PSYCHOLOGICAL STATISTICS

INTRODUCTION TO T-TEST
ANALYSIS

PRESENTED BY: JOHN KENNETH PADEL

 01 What is t-test?

Introduction to terminologies and definitions used in t-test analysis.

 t-test

A t-test is an inferential statistic used to determine if

there is a significant difference between the means of
two groups and how they are related. T-tests are used
when the data sets follow a normal distribution and
have unknown variances.
 Assumptions of t-test

1. Our dependent variable should be measured at the

interval or ratio level (i.e., continuous).
2. There should be no significant outliers.
3. Your dependent variable should be approximately
normally distributed.
 Hypothesis of t-test
H0: There is no significant differences between group
A and group B.
Group A = Group B

H1: There is a significant differences between group

A and group B.
Group A  Group B
 t-value or t-statistics

The t-value, also known as the t-statistic, is a measure of

how large the difference between two sample means is
relative to the variability within the samples.
The t-value, also known as the t-statistic, is a measure of
how large the difference between two sample means is
relative to the variability within the samples.
 t-value or t-statistics
If the absolute value of the t-value is large, it suggests that
the difference between the means is not likely to have
occurred by chance alone, and thus the groups are likely to
be different in a meaningful way.
On the other hand, if the absolute value of the t-value is
small, it suggests that the difference between the means
may be due to chance, and thus the groups may not be
meaningfully different.
 t-value or t-statistics

The t-value is often compared to a critical value based on

the degrees of freedom and the level of significance
(alpha) to determine whether the difference between the
means is statistically significant or not.
 Critical Value

The critical value is a value used in hypothesis testing that

defines the boundary between the region where the null
hypothesis can be rejected and the region where it cannot
be rejected. In a t-test, the critical value is determined
based on the degrees of freedom and the level of
significance (alpha).
 Critical Value

If the absolute value of the t-value is greater than the

critical value, then the null hypothesis is rejected and the
alternative hypothesis is accepted. If the absolute value of
the t-value is less than the critical value, then the null
hypothesis is not rejected.
 Degrees of Freedom

Degrees of freedom refers to the number of values in a

sample that are free to vary without changing the sample
mean.
For example, let's say you have a sample of five numbers: 4, 6,
8, 10, and 12. The mean of this sample is 8. If you change any
one of the values, the mean of the sample will also change.
However, if you fix the mean of the sample at 8, you can only
change four of the values freely, since changing the fifth value
will affect the mean. Thus, the degrees of freedom for this
sample is 4, because you have four values that are free to vary
without changing the sample mean.
 Level of Significance

The level of significance, also known as alpha, is the

probability of making a Type I error, which is the rejection
of the null hypothesis when it is actually true. The level of
significance is typically set at 0.05, which means that there
is a 5% chance of making a Type I error.
 Level of Significance
The level of significance determines how confident we
want to be in our decision to reject or not reject the null
hypothesis.
If the p-value is less than or equal to the level of
significance, then the null hypothesis is rejected. If the p-
value is greater than to the level of significance, then you
failed to reject the null hypothesis.
 Effect Size

Effect size tells you how meaningful the relationship

between variables or the difference between groups is. It
indicates the practical significance of a research outcome.
A large effect size means that a research finding has
practical significance, while a small effect size indicates
limited practical applications.
 Effect Size
According to Cohen's guidelines, a d value of 0.20 can be considered a small
effect size, a d value of 0.50 can be considered a medium effect size, and a d
value of 0.80 or higher can be considered a large effect size.

Positive Value: A value between 0 to 1 shows a direct relationship between the

two variables. Negative Value: A value between -1 to 0 indicates an inverse
relationship between the two variables. Zero Value: When the result is 0, there is
no relationship between the two variables.
 Mean Difference

The mean difference, or difference in means,

measures the absolute difference between the
mean value in two different groups. In clinical
trials, it gives you an idea of how much difference
there is between the averages of the
experimental group and control groups.
 Test of Normality
Shapiro-Wilk and Kolmogorov Smirnov are two most popular
test of normality.
The Shapiro–Wilk test is more appropriate method for small
sample sizes (<50 samples) although it can also be handling on
larger sample size while Kolmogorov–Smirnov test is used for N
≥ 50. For both of the above tests, null hypothesis states that
data are taken from normal distributed population.
 Test of Homogeneity

Levene's test is used to test if the samples have equal variances.

