-
parametric statistical hypothesis test used either
to test the location of a population based on a
sample of data, or to compare the locations of
two populations using two matched samples.
The one-sample version serves a purpose similar
to that of the one-sample Student's t-test.
The Wilcoxon test, which can refer to either the
rank sum test or the signed rank test version, is a
nonparametric statistical test that compares
two paired groups. The tests essentially
calculate the difference between sets of pairs
and analyze these differences to establish if
they are statistically significantly different from
one another.
KEY TAKEAWAYS
• The Wilcoxon test compares two paired
groups and comes in two versions, the
rank sum test, and signed rank test.
• The goal of the test is to determine if two
or more sets of pairs are different from
one another in a statistically significant
manner.
• Both versions of the model assume that
the pairs in the data come from
dependent populations, i.e., following
the same person or share price through
time or place.
Understanding the Wilcoxon Test
The rank sum and signed rank tests were both
proposed by American statistician Frank
Wilcoxon in a groundbreaking research paper
published in 1945. The tests laid the foundation
for hypothesis testing of nonparametric
statistics, which are used for population data
that can be ranked but do not have numerical
values, such as customer satisfaction or music
reviews. Nonparametric distributions do not
have parameters and cannot be defined by
an equation as parametric distributions can.
The types of questions that the Wilcoxon test
can help answer include things like:
21
 •Are test scores different from 5th
grade to 6th grade for the same
students?
• Does a particular drug have an
effect on health when tested on the
same individuals?
These models assume that the data comes
from two matched, or dependent, populations,
following the same person or stock through
time or place. The data is also assumed to be
continuous as opposed to discrete. Because it is
a nonparametric test, it does not require a
particular probability distribution of the
dependent variable in the analysis.
Types of the Wilcoxon Test
• The Wilcoxon rank sum test can be used
to test the null hypothesis that two
populations have the same continuous
distribution. A null hypothesis is a
statistical test that says there's no
significant difference between two
populations or variables. The base
assumptions necessary to employ the
rank sum test is that the data are from
the same population and are paired,
the data can be measured on at least
an interval scale, and the data were
chosen randomly and independently.
• The Wilcoxon signed rank test assumes
that there is information in the
magnitudes and signs of the differences
between paired observations. As the
nonparametric equivalent of the paired
student's t-test, the signed rank can be
used as an alternative to the t-test when
AE | 2D
the population data does not follow a
normal distribution.
Calculating a Wilcoxon Test Statistic
The steps for arriving at a Wilcoxon signed rank
test statistic, W, are as follows:
1. For each item in a sample of n items,
obtain a difference score, Di, between
two measurements (i.e., subtract one
from the other).
2. Neglect then positive or negative signs
and obtain a set of n absolute
differences |Di|.
3. Omit difference scores of zero, giving
you a set of n non-zero absolute
difference scores, where n' ≤ n. Thus, n'
becomes the actual sample size.
4. Then, assign ranks Ri from 1 to n to each
of the |Di| such that the smallest
absolute difference score gets rank 1
and the largest gets rank n. If two or
more |Di| are equal, they are each
assigned the average rank of the ranks
they would have been assigned
individually had ties in the data not
occurred.
5. Now reassign the symbol "+" or "–" to
each of the n ranks Ri, depending on
whether Di was originally positive or
negative.
6. The Wilcoxon test statistic W is
subsequently obtained as the sum of the
positive ranks.
Chi square test
A Pearson’s chi-square test is a statistical test for
categorical data. It is used to determine
whether your data are significantly different
from what you expected. There are two types
of Pearson’s chi-square tests:
• The chi-square goodness of fit test is
used to test whether the frequency
distribution of a categorical variable is
different from your expectations.
• The chi-square test of independence is
used to test whether two categorical
variables are related to each other.
Chi-square is often written as Χ2 and is
pronounced “kai-square” (rhymes with “eye-
square”). It is also called chi-squared.
What is a chi-square test?
Pearson’s chi-square (Χ2) tests, often referred to
simply as chi-square tests, are among the most
common nonparametric tests. Nonparametric
tests are used for data that don’t follow the
assumptions of parametric tests, especially the
assumption of a normal distribution.
If you want to test a hypothesis about the
distribution of a categorical variable you’ll
need to use a chi-square test or another
nonparametric test. Categorical variables can
be nominal or ordinal and represent groupings
such as species or nationalities. Because they
can only have a few specific values, they can’t
have a normal distribution.
Note: Parametric tests can’t test hypotheses
about the distribution of a categorical variable,
but they can involve a categorical variable as
an independent variable (e.g., ANOVAs).
22
Test hypotheses about frequency distributions
There are two types of Pearson’s chi-square
tests, but they both test whether the observed
frequency distribution of a categorical variable
is significantly different from its expected
frequency distribution. A frequency distribution
describes how observations are distributed
between different groups.
Frequency distributions are often displayed
using frequency distribution tables. A frequency
distribution table shows the number of
observations in each group. When there are
two categorical variables, you can use a
specific type of frequency distribution table
called a contingency table to show the number
of observations in each combination of groups.
Example 1. Bird species at a bird feeder.
Frequency of visits by bird species at a bird
feeder during a 24-hour period.
A chi-square test (a chi-square goodness of fit
test) can test whether these observed
AE | 2D
frequencies are significantly different from what
was expected, such as equal frequencies.
Example 2. Handedness and nationality
Contingency table of the handedness of a
sample of Americans and Canadians
A chi-square test (a test of independence) can
test whether these observed frequencies are
significantly different from the frequencies
expected if handedness is unrelated to
nationality.
Chi-square formula
Both of Pearson’s chi-square tests use the same
formula to calculate the test statistic, chi-
square (Χ2):
Where:
• Χ2 is the chi-square test statistic
• Σ is the summation operator (it means
“take the sum of”)
• O is the observed frequency
• E is the expected frequency
The larger the difference between the
observations and the expectations (O − E in the
equation), the bigger the chi-square will be. To
decide whether the difference is big enough to
be statistically significant, you compare the chi-
square value to a critical value.
When to use a chi-square test
A Pearson’s chi-square test may be an
appropriate option for your data if all of the
following are true:
1. You want to test a hypothesis about one
or more categorical variables. If one or
more of your variables is quantitative,
you should use a different statistical test.
Alternatively, you could convert the
quantitative variable into a categorical
variable by separating the observations
into intervals.
