1.
Types of Variables
● Quantitative (Scale):numeric values (e.g., scores,age, salary)
● Categorical (Nominal/Ordinal):groups or categories(e.g., gender, department)
Understanding your variable type isessentialto selecting the correct test.
2. Types of Hypotheses
● Null Hypothesis (H₀):No difference or no effect
● Alternative Hypothesis (H₁):There is a difference or effect
● Hypothesis testing relies onp-valuesto decide whether to reject H₀
3. Confidence Intervals
● A
95% confidence intervalgives a range in which we’re95% confident thetrue
population parameterlies
● Used to support hypothesis test results
4. How to Choose the Right Statistical Test
You choose the test based on:
● Thetype of variables(categorical vs. scale)
● Thenumber of groupsor predictors
● Whether you're testing for adifference, anassociation,or aprediction
Includes:
● T-tests
● ANOVA
● Regression (simple, multiple, logistic)
● Chi-square
5. Good Practices & Warnings
● Alwayscheck assumptions: normality, independence,sample size
● Don't run tests blindly — first explore the data (descriptive stats, graphs)
● Understand whateach test result actually means(statisticalvs. practical significance)
e notice here that theaverage Test_Scorein the population is significantly different from the
W
reference value of 50.
To validate this using aOne-Sample T-Test, we testthe following hypotheses:
● H
₀: There is no significant difference between thepopulation’s average score and the
reference value of 50.
● H
₁: There is a significant difference between thepopulation’s average score and the
reference value of 50.
ith at-value of 56.543,p < .001, and amean differenceof 26.09, we clearly reject the null
W
hypothesis.
This means that the average Test_Score (M = 76.09) isstatistically and substantially higher
than 50.
o based on the result of our t-test we can say the valid hypothese isH1because the p value is <
S
0.05
e notice here that there’sno significant differencebetween male and female participants in
W
terms of their average test scores (M = 75.98 for males, M = 76.19 for females).
o confirm this result using anIndependent SamplesT-Test, we evaluate the following
T
hypotheses:
● H
₀: There is no significant difference in the mean Test_Score between male and female
participants.
● H
₁: There is a significant difference in the meanTest_Score between male and female
participants.
Based on the results:
● T
heLevene’s Testfor equality of variances has a p-value of.967, so we assumeequal
variances.
● Thet-test p-valueis.816, which isgreater than 0.05.
herefore, wefail to reject the null hypothesis(H₀), and conclude thatthere is no statistically
T
significant differencebetween male and female test scores in this sample.
e observe that the meanPre_Training_Scoreis60.84,while thePost_Training_Scoreis
W
65.89.
Based on this initial data, we can say that trainingmight have hadan effect on scores.
To confirm this result, we apply apaired-samplest-testand test the following hypotheses:
● H₀: Training hadno effecton scores.
● H ₁: Training hadan effecton scores.
res
● Themean differenceis-5.05
● Thep-valueis.000, which isless than 0.05
ased on the p-value (< 0.05), wereject H₀and conclude thattraining had a
B
statistically significant effect on scores.
sing
U Satisfaction_Score , we compared the mean satisfactionlevels across
Department groups (HR, IT, Marketing) using a One-Way ANOVA with Tukey post hoc
test.
The post hoc results show that:
● HR has the highest satisfaction score
● IT is in the middle
● Marketing has the lowest
ll pairwise comparisons arestatistically significant(p < .001), as indicated by theSig. column
A
in the Tukey test.
here aresignificant differencesin satisfaction levels betweenall three
T
departments. Each department’s average satisfaction score issignificantly
differentfrom the others.
e used a binary logistic regression to predict the likelihood of being hired based on three
W
variables:GPA,Work_Experience, andInterview_Score.
he overall model isstatistically significant(Chi²= 15.717, p = .001), which means that at least
T
one predictor has an effect. The model explains about17.9%of the variation in hiring
(Nagelkerke R² = .179).