Equal variances across samples is called homogeneity of
variance. Some statistical tests (e.g., t-test and ANOVA) assume
that variances are equal across groups or samples. The Levene’s
test can be used to verify that assumption.
 Test of Homogeneity

Levene's test is used to test if the samples have equal variances.

Equal variances across samples is called homogeneity of
variance. Some statistical tests (e.g., t-test and ANOVA) assume
that variances are equal across groups or samples. The Levene’s
test can be used to verify that assumption.
 Test of Homogeneity
When we say that the data is homogenous, it means that the
variance or spread of scores is similar or equal across groups
being compared.
Homogeneous data is important because it ensures that the
comparison of means between the groups is not biased due to
differences in variance. If the variance is not homogenous, then it
can lead to inflated or deflated results and may affect the
accuracy of the t-test.
 1 is true dichotomous? Phi Coefficient
Chi-square Between 2
Test of dichotomous? Both are artif. dichotomous? Tetrachoric Coefficient
Independence
Did you conducted NO Between 1 True dichotomous? Point-Biserial Correlation
a random continuous and 1
sampling? Between 2 Artificial dichotomous?
Nonparametric dichotomous? Biserial Correlation
nominal
Tests data? Small sample? Kendall Tau-b
Between 2
ordinal data?
Do you want to Large sample? Spearman Rho
YES NO see the
Simple Linear relationship? Between 2 continuous data? Pearson’s r
Regression
Is the data But I have two
Two-way 1?
homogenous and 1? independent One-way ANOVA
Do you want to ANOVA
normally distributed variables? Repeated
predict?
Measures
How many
Do you have How many
YES predictors? 2?
three or more groups do you One-way ANOVA
NO have?
sets of data?
2 or more?
Do you want to 1? Paired-sample
Multiple Linear compare Do you have How many t-test
Regression means? two sets of groups do you
data? have? Independent-
Is the data 2?
continuous (i.e., Parametric sample t-test
YES
intervals or ratio?) Tests Do you have
One-sample t-test
one set of data?
 One-Sample t-test

The one-sample t-test is used to determine whether a

sample comes from a population with a specific mean. This
population mean is not always known, but is sometimes
hypothesized.
SAMPLE PROBLEM FOR ONE- Battery Life (hours)
SAMPLE T-TEST: 1 18.5
2 20.2
3 22.1
4 19.6
A company produces batteries for electronic 5 17.9
6 21
devices and claims that the average battery 7 19.8
life is 20 hours. To test this claim, a sample 8 20.1
of 25 batteries is selected at random and 9 18.4
10 21.4
their battery life (in hours) is recorded. 11 19.7
12 17.6
13 22
H0: There is no significant difference 14 21.8
15 18.9
between the hypothesized battery life and 16 19.5
the actual battery life. 17 20.5
18 20.3
H1: There is a significant difference between 19 17.7
the hypothesized battery life and the actual 20 18.3
21 19.2
battery life. 22 20.9
23 18.8
24 19.1
25 20.7
A one-sample t-test was conducted to determine if the mean life of batteries produced by a
company is different from 20 hours. A sample of 25 batteries was taken, and the sample
mean was found to be 19.8 hours with a standard deviation of 1.33 hours.

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data followed a normal distribution (W = 0.969, p = 0.615.)

The results of the t-test revealed that there is no significant difference between the sample
mean and the hypothesized population mean of 20 hours (t[24] = -0.9, p = 0.377, 95% CI =
[-.790 – 0.310]), two-tailed. The effect size was small (Cohen's d = -0.180). Therefore, the
researchers failed to reject the null hypothesis, and that the hypothesized battery life is not
different from the actual battery life.
SAMPLE PROBLEM FOR ONE-SAMPLE T-TEST:

A nutritionist claims that their new weight loss program results in an average weight loss of at
least 15 pounds in one month. To test this claim, a random sample of 20 participants who
completed the program was taken, and their weight loss was recorded. The sample mean
weight loss was found to be 6.11 pounds with a standard deviation of 0.670 pounds.

Conduct a one-sample t-test to determine whether the mean weight loss in the program is
significantly different from 5 pounds at a 95% confidence level.

H0: The mean weight loss in the program is not significantly different from 15 pounds

H1: The mean weight loss in the program is significantly different from 15 pounds.
A one-sample t-test was conducted to determine whether the mean weight loss in the
program is significantly different from 5 pounds at a 95% confidence level. A sample of N =
20 participants was taken, and the sample mean was found to be M = 6.11 lbs. with a
standard deviation of SD = 0.670 lbs.