2. The sample was randomly selected from
the population.
3. There are a minimum of five
observations expected in each group or
combination of groups.
Types of chi-square tests
The two types of Pearson’s chi-square tests are:
• Chi-square goodness of fit test
• Chi-square test of
independence
Mathematically, these are actually the same
test. However, we often think of them as
different tests because they’re used for
different purposes.
Chi-square goodness of fit test
You can use a chi-square goodness of fit test
when you have one categorical variable. It
allows you to test whether the frequency
distribution of the categorical variable is
23
significantly different from your expectations.
Often, but not always, the expectation is that
the categories will have equal proportions.
Example: Hypotheses for chi-square goodness
of fit test
Expectation of equal proportions
• Null hypothesis (H0): The bird species visit
the bird feeder in equal proportions.
• Alternative hypothesis (HA): The bird
species visit the bird feeder in different
proportions.
Expectation of different proportions
• Null hypothesis (H0): The bird species visit
the bird feeder in the same proportions
as the average over the past five years.
• Alternative hypothesis (HA): The bird
species visit the bird feeder in different
proportions from the average over the
past five years.
Chi-square test of independence
You can use a chi-square test of independence
when you have two categorical variables. It
allows you to test whether the two variables are
related to each other. If two variables are
independent (unrelated), the probability of
belonging to a certain group of one variable
isn’t affected by the other variable.
Example: Chi-square test of independence
• Null hypothesis (H0): The proportion of
people who are left-handed is the same
for Americans and Canadians.
AE | 2D
• Alternative hypothesis (HA): The
proportion of people who are left-
handed differs between nationalities.
Other types of chi-square tests
Some consider the chi-square test of
homogeneity to be another variety of Pearson’s
chi-square test. It tests whether two populations
come from the same distribution by
determining whether the two populations have
the same proportions as each other. You can
consider it simply a different way of thinking
about the chi-square test of independence.
McNemar’s test is a test that uses the chi-square
test statistic. It isn’t a variety of Pearson’s chi-
square test, but it’s closely related. You can
conduct this test when you have a related pair
of categorical variables that each have two
groups. It allows you to determine whether the
proportions of the variables are equal.
Example: McNemar’s test
Suppose that a sample of 100 people is offered
two flavors of ice cream and asked whether
they like the taste of each.
Contingency table of ice cream flavor
preference
Like chocolate Dislike chocolate
Like vanilla 47 32
Dislike 8 13
vanilla
• Null hypothesis (H0): The proportion of
people who like chocolate is the same
as the proportion of people who like
vanilla.
• Alternative hypothesis (HA): The
proportion of people who like chocolate
is different from the proportion of people
who like vanilla.
There are several other types of chi-square tests
that are not Pearson’s chi-square tests,
including the test of a single variance and the
likelihood ratio chi-square test.
How to perform a chi-square test
The exact procedure for performing a
Pearson’s chi-square test depends on which
test you’re using, but it generally follows these
steps:
1. Create a table of the observed and
expected frequencies. This can
sometimes be the most difficult step
because you will need to carefully
consider which expected values are
most appropriate for your null
hypothesis.
2. Calculate the chi-square value from
your observed and expected
frequencies using the chi-square
formula.
3. Find the critical chi-square value in a
chi-square critical value table or using
statistical software.
4. Compare the chi-square value to the
critical value to determine which is
larger.
5. Decide whether to reject the null
hypothesis. You should reject the null
hypothesis if the chi-square value is
24
 greater than the critical value. If you
reject the null hypothesis, you can
conclude that your data are
significantly different from what you
expected.
How to report a chi-square test
If you decide to include a Pearson’s chi-square
test in your research paper, dissertation or
thesis, you should report it in your results section.
You can follow these rules if you want to report
statistics in APA Style:
• You don’t need to provide a reference
or formula since the chi-square test is a
commonly used statistic.
• Refer to chi-square using its Greek
symbol, Χ2. Although the symbol looks
very similar to an “X” from the Latin
alphabet, it’s actually a different
symbol. Greek symbols should not be
italicized.
• Include a space on either side of the
equal sign.
• If your chi-square is less than zero, you
should include a leading zero (a zero
before the decimal point) since the chi-
square can be greater than zero.
• Provide two significant digits after the
decimal point.
• Report the chi-square alongside its
degrees of freedom, sample size, and p
value, following this format: Χ2 (degrees
of freedom, N = sample size) = chi-
square value, p = p value).
Example: Reporting a chi-square test
AE | 2D
There was no significant relationship between
handedness and nationality, Χ2 (1, N = 428) =
0.44, p = .505.
Linear Regression
Linear regression analysis is used to predict the
value of a variable based on the value of
another variable. The variable you want to
predict is called the dependent variable. The
variable you are using to predict the other
variable's value is called the independent
variable.
Regression models describe the relationship
between variables by fitting a line to the
observed data. Linear regression models use a
straight line, while logistic and nonlinear
regression models use a curved line. Regression
allows you to estimate how a dependent
variable changes as the independent
variable(s) change.
Simple Linear Regression
Simple linear regression is used to estimate the
relationship between two quantitative
variables. You can use simple linear regression
when you want to know:
1. How strong the relationship is between
two variables (e.g., the relationship
between rainfall and soil erosion).
2. The value of the dependent variable at
a certain value of the independent
variable (e.g., the amount of soil erosion
at a certain level of rainfall)
Simple linear regression example: You are a
social researcher interested in the relationship
between income and happiness. You survey
500 people whose incomes range from 15k to
75k and ask them to rank their happiness on a
scale from 1 to 10.
Your independent variable (income) and
dependent variable (happiness) are both
quantitative, so you can do a regression
analysis to see if there is a linear relationship
between them.
Note: If you have more than one independent
variable, use multiple linear regression instead.
Assumptions of simple linear regression
Simple linear regression is a parametric test,
meaning that it makes certain assumptions
about the data. These assumptions are:
1. Homogeneity of variance
(homoscedasticity): the size of the error
in our prediction doesn’t change
significantly across the values of the
independent variable.
2. Independence of observations: the
observations in the dataset were
collected using statistically valid
sampling methods, and there are no
hidden relationships among
observations.
3. Normality: The data follows a normal
distribution.
Linear regression makes one additional
assumption:
4. The relationship between the
independent and dependent variable is
linear: the line of best fit through the
25
 data points is a straight line (rather than
a curve or some sort of grouping factor).
