Tutorial: Statistical Tests
for Mean Differences
Number Analytics
Coca cola vs Pepsi, taste better?
How to test blind taste test results
Is there significant taste difference?
Statistical Tests for Differences
Choose the correct statistic test
Area of Applica Level Sca
Subgroups Test Example
tion ling
Hypotheses
Do customer industry types di
about Nominal All Chi-square
ffer by company size ?
frequency
One Sample Is the purchase frequency
One
T-test different from 1.5?
Is the purchase frequency
Independent greater for email promotion
Samples T-test responders than that for non-r
Metric
esponders?
Hypotheses (Interval Two
about means or Is the average brand score diff
ratio) Paired Sample
erent after exposure to a new
T-test
TV ad? (before and after)
One-way Is the purchase frequency
Three or more
ANOVA different by company size?
One Sample T-test
One Sample T-test
Is the overall rating significantly
different than 4?
You can choose your own file
by uploading it to the cloud.
One Sample T-test
3 STEPS! Easy to apply!
Then manually enter the
First select the test variable test value and choose th
in your data file e sided test.
Now click on ‘Run’!
One Sample t-test
Conclusion: Rating is not different
from 4 on average.
Independent Samples T-test
Independent Samples T-test
Now we’re interested to know whether
the rating for female group is
significantly different from rating for
male group.
Independent Samples T-test
Select the test variable
Choose the groupin
g variable
Let’s run results!
Independent Samples T-test
Paired sample t-test
Paired Sample T-test
Unlike the independent samples T-test
, paired sample T-test is applied to the
same group comparing two outcome
s like music preference rating scores
across different songs
Independent Samples T-test
Select the test variable 1
Choose the test variable 2
Let’s run results!
Independent Samples T-test
ANOVA
ANOVA
How about the rating among
different ethnical groups?
ANOVA
ANOVA
P value 0.32 (>0.05)
Conclusion: Rating is indifferent
across ethnicity.
P Value: Exact Probability of getting
a computed test statistic that is due
to chance. The smaller the p value,
the smaller the probability that the
observed result occurred by chance
Cross-tab (Chi-square test)
Cross-tab (Chi-square test)
• Cross-tab is a frequency table of two
or three variables
• Used to examine association between
two or 3 variables (usually 2)
• H0: there is a relation between variable X
and variable Y
• Variables take a limited number of
values, for example:
Consumers: gender, ethnicity
Business: industry, company size
Cross-tab (Chi-square test)
Cross-tab (Chi-square test)
P value greater than 0.05, reject H0.
Conclusion: There is no relation between
gender and ethnicity