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data followed a normal distribution (W = 0.966, p = 0.679.)

The results of the t-test revealed that the mean weight loss in the program is significantly
different from 15 pounds (t[19] = -59.4, p < .001, MD = -8.89), two-tailed. The effect size was
large (Cohen's d = -13.1, 95% CI [-17.3 – 8.86]). Therefore, the researchers reject the null
hypothesis, and that the mean weight loss in the program is significantly different from 15
pounds.
Independent-Sample t-test

The independent-samples t-test (or independent t-test, for

short) compares the means between two unrelated groups
on the same continuous, dependent variable.
 SAMPLE PROBLEM FOR INDEPENDENT-SAMPLE T-TEST:

A researcher wants to test whether there is a significant difference in the average amount
of time it takes for two different groups of students to complete a standardized test.
Group A consists of 20 students who used a traditional paper-and-pencil format, while
Group B consists of 18 students who used a computerized format. The researcher wants
to use an independent samples t-test to compare the means of the two groups and
determine whether the difference is statistically significant.

H0: There is no significant difference between Group A and Group B in terms of the
average amount of time to complete a standardized test.

H1: There is a significant difference between Group A and Group B in terms of the
average amount of time to complete a standardized test.
An independent-sample t-test was conducted to determine whether Group A and Group B
are statistically different in terms of the average amount of time to complete a standardized
test at 95% confidence interval. The mean score for Group A was M = 23.4 (SD = 2.82, N =
20), and the mean score for Group B was M = 22.1 (SD = 2.29, N = 18).

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data followed a normal distribution (W = 0.978, p = 0.657). The
Levene's test for equality of variances was not significant (F[1, 36] = 1.02, p = 0.320),
indicating that the assumption of equal variances was met.

The results of the independent-sample t-test revealed that Group A and Group B are not
statistically different in terms of the average amount of time to complete a standardized test
(t[36] = 1.66, p = .105, MD = 1.39), two-tailed. The effect size was moderate (Cohen's d =
.540, 95% CI [-0.126 – 1.19]). Therefore, the researchers failed to reject the null hypothesis,
and that the average time to complete a standardized test of Group A and Group B are not
different from one another.
An independent-sample t-test was conducted to
determine whether Group A and Group B are statistically
different in terms of the average amount of time to
complete a standardized test at 95% confidence interval.
The mean score for Group A was M = 23.4 (SD = 2.82, N =
20), and the mean score for Group B was M = 22.1 (SD =
2.29, N = 18)
An independent-sample t-test was conducted to
determine whether Group A and Group B are statistically
different in terms of the average amount of time to
complete a standardized test at 95% confidence interval.
The mean score for Group A was M = 23.4 (SD = 2.82, N =
20), and the mean score for Group B was M = 22.1 (SD =
2.29, N = 18)
The normality assumption of the data was tested using
the Shapiro-Wilk test, and the results of the test indicated
that the data followed a normal distribution (W = 0.978, p
= 0.657). The Levene's test for equality of variances was
not significant (F[1, 36] = 1.02, p = 0.320), indicating that
the assumption of equal variances was met.
The normality assumption of the data was tested using
the Shapiro-Wilk test, and the results of the test indicated
that the data followed a normal distribution (W = 0.978, p
= 0.657). The Levene's test for equality of variances was
not significant (F[1, 36] = 1.02, p = 0.320), indicating that
the assumption of equal variances was met.
The results of the independent-sample t-test revealed that Group A and Group B are not
statistically different in terms of the average amount of time to complete a standardized test
(t[36] = 1.66, p = .105, MD = 1.39), two-tailed. The effect size was moderate (Cohen's d =
.540, 95% CI [-0.126 – 1.19]). Therefore, the researchers failed to reject the null hypothesis,
and that the average time to complete a standardized test of Group A and Group B are not
different from one another.
The results of the independent-sample t-test revealed that Group A and Group B are not
statistically different in terms of the average amount of time to complete a standardized test
(t[36] = 1.66, p = .105, MD = 1.39), two-tailed. The effect size was moderate (Cohen's d =
.540, 95% CI [-0.126 – 1.19]). Therefore, the researchers failed to reject the null hypothesis,
and that the average time to complete a standardized test of Group A and Group B are not
different from one another.
 SAMPLE PROBLEM FOR INDEPENDENT-SAMPLE T-TEST:

A company is interested in testing a new training program for their customer service
representatives. They randomly select 30 customer service representatives and assign
them to either the control group or the experimental group. The control group will
receive the usual training program, while the experimental group will receive the new
training program. After the training, the company wants to know if there is a significant
difference in the average customer satisfaction ratings between the two groups. The
customer satisfaction ratings are measured on a scale from 1 to 10.