If your data do not meet the assumptions of
homoscedasticity or normality, you may be
able to use a nonparametric test instead, such
as the Spearman rank test.
Example. Data that doesn’t meet the
assumptions.
You think there is a linear relationship between
cured meat consumption and the incidence of
colorectal cancer in the U.S. However, you find
that much more data has been collected at
high rates of meat consumption than at low
rates of meat consumption, with the result that
there is much more variation in the estimate of
cancer rates at the low range than at the high
range. Because the data violate the
assumption of homoscedasticity, it doesn’t work
for regression, but you perform a Spearman
rank test instead.
If your data violate the assumption of
independence of observations (e.g., if
observations are repeated over time), you may
be able to perform a linear mixed-effects
model that accounts for the additional
structure in the data.
How to perform a simple linear regression
Simple linear regression formula
The formula for a simple linear regression is:
• y is the predicted value of the
dependent variable (y) for any given
value of the independent variable (x).
AE | 2D
• B0 is the intercept, the predicted value
of y when the x is 0.
• B1 is the regression coefficient – how
much we expect y to change as x
increases.
• x is the independent variable ( the
variable we expect is influencing y).
• e is the error of the estimate, or how
much variation there is in our estimate of
the regression coefficient.
Linear regression finds the line of best fit line
through your data by searching for the
regression coefficient (B1) that minimizes the
total error (e) of the model.
Example 1.
You have to study the relationship between the
monthly e-commerce sales and the online
advertising costs. You have the survey results for
7 online stores for the last year.
Your task is to find the equation of the straight
line that fits the data best.
The following table represents the survey results
from the 7 online stores.
We can see that there is a positive relationship
between the monthly e-commerce sales (Y)
and online advertising costs (X).
“The positive correlation means that the values
of the dependent variable (y) increase when
the values of the independent variable (x) rise.”
So, if we want to predict the monthly e-
commerce sales from the online advertising
costs, the higher the value of advertising costs,
the higher our prediction of sales.
We will use the above data to build our Scatter
diagram.
Now, let’ see how the Scatter diagram looks
like:
The Scatter plot shows how much one variable
affects another. In our example, above Scatter
plot shows how much online advertising costs
affect the monthly e-commerce sales. It shows
their correlation.
Let’s see the simple linear regression equation.
Y = Β0 + Β1X
Y= 125.8 + 171.5*X
26
Note: You can find easily the values for Β0 and
Β1 with the help of paid or free statistical
software, online linear regression calculators or
Excel. All you need are the values for the
independent (x) and dependent (y) variables
(as those in the above table).
Now, we have to see our regression line:
Graph of the Regression Line:
Diagram 2
Linear regression aims to find the best-fitting
straight line through the points. The best-fitting
line is known as the regression line.
If data points are closer when plotted to
making a straight line, it means the correlation
between the two variables is higher. In our
example, the relationship is strong.
The orange diagonal line in diagram 2 is the
regression line and shows the predicted score
on e-commerce sales for each possible value
of the online advertising costs.
Interpretation of the Results
AE | 2D
The slope of 171.5 shows that each increase of
one unit in X, we predict the average of Y to
increase by an estimated 171.5 units.
The formula estimates that for each increase of
1 dollar in online advertising costs, the
expected monthly e-commerce sales are
predicted to increase by $171.5.
This was a simple linear regression example for a
positive relationship in business. Let’s see an
example of the negative relationship.
Example 2.
You have to examine the relationship between
the age and price for used cars sold in the last
year by a car dealership company.
Here is the table of the data:
Now, we see that we have a negative
relationship between the car price (Y) and car
age(X) – as car age increases, price decreases.
When we use the simple linear regression
equation, we have the following results:
Y = Β0 + Β1X
Y = 7836 – 502.4*X
Let’s use the data from the table and create
our Scatter plot and linear regression line:
Diagram 3:
Interpretation of the Results
With an estimated slope of – 502.4, we can
conclude that the average car price
decreases $502.2 for each year a car increases
in age.
a. Multiple Linear Regression
Multiple linear regression is used to estimate the
relationship between two or more independent
variables and one dependent variable. You
can use multiple linear regression when you
want to know:
27
 1. How strong the relationship is between
two or more independent variables and
one dependent variable (e.g. how
rainfall, temperature, and amount of
fertilizer added affect crop growth).
2. The value of the dependent variable at
a certain value of the independent
variables (e.g. the expected yield of a
crop at certain levels of rainfall,
temperature, and fertilizer addition).
Multiple Linear Regression Example
You are a public health researcher interested in
social factors that influence heart disease. You
survey 500 towns and gather data on the
percentage of people in each town who
smoke, the percentage of people in each town
who bike to work, and the percentage of
people in each town who have heart disease.
Because you have two independent variables
and one dependent variable, and all your
variables are quantitative, you can use multiple
linear regression to analyze the relationship
between them.
Assumptions of Multiple Linear Regression
Multiple linear regression makes all of the same
assumptions as simple linear regression:
Homogeneity of variance (homoscedasticity):
the size of the error in our prediction doesn’t
change significantly across the values of the
independent variable.
Independence of observations: the
observations in the dataset were collected
using statistically valid sampling methods, and
there are no hidden relationships among
variables.
AE | 2D
In multiple linear regression, it is possible that
some of the independent variables are actually
correlated with one another, so it is important
to check these before developing the
regression model. If two independent variables
are too highly correlated (r2 > ~0.6), then only
one of them should be used in the regression
model.
Normality: The data follows a normal
distribution.
Linearity: the line of best fit through the data
points is a straight line, rather than a curve or
some sort of grouping factor.
How to perform a Multiple Linear Regression
Multiple linear regression formula
The formula for a multiple linear regression is:
• = the predicted value of the
dependent variable
• = the y-intercept (value of y when all
other parameters are set to 0)
• = the regression coefficient ( )
of the first independent variable ( ) ExxonMobil, for example, depends on more
(a.k.a. the effect that increasing the
value of the independent variable has
on the predicted y value)
• … = do the same for however many
independent variables you are testing
• = the regression coefficient of
the last independent variable
• = model error (a.k.a. how much
variation there is in our estimate of )
To find the best-fit line for each independent
variable, multiple linear regression calculates
three things:
• The regression coefficients that lead to
the smallest overall model error.
• The t statistic of the overall model.