Looking at the individual predictors:
● I nterview_Scoreis statistically significant (p =.001), so we can say it has areal effect
on the chance of being hired. The higher the score, the higher the odds (Exp(B) = 1.13).
● Work_Experienceisborderline(p = .053), so it mighthave a small effect.
● GPAisnot significant(p = .965), so it doesn't influence hiring in this model.
Conclusion:Interview performance is the best predictorof being hired in this case.
We used asimple linear regressionto predictSalarybased onYears_Experience.
he model isstatistically significant(p = .000) and explains avery large portion of the
T
variationin salary (R² = .870). This means that87%of the variation in salarycan be
explained by years of experience.
Looking at the coefficients:
● Thep-value for Years_Experience is .000, so it'sstatistically significant.
● T
heB coefficient is 4103.86, which means thateach additional year of experience
increases salary by approximately 4104 units.
Conclusion:
ears of experience has a strong and significant impact on salary.The more
Y
experience someone has, the higher their salary is likely to be.
e tested the association betweenSmoking_StatusandExercise_Levelusing theChi-Square
W
Test of Independence.
● H₀: There is no association between smoking and exerciselevel.
● H₁: There is an association between smoking and exerciselevel.
Looking at the table:
● TheChi-square value = 50.347, withdf = 2
● Thep-value = .000→ which isless than 0.05
Conclusion:
e reject H₀. There is astatistically significant associationbetween
W
Smoking_StatusandExercise_Level.
In other words,exercise habits differ depending on whether someone smokes or
not.
95% confidence interval means that if we repeated the study many times, about 95% of
A
the calculated intervals would contain the true population parameter.
e use a t-test when comparing the means oftwo groups, and a one-way ANOVA when
W
comparing the means acrossthree or more groups.
One way anova.
o compare the means between two groups, you should use a t-test (specifically, an
T
independent samples t-test).
redictor
P utcome
O Test to Use
(Independent) (Dependent)
Quantitative Quantitative imple or Multiple Linear
S
Regression
Quantitative Categorical Logistic Regression
Categorical Quantitative -Test(if 2 groups) orANOVA(if
T
3+ groups)
Categorical Categorical Chi-Square Test (χ²)
Chi-Square (χ²) Test — When and How (in SPSS)
Use when:
● Both variables arecategorical
● You want to testassociationorindependence
SPSS Steps:
pgsql
CopierModifier
Analyze → Descriptive Statistics → Crosstabs → Statistics
→ Chi-square
H1 < 0.05- there’s a difference
H0> there’s no difference
On sample t-test= comparing a number
Independent t-test eg: gender
Paired t-test = same sample before and after
#
Question
Test to Use
Reason
1
Using the variable Stress_Score, One-Sample
You're
test whether the average is
T-Test
comparing a
significantly different from 50.
sample mean to
a fixed value
(50)
2
Using Test_Score as the outcome, Independent
Two independent
compare the means between Male
Samples
groups (Male vs
and Female participants.
T-Test
Female),
quantitative
outcome
3
Using Pre_Training_Score and Paired
Same
Post_Training_Score, evaluate
Samples
participants
whether training had an effect
T-Test
tested before
on scores.
and after
training
4
Use Satisfaction_Score to One-Way
Comparing means
compare mean satisfaction levels
ANOVA
across 3+
across Department groups (HR,
groups
IT, Marketing).
5
Model the likelihood of being Binary
Outcome is
Hired using predictors GPA,
Logistic
binary (Hired:
Work_Experience, and
Regression
Yes/No),
Interview_Score.
predictors are
quantitative
6
Using Years_Experience to Simple
One
predict Salary.
Linear
quantitative
Regression
predictor and
one
quantitative
outcome
7
Test whether there is an Chi-Square
Both variables
association between
Test of
are categorical
Smoking_Status and
Independence
– checking
Exercise_Level.
(χ²)
association