H0: There is no significant difference between the control group and experimental group
in terms of the average customer satisfaction ratings.

H1: There is a significant difference between the control group and experimental group in
terms of the average customer satisfaction ratings.
An independent-sample t-test was conducted to determine if there is significant difference
between the control group and experimental group in terms of the average customer
satisfaction ratings at 95% confidence interval. The mean score for experimental group was
M = 8.5 (SD = 0.519, N = 14), and the mean score for control group was M = 5.88 (SD =
0.806, N = 16).

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data did not followed a normal distribution (W = 0.893, p =
0.006). The Levene's test for equality of variances was significant (F[1, 28] = 1.79, p = 0.192),
indicating that the assumption of equal variances was met.

The results of the independent-sample t-test revealed that there is a significant difference
between the control group and experimental group in terms of the average customer
satisfaction ratings (t[28] = 10.4, p = < .001, MD = 2.63), two-tailed. The effect size was large
(Cohen's d = 3.82, 95% CI [2.18 – 5.42]). Therefore, the researchers reject the null hypothesis,
and that the average customer satisfaction rating between the experimental and control
group are statistically different from one another.
 Dependent-Sample t-test

The dependent t-test (called the paired-samples t-test in

SPSS Statistics) compares the means between two related
data (gathered from one group) on the same continuous,
dependent variable.
 SAMPLE PROBLEM FOR PAIRED-SAMPLE T-TEST:

Suppose a psychologist is interested in determining whether a new therapy for reducing

anxiety is effective. She recruits 20 participants who all suffer from moderate levels of
anxiety. Each participant completes a standardized anxiety measure (out of 100) and is
then given the new therapy. After the therapy is completed, the same standardized
anxiety measure is given to each participant. The psychologist wants to know if there is a
significant difference in anxiety levels before and after the therapy.

H0: There is no significant difference in terms of the average anxiety level of participants
before and after the therapy.

H1: There is a significant difference in terms of the average anxiety level of participants
before and after the therapy.
A paired-sample t-test was conducted to determine if there is a significant difference in
terms of average anxiety levels before and after the therapy at 95% confidence interval. The
mean anxiety level before the therapy was M = 74 (SD = 11.1, N = 20), and the mean anxiety
level after the therapy was M = 65.8 (SD = 11.2, N = 20).

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data did not followed a normal distribution (W = 0.886, p =
0.023).

The results of the independent-sample t-test revealed that there is a significant difference in
terms of average anxiety levels before and after the therapy (t[19] = 11.7, p = < .001, MD =
8.30), two-tailed. The effect size was large (Cohen's d = 2.62, 95% CI [1.68 – 3.55]). Therefore,
the researchers reject the null hypothesis, and that the average anxiety levels of participants
before the treatment is significantly different from their anxiety levels after the treatment.
 SAMPLE PROBLEM FOR PAIRED-SAMPLE T-TEST:

A researcher is interested in investigating the effectiveness of a new exercise program in

improving cardiovascular health. The researcher randomly selects 25 participants with
pre-existing cardiovascular conditions and measures their resting heart rate before and
after completing the exercise program. The researcher wants to know if there is a
significant difference in the average resting heart rate before and after completing the
exercise program.

H0: There is no significant difference in terms of the average resting heart rate before and
after completing the exercise program.

H1: There is a significant difference in terms of the average resting heart rate before and
after completing the exercise program.
A paired-sample t-test was conducted to determine if there is a significant difference in the
average resting heart rate before and after completing the exercise program at 95%
confidence interval. The mean resting heart rate before the program was M = 10.4 (SD =
2.35, N = 25), and the mean resting heart rate after the program was M = 11.7 (SD = 2.54, N
= 25).

The normality assumption of the data was tested using the Shapiro-Wilk test, and the results
of the test indicated that the data did not followed a normal distribution (W = 0.887, p =
0.010).

The results of the independent-sample t-test revealed that there is a significant difference in
the average resting heart rate before and after completing the exercise program (t[24] = -
5.30, p = < .001, MD = -1.28), two-tailed. The effect size was large (Cohen's d = -1.06, 95%
CI [-1.54 – -0.560]). Therefore, the researchers reject the null hypothesis, and that the
average resting heart rate before the program is statistically different from the average
resting heart rate after completing the program.