• The associated p value (how likely it is
that the t statistic would have occurred
by chance if the null hypothesis of no
relationship between the independent
and dependent variables was true).
It then calculates the t statistic and p value for
each regression coefficient in the model.
Example.
An analyst may want to know how the
movement of the market affects the price of
ExxonMobil (XOM). In this case, their linear
equation will have the value of the S&P 500
index as the independent variable, or
predictor, and the price of XOM as the
dependent variable.
In reality, multiple factors predict the outcome
of an event. The price movement of
than just the performance of the overall
market. Other predictors such as the price of
oil, interest rates, and the price movement of oil
futures can affect the price of Exon Mobil
(XOM) and the stock prices of other oil
companies. To understand a relationship in
which more than two variables are present,
multiple linear regression is used.
28
Multiple linear regression (MLR) is used to
determine a mathematical relationship among
several random variables.1 In other terms, MLR
examines how multiple independent variables
are related to one dependent variable. Once
each of the independent factors has been
determined to predict the dependent variable,
the information on the multiple variables can
be used to create an accurate prediction on
the level of effect they have on the outcome
variable. The model creates a relationship in
the form of a straight line (linear) that best
approximates all the individual data points.
Referring to the MLR equation above, in our
example:
• yi = dependent variable—the price of
XOM
• xi1 = interest rates
• xi2 = oil price
• xi3 = value of S&P 500 index
• xi4= price of oil futures
• B0 = y-intercept at time zero
• B1 = regression coefficient that measures
a unit change in the dependent
variable when xi1 changes - the change
in XOM price when interest rates change
• B2 = coefficient value that measures a
unit change in the dependent variable
when xi2 changes—the change in XOM
price when oil prices change
The least-squares estimates—B0, B1, B2…Bp—are
usually computed by statistical software. As
many variables can be included in the
regression model in which each independent
AE | 2D
variable is differentiated with a number—1,2, 3,
4...p. The multiple regression model allows an
analyst to predict an outcome based on
information provided on multiple explanatory
variables.
Still, the model is not always perfectly accurate
as each data point can differ slightly from the
outcome predicted by the model. The residual
value, E, which is the difference between the
actual outcome and the predicted outcome, is
included in the model to account for such slight
variations.
Assuming we run our XOM price regression
model through a statistics computation
software, that returns this output:
Interpretation of the Results
An analyst would interpret this output to mean
if other variables are held constant, the price of
XOM will increase by 7.8% if the price of oil in
the markets increases by 1%. The model also
shows that the price of XOM will decrease by
1.5% following a 1% rise in interest rates. R2
indicates that 86.5% of the variations in the
stock price of Exxon Mobil can be explained by
changes in the interest rate, oil price, oil futures,
and S&P 500 index.
Logistic Regression
Logistic regression is the type of statistical
model (also known as logit model) is often used
for classification and predictive analytics.
Logistic regression estimates the probability of
an event occurring, such as voted or didn't
vote, based on a given dataset of
independent variables.
Logistic regression is a supervised machine
learning algorithm that accomplishes binary
classification tasks by predicting the probability
of an outcome, event, or observation. The
model delivers a binary or dichotomous
outcome limited to two possible outcomes:
yes/no, 0/1, or true/false.
Logical regression analyzes the relationship
between one or more independent variables
and classifies data into discrete classes. It is
extensively used in predictive modeling, where
the model estimates the mathematical
probability of whether an instance belongs to a
specific category or not.
For example, 0 – represents a negative class; 1 –
represents a positive class. Logistic regression is
commonly used in binary classification
problems where the outcome variable reveals
either of the two categories (0 and 1).
Some examples of such classifications and
instances where the binary response is
expected or implied are:
1. Determine the probability of heart attacks:
With the help of a logistic model, medical
practitioners can determine the relationship
between variables such as the weight, exercise,
etc., of an individual and use it to predict
whether the person will suffer from a heart
attack or any other medical complication.
29
2. Possibility of enrolling into a university:
Application aggregators can determine the
probability of a student getting accepted to a
particular university or a degree course in a
college by studying the relationship between
the estimator variables, such as GRE, GMAT, or
TOEFL scores.
3. Identifying spam emails: Email inboxes are
filtered to determine if the email
communication is promotional/spam by
understanding the predictor variables and
applying a logistic regression algorithm to
check its authenticity.
Key Advantages of Logistic Regression
The logistic regression analysis has several
advantages in the field of machine learning.
1. Easier to implement machine learning
methods:
2. Suitable for linearly separable datasets
3. Provides valuable insights:
Logistic Regression Equation and Assumptions
AE | 2D
Logistic regression uses a logistic function called The following equation represents logistic
a sigmoid function to map predictions and their regression:
probabilities. The sigmoid function refers to an
S-shaped curve that converts any real value to
a range between 0 and 1.
Moreover, if the output of the sigmoid function
(estimated probability) is greater than a
predefined threshold on the graph, the model
predicts that the instance belongs to that class.
If the estimated probability is less than the
predefined threshold, the model predicts that here,
the instance does not belong to the class. • x = input value
For example, if the output of the sigmoid • y = predicted output
function is above 0.5, the output is considered • b0 = bias or intercept term
as 1. • b1 = coefficient for input (x)
This equation is similar to linear regression,
On the other hand, if the output is less than 0.5, where the input values are combined linearly to
the output is classified as 0. Also, if the graph predict an output value using weights or
goes further to the negative end, the predicted coefficient values. However, unlike linear
value of y will be 0 and vice versa. In other regression, the output value modeled here is a
words, if the output of the sigmoid function is binary value (0 or 1) rather than a numeric
0.65, it implies that there are 65% chances of value.
the event occurring; a coin toss, for example.
The sigmoid function is referred to as an
activation function for logistic regression and is
defined as:
where,
• e = base of natural logarithms
• value = numerical value one wishes to
transform
30
 While implementing logistic regression, one
needs to keep in mind the following key
assumptions:
2. Little or no multicollinearity between the
predictor/explanatory variables
3. Linear relationship of independent variables
to log odds
4. Prefers large sample size
5. Problem with extreme outliers
6. Consider independent observations

The assumption can be verified by plotting

residuals against time, which signifies the order
of observations. The plot helps in determining
the presence or absence of a random pattern.
If a random pattern is present or detected, this
assumption may be considered violated.
Key properties of the logistic regression
equation Types of Logistic Regression with Examples
Typical properties of the logistic regression Logistic regression is classified into binary,
equation include: multinomial, and ordinal. Each type differs from
• Logistic regression’s dependent variable the other in execution and theory. Let’s
obeys ‘Bernoulli distribution’ understand each type in detail.
• Estimation/prediction is based on 1. Binary logistic regression
‘maximum likelihood.’ Binary logistic regression predicts the
• Logistic regression does not evaluate the relationship between the independent and
coefficient of determination (or R binary dependent variables.
squared) as observed in linear 2. Multinomial logistic regression
regression’. Instead, the model’s fitness is A categorical dependent variable has two or
assessed through a concordance. more discrete outcomes in a multinomial
regression type. This implies that this regression
For example, KS or Kolmogorov-Smirnov type has more than two possible outcomes.
statistics look at the difference between Examples:
cumulative events and cumulative non-events 1. Let’s say you want to predict the most
to determine the efficacy of models through popular transportation type for 2040.
credit scoring. Here, transport type equates to the
Logistic Regression Best Practices
dependent variable, and the possible
1. The dependent/response variable is binary or outcomes can be electric cars, electric
dichotomous trains, electric buses, and electric bikes.

AE | 2D 31
 2. Predicting whether a student will join a
college, vocational/trade school, or
corporate industry.
3. Estimating the type of food consumed
by pets, the outcome may be wet food,
dry food, or junk food.
3.Ordinal logistic regression
Ordinal logistic regression applies when the
dependent variable is in an ordered state (i.e.,
ordinal). The dependent variable (y) specifies
an order with two or more categories or levels.
Examples: Dependent variables represent,
1. Formal shirt size: Outcomes = In this figure, model B represents a better fit
XS/S/M/L/XL than model A. This is because, although model
2. Survey answers: Outcomes = A shows high variability, model B seems to be
Agree/Disagree/Unsure more precise.
3. Scores on a math test: Outcomes = 4. Appropriately interpret the results
Poor/Average/Good A logistic model is accurate when it has a fine-
tuned build strategy and when the
Logistic Regression Best Practices for 2022 interpretation of the results produced by it is
Logistic regression can produce an accurate made right. Generally, a model is rated purely
Representation of Two Logistic Regression by analyzing the statistical significance of the
model if some best practices are followed, from
Models estimates. However, not much attention is
independent variable selection and choice of
model building strategy to validating the model given to the magnitude of the coefficients.
1. Identify dependent variables to ensure the Thus, interpreting the coefficients and
results.
model’s consistency discussing how the results relate to the research
2. Discover the technical requirements of the hypothesis or question is one of the good
model practices for logistic regression.
Coefficients are easy to interpret in linear
3. Estimate the model and evaluate the regression but not in logistic regression, as the
goodness of the fit estimates produced in the latter are not as
intuitive. In logistic type regression, the logit
transformation reveals the independent
variable’s impact on the variation of the
dependent variable’s natural logarithm of the
odds.

AE | 2D 32
5. Validate observed results
Another critical practice that researchers can
implement is validating the observed results
with a subsample of the original dataset. This
practice makes the model results more reliable,
especially when working with smaller samples.
6.Analysis of Variance (ANOVA)
ANOVA stands for Analysis of Variance. It is a
statistical method used to analyze the
differences between the means of two or more
groups or treatments. It is often used to
determine whether there are any statistically
significant differences between the means of
different groups.
Analysis of variance (ANOVA) is an analysis tool
used in statistics that splits an observed
aggregate variability found inside a data set
into two parts: systematic factors and random
factors. The systematic factors have a statistical
influence on the given data set, while the
random factors do not. Analysts use the
ANOVA test to determine the influence that
independent variables have on the dependent
variable in a regression study.
The t- and z-test methods developed in the 20th
century were used for statistical analysis until
1918, when Ronald Fisher created the analysis
of variance method. ANOVA is also called the
Fisher analysis of variance, and it is the
extension of the t- and z-tests. The term
became well-known in 1925, after appearing in
Fisher's book, "Statistical Methods for Research
Workers."3 It was employed in experimental
AE | 2D
psychology and later expanded to subjects
that were more complex.
Key Takeaways
• Analysis of variance, or ANOVA, is a
statistical method that separates
observed variance data into different
components to use for additional tests.
• A one-way ANOVA is used for three or
more groups of data, to gain
information about the relationship
between the dependent and
independent variables.
• If no true variance exists between the
groups, the ANOVA's F-ratio should
equal close to 1.
Formula for ANOVA
where:
F = ANOVA coefficient
MST = Mean sum of squares due to treatment
MSE = Mean sum of squares due to error
What Does the Analysis of Variance Reveal?
The ANOVA test is the initial step in analyzing
factors that affect a given data set. Once the
test is finished, an analyst performs additional
testing on the methodical factors that
measurably contribute to the data set's
inconsistency. The analyst utilizes the ANOVA
test results in an f-test to generate additional
data that aligns with the proposed regression
models.
The ANOVA test allows a comparison of more
than two groups at the same time to determine
whether a relationship exists between them.
The result of the ANOVA formula, the F statistic
(also called the F-ratio), allows for the analysis
of multiple groups of data to determine the
variability between samples and within
samples.
If no real difference exists between the tested
groups, which is called the null hypothesis, the
result of the ANOVA's F-ratio statistic will be
close to 1. The distribution of all possible values
of the F statistic is the F-distribution. This is
actually a group of distribution functions, with
two characteristic numbers, called the
numerator degrees of freedom and the
denominator degrees of freedom.
Example of How to Use ANOVA
A researcher might, for example, test students
from multiple colleges to see if students from
one of the colleges consistently outperform
students from the other colleges. In a business
application, an R&D researcher might test two
different processes of creating a product to see
if one process is better than the other in terms
of cost efficiency.
The type of ANOVA test used depends on a
number of factors. It is applied when data
needs to be experimental. Analysis of variance
is employed if there is no access to statistical
software resulting in computing ANOVA by
hand. It is simple to use and best suited for small
samples. With many experimental designs, the
sample sizes have to be the same for the
various factor level combinations.
33
ANOVA is helpful for testing three or more
variables. It is similar to multiple two-sample t-
tests. However, it results in fewer type I errors
and is appropriate for a range of issues. ANOVA
groups differences by comparing the means of
each group and includes spreading out the
variance into diverse sources. It is employed
with subjects, test groups, between groups and
within groups.
One-Way ANOVA Versus Two-Way ANOVA
There are two main types of ANOVA: one-way
(or unidirectional) and two-way. There also
variations of ANOVA. For example, MANOVA
(multivariate ANOVA) differs from ANOVA as
the former tests for multiple dependent
variables simultaneously while the latter
assesses only one dependent variable at a
time. One-way or two-way refers to the number
of independent variables in your analysis of
variance test. A one-way ANOVA evaluates the
impact of a sole factor on a sole response
variable. It determines whether all the samples
are the same. The one-way ANOVA is used to
determine whether there are any statistically
significant differences between the means of
three or more independent (unrelated) groups.
A two-way ANOVA is an extension of the one-
way ANOVA. With a one-way, you have one
independent variable affecting a dependent
variable. With a two-way ANOVA, there are
two independents. For example, a two-way
ANOVA allows a company to compare worker
productivity based on two independent
variables, such as salary and skill set. It is utilized
to observe the interaction between the two
AE | 2D
factors and tests the effect of two factors at the
same time.
How does ANOVA differ from a T test?
ANOVA differs from T tests in that ANOVA can
compare three or more groups while T tests are
only useful for comparing two groups at one
time.
What is Analysis of Covariance (ANCOVA)?
Analysis of Covariance combines ANOVA and
regression. It can be useful for understanding
within-group variance that ANOVA tests do not
explain.
Does ANOVA rely on any assumptions?
Yes, ANOVA tests assume that the data is
normally distributed and that the levels of
variance in each group is roughly equal. Finally,
it assumes that all observations are made
independently. If these assumptions are not
accurate, ANOVA may not be useful for
comparing groups.
The Bottom Line
ANOVA is a good way to compare more than
two groups to identify relationships between
them. The technique can be used in scholarly
settings to analyze research or in the world of
finance to try to predict future movements in
stock prices. Understanding how ANOVA works
and when it may be a useful tool can be
helpful for advanced investors.
a. Post Hoc Test
A priori comparisons are performed before the
data are collected, and post-hoc (or a
posteriori) comparisons are done after the data
have been collected. When the null hypothesis
of an analysis of variance (ANOVA) model is
rejected, post-hoc tests are used to identify the
population means that are different.
Post hoc in Latin means ‘after this’. Simply put, a
post-hoc analysis refers to a statistical analysis
specified after a study has been concluded
and the data collected. A post-hoc test is done
to identify exactly which groups differ from
each other. Therefore, such tests are also called
multiple comparison tests.
Purpose of post-hoc test
When the null hypothesis is rejected in an
omnibus test (a test that provides overall results
for study data, e.g. ANOVA), it means that at
least one parameter is significant. The question
to be asked at this stage is:
Which groups significantly differ from the others
in terms of the mean?
Types of studies that commonly use post-hoc
analysis
A post-hoc analysis can be conducted for
proportions and frequencies, but it is mostly
used for testing mean differences. The following
types of research involve post-hoc analyses.
A. In any discipline, studies investigating
differences between groups will use post-hoc
tests when the null hypothesis of an ANOVA
model is rejected.
Here is an example.
A researcher wants to investigate differences in
the effectiveness of TikTok, Instagram and
Facebook influencers in promoting a
nutraceutical brand. Let’s say that, by ANOVA,
the null hypothesis (that all three influencer
types have similar effectiveness) is rejected. A
34
post-hoc pairwise comparison may then reveal
that Instagram influencers have a significantly
higher effectiveness in promoting the brand
than TikTok and Facebook influencers, while the
latter two are similar.
B. In medicine, post-hoc analyses may be used
in clinical trials if the original hypothesis does
not hold (e.g. the primary outcome being the
antidiabetic effect of a drug).
Triallists then re-examine the dataset for other
outcomes (not originally planned, e.g.
improvement in renal outcomes in diabetes
patients) and perform statistical analysis to
determine other valuable results from the trial.
Note: For most clinical trials, the research
questions and statistical tests must be defined
before observing the research outcomes, even
before the first patient is enrolled. Primary,
secondary and exploratory outcome measures
should be established beforehand, while post-
hoc outcome measures can be specified after
the trial has started. This ‘pre-registration’ avoids
the practice of outcome switching (reporting
something different from what was originally
planned). Pre-specified and post-hoc outcome
measures must be clearly indicated in the
analysis section, in a way that makes it possible
to readily distinguish between them.
C. Analyses of pooled data from completed
trials comprise a type of post-hoc study as well.
a.1 Bonferrone Test
The Bonferroni test is a type of multiple
comparison test used in statistical analysis.
When performing a hypothesis test with multiple
comparisons, eventually a result could occur
that appears to demonstrate statistical
AE | 2D
significance in the dependent variable, even
when there is none.
If a particular test, such as a linear regression,
thus yields correct results 99% of the time,
running the same regression on 100 different
samples could lead to at least one false-
positive result at some point. The Bonferroni test
attempts to prevent data from incorrectly
appearing to be statistically significant like this
by making an adjustment during comparison
testing.
Key Takeaways
• The Bonferroni test is a statistical test
used to reduce the instance of a false
positive.
• In particular, Bonferroni designed an
adjustment to prevent data from
incorrectly appearing to be statistically
significant.
• An important limitation of Bonferroni
correction is that it may lead analysts to
mix actual true results.
Understanding the Bonferroni Test
The Bonferroni test, also known as "Bonferroni
correction" or "Bonferroni adjustment" suggests
that the p-value for each test must be equal to
its alpha divided by the number of tests
performed.
The Bonferroni test is a multiple-comparison
correction used when several dependent or
independent statistical tests are being
performed simultaneously. The reason is that
while a given alpha value may be appropriate
for each individual comparison, it is not
appropriate for the set of all comparisons. In
order to eliminate multiple spurious positives,
the alpha value needs to be lowered to
account for the number of comparisons being
performed.
The test is named for the Italian mathematician
who developed it, Carlo Emilio Bonferroni
(1892–1960).1 Other types of multiple
comparison tests include Scheffé's test and the
Tukey-Kramer method test. A criticism of the
Bonferroni test is that it is too conservative and
may fail to catch some significant findings.
In statistics, a null hypothesis is essentially the
belief that there's no statistical difference
between two data sets being compared.
Hypothesis testing involves testing a statistical
sample to confirm or reject a null hypothesis.
The test is performed by taking a random
sample of a population or group. While the null
hypothesis is tested, the alternative hypothesis is
also tested, whereby the two results are
mutually exclusive.
However, with any testing of a null hypothesis,
there's the expectation that a false-positive
result could occur. This is formally called a Type
I error, and as a result, an error rate that reflects
the likelihood of a Type I error is assigned to the
test. In other words, a certain percentage of
the results will likely yield a false positive.
Using Bonferroni Correction
For example, an error rate of 5% might typically
be assigned to a statistical test, meaning that
5% of the time there will likely be a false
positive. This 5% error rate is called the alpha
level. However, when many comparisons are
being made in an analysis, the error rate for
35
each comparison can impact the other results,
creating multiple false positives.
Bonferroni designed his method of correcting
for the increased error rates in hypothesis
testing that had multiple comparisons.
Bonferroni's adjustment is calculated by taking
the number of tests and dividing it into the
alpha value. Using the 5% error rate from our
example, two tests would yield an error rate of
0.025 or (.05/2) while four tests would therefore
have an error rate of .0125 or (.05/4). Notice
that the error rate decreases as the sample size
increases.
a.2 Scheffe Test
The Scheffé test is used to make unplanned
comparisons, rather than pre-planned
comparisons, among group means in an
analysis of variance (ANOVA) experiment. The
Scheffé test has the advantage of giving the
experimenter the flexibility to test any
comparisons that appear interesting.
The Scheffe Test (also called Scheffe’s
procedure or Scheffe’s method) is a post-hoc
test used in Analysis of Variance. It is named for
the American statistician Henry Scheffe. After
you have run ANOVA and got a significant F-
statistic (i.e. you have rejected the null
hypothesis that the means are the same), then
you run Sheffe’s test to find out which pairs of
means are significant. The Scheffe test corrects
alpha for simple and complex mean
comparisons. Complex mean comparisons
involve comparing more than one pair of
means simultaneously.
Out of the three mean comparisons test you
can run (the other two are Fisher’s LSD and
AE | 2D
Tukey’s HSD). The Scheffe test is the most
flexible, but it is also the test with the lowest
statistical power. Deciding which test to run
largely depends on what comparisons you’re
interested in:
• If you only want to make pairwise
comparisons, run the Tukey procedure
because it will have a narrower
confidence interval.
• If you want to compare all possible
simple and complex pairs of means, run
the Scheffe test as it will have a narrower
confidence interval.
Calculating the Test
Only run this test if you have rejected the null
hypothesis in an ANOVA test, indicating that the
means are not the same. Otherwise, the means
are equal and so there is no point in running this
test.
• The null hypothesis for the test is that all
means are the same: H0: μi = μj.
• The alternate hypothesis is that the
means are not the same: H0: μi ≠ μj.
Like most statistical tests, you find a critical
value and then compare it with a test statistic.
Reject the null hypothesis if the Scheffe test
statistic is greater than the critical value.
The steps that follow are for calculating simple
pairwise comparisons by hand for four samples
A B C and D. You’ll need descriptive statistics
data (the sample means) plus data from
ANOVA (like mean squared error).
Sample Means for Samples ABCD:
• A = 36.00,
• B = 34.50,
• C = 35.60,
• D = 36.21.
ANOVA Data (from the ANOVA output)
• F-value = 3.24 (from the f-table, with 3
numerator degrees of freedom and 16
denominator degrees of freedom at an
α of 0.05).
• Mean Squared Error: 0.028.
Steps
Step 1: Calculate the absolute values of pair
wise differences between sample means. You’ll
have to figure out all the possible combinations.
For four samples, there are 6 possible
combinations of two: AB AC AD BC BD and CD.
For example, for AB the absolute difference
(|A-B|) is 36.00 – 34.50 = 1.50.
Step 2: Use the following formula to find a set of
Scheffe formula values (you will use these in the
next Step):
Where:
• (k – 1) is the between samples degrees
of freedom. For this set of data, that’s 3
(4 – 1 = 3).
36
 • f , 1 * n − k k − ; α = the f-value (from
ANOVA). For this set of data, the f-value
is 3.24.
• MSE = mean square error (from ANOVA).
It’s easier than it looks! Once you’ve found the
df and MSE from your ANOVA output, you only
need to change the sample sizes in the formula
for each comparison. (The sample are A = 5, B
= 6, C = 4, D = 5).
For example, comparing A and B we get:
And comparing A and C we get:
a.3 Fisher LSD
Fisher's least significant difference (LSD)
procedure is a two-step testing procedure for
pairwise comparisons of several treatment
groups.
A one-way ANOVA is used to determine
whether or not there is a statistically significant
difference between the means of three or
more independent groups.
The hypotheses used in an ANOVA are as
follows:
Step 3: Compare the values in Step 1 to the
H0: The means are equal for each group.
values in Step 2 and find any values that are
HA: At least one of the means is different from
larger than the comparison values. Larger
the others.
values are statistically significant at your chosen
If the p-value from the ANOVA is less than some
alpha level (in this case, 0.05). For this example,
significance level (like α = .05), we can reject
the only value that isn’t significant is AD,
the null hypothesis and conclude that at least
because 0.21 is smaller than 0.33.
one of the group means is different from the
others.
But in order to find out exactly which groups are
different from each other, we must conduct a
post-hoc test.
AE | 2D
One commonly used post-hoc test is Fisher’s
least significant difference test.
To perform this test, we first calculate the
following test statistic:
where:
• t.025, DFw: The t-critical value from the t-
distribution table with α = .025 and DFw is
the degrees of freedom within groups
from the ANOVA table.
• MSW: The mean squares within groups
from the ANOVA table.
• n1, n2: The sample sizes of each group
We can then compare the mean difference
between each group to this test statistic. If the
absolute value of the mean difference
between two groups is greater than the test
statistic, we can declare that there is a
statistically significant difference between the
group means.
The following example shows how to perform
Fisher’s least significant difference test in
practice.
Example: Fisher’s Least Significance Difference
Test
Suppose a professor wants to know whether or
not three different studying techniques lead to
different exam scores among students. To test
this, she randomly assigns 10 students to use
each studying technique and records their
exam scores.
37
The following table shows the exam scores for
each student based on the studying technique
they used:
The professor performs a one-way ANOVA and
get the following results:
Since the p-value in the ANOVA table (.018771)
is less than .05, we can conclude that not all of
AE | 2D
the mean exam scores between the three
groups are equal.
Thus, we can proceed to perform Fisher’s least
significant difference test to determine which
group means are different.
Using values from the output of the ANOVA, we
can calculate Fisher’s test statistic as:
We can then calculate the absolute mean
difference between each group:
The absolute mean differences between
technique 1 vs. technique 2 and technique 1
vs. technique 3 are greater than Fisher’s test
statistic, thus we can conclude that these
techniques lead to statistically significantly
different mean exam scores.
Interpretation of the Result
We can also conclude that there is no
significant difference in mean exam scores
between technique 2 and technique 3.
a.4 Tukey LSD/Tukey’s Range Test
Tukey's range test, also known as Tukey's test,
Tukey method, Tukey's honest significance test,
or Tukey's HSD (honestly significant difference)
test, is a single-step multiple comparison
procedure and statistical test. It can be used to
find means that are significantly different from
each other.
Named after John Tukey, it compares all
possible pairs of means, and is based on a
studentized range distribution (q) (this
distribution is similar to the distribution of t from
the t-test. See below).
Tukey's test compares the means of every
treatment to the means of every other
treatment; that is, it applies simultaneously to
the set of all pairwise comparisons
and identifies any difference between two
means that is greater than the expected
standard error. The confidence coefficient for
the set, when all sample sizes are equal, is
exactly 1 - α for any 0 ≤ α ≤ 1 . For unequal
sample sizes, the confidence coefficient is
greater than 1 − α. In other words, the Tukey
method is conservative when there are
unequal sample sizes.
This test is often followed by the Compact
Letter Display (CLD) statistical procedure to
render the output of this test more transparent
to non-statistician audiences.
Assumptions
1. The observations being tested are
independent within and among the
groups.
38
 2. The groups associated with each mean
in the test are normally distributed.
3. There is equal within-group variance
across the groups associated with each
mean in the test (homogeneity of
variance).
The test statistic
Tukey's test is based on a formula very similar to
that of the t-test. In fact, Tukey's test is
essentially a t-test, except that it corrects for
family-wise error rate.
The formula for Tukey's test is
where YA is the larger of the two means being
compared, YB is the smaller of the two means
being compared, and SE is the standard error
of the sum of the means.
This qs value can then be compared to a q
value from the studentized range distribution. If
the qs value is larger than the critical value qα
obtained from the distribution, the two means
are said to be significantly different at level α : 0
≤ α ≤ 1.
Since the null hypothesis for Tukey's test states
that all means being compared are from the
same population (i.e. μ1 = μ2 = μ3 = ... = μk), the
means should be normally distributed
(according to the central limit theorem). This
gives rise to the normality assumption of Tukey's
test.
The studentized range (q) distribution
The Tukey method uses the studentized range
distribution. Suppose that we take a sample of
AE | 2D
size n from each of k populations with the same
normal distribution N(μ, σ2) and suppose that
ymin is the smallest of these sample means
and ymax is the largest of these sample means,
and suppose S2 is the pooled sample variance
from these samples. Then the following random
variable has a Studentized range distribution.
This value of q is the basis of the critical value of
q, based on three factors:
1. α (the Type I error rate, or the probability
of rejecting a true null hypothesis)
2. k (the number of populations)
3. df (the number of degrees of freedom
(N – k) where N is the total number of
observations)
The distribution of q has been tabulated and
appears in many textbooks on statistics. In some
tables the distribution of q has been tabulated
without the factor. To understand which
table it is, we can compute the result for k = 2
and compare it to the result of the Student's t-
distribution with the same degrees of freedom
and the same α. In addition, R offers a
cumulative distribution function (ptukey) and a
quantile function (qtukey) for α.
Confidence limits
The Tukey confidence limits for all pairwise
comparisons with confidence coefficient of at
least 1 − α are
where n i and n j are the sizes of groups i and j
respectively. The degrees of freedom for the
whole design is also applied.
Comparing ANOVA and Tukey–Kramer tests
Both ANOVA and Tukey–Kramer tests are based
on the same assumptions. However, these two
tests for k groups (i.e. μ1 = μ2 = ... = μk) may result
in logical contradictions when k> 2, even if the
assumptions hold. It is possible to generate a set
of pseudorandom samples of strictly positive
measure such that hypothesis μ1 = μ2 is rejected
at significance level 1 – α > 0.95 while μ1 = μ2 =
μ3 is not rejected even at 1,- α = 0.975.
a.5 Dunnet Test
Dunnett’s Test (also called Dunnett’s Method or
Dunnett’s Multiple Comparison) compares
means from several experimental groups
against a control group mean to see is there is
a difference. When an ANOVA test has
significant findings, it doesn’t report which pairs
of means are different. Dunnett’s can be used
after the ANOVA has been run to identify the
pairs with significant differences.
One fixed “control” group is compared to all of
the other samples, so it should only be used
when you have a control group. If you don’t
have a control group, use Tukey’s Test.
Calculation
As Dunnett’s compares two groups, it acts
similarly to a t-test. The following formula gives
you a value that you can use to compare
mean differences. The formula is:
39

Step 1: Look up the tDunnett critical value in the
Dunnett-critical value table. You’ll need:
• Your chosen alpha level (usually 5%),
• Sample size, n,
• Degrees of freedom from the ANOVA
“Within Groups” output. This is the “n” in
the left hand column of the table.
For example, let’s say you had an alpha level of
5%, a sample size of 5 and 20 df. The Dunnett
critical value is 2.65.
Plug the value into the formula:
Step 2: Find the mean squares (MS) of the within
group in the ANOVA source table. Plug that
value into the above formula. For example, let’s
say your MS within is 15.39:
AE | 2D
Step 3: Find “n”, the number in one group. For
this example, that’s 5:
Solving the formula, we get:
DDunnett = 2.65 * 2.481 = 6.575.
The answer (6.575) is the critical distance
between means. If the distance between a
control group mean and an experimental
group mean is greater than 6.575, then that
distance is significant. For example, let’s say you
had a control group mean of 52 and three
experimental group means of 55, 64 and 72.
The differences between means (absolute
values) are:
• Group 1 55 – 52 = 3
• Group 2 64 – 52 = 12
• Group 3 72 – 52 = 20
Interpretation of the Result
Only groups 2 and 3 exceed the Dunnett’s
distance of 6.575, so they are the significant
pairs.
40