See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/363793117

THE USE OF TWO-SAMPLE t-TEST IN THE REAL DATA

Article in Advances and Applications in Statistics · September 2022

DOI: 10.17654/0972361722071

CITATIONS READS
2 6,820

2 authors:

Mowafaq Muhammed Tawfeeq Al-kassab Aveen Hameed Majeed

Knowledge University Tishik International University
81 PUBLICATIONS 168 CITATIONS 1 PUBLICATION 2 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mowafaq Muhammed Tawfeeq Al-kassab on 03 July 2024.

The user has requested enhancement of the downloaded file.

 Advances and Applications in Statistics
© 2022 Pushpa Publishing House, Prayagraj, India
http://www.pphmj.com
http://dx.doi.org/10.17654/0972361722071
Volume 81, 2022, Pages 13-22 P-ISSN: 0972-3617

THE USE OF TWO-SAMPLE t-TEST IN THE REAL DATA

Abstract

The t-test is one of the most commonly used statistical methods. It was
developed and accredited by William Gosset, Karl Pearson and R.
Fisher in the 19th century. The test was further developed to the two-
sample test (Snedecor and Cochran [10]) which is used to determine
whether two populations are equal. A common application of the two-
sample t-test is to test whether a process or treatment is superior to a
current process or treatment. In this research, using the two-sample
t-test, a comparison between the students in the three grades of the
Department of Mathematics Education, Tishk International University,
is made to see whether there is a significant difference between the
grade point averages (GPAs) for the second, third and fourth grades,
Received: August 10, 2022; Accepted: September 15, 2022
2020 Mathematics Subject Classification: 62P99.
Keywords and phrases: grade points average (GPA), descriptive statistics, pooled variance,
two-sample t-test, significant difference.
How to cite this article: Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed, The use
of two-sample t-test in the real data, Advances and Applications in Statistics 81 (2022), 13-22.
http://dx.doi.org/10.17654/0972361722071
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Published Online: September 22, 2022
14 Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed
in addition to gender. The result showed that there is no significant
difference on the average scores between grade 4 and grade 3, also no
significant difference between grade 4 and grade 2, but there is a
significant difference between grade 3 and grade 2. Similarly, there is
no significant difference on the average scores due to gender.

1. Introduction

The two-sample t-test is a statistical test for comparing two groups’

means. In pain research, it is one of the most commonly utilized statistical
hypothesis tests (Yim et al. [12]). A test of significance is a formal technique
for comparing observable facts to a claim, also known as a hypothesis,
whose veracity is being determined. The claim is a declaration concerning a
parameter, such as the population percentage P or the population mean µ. A
significance test’s results are given as a probability that indicates how well
the data and the claim coincide. Xu et al. [11] mentioned that a significance
test begins with a thorough description of the statements being compared.
The null hypothesis is the assertion that is examined by a statistical test,
denoted by ( H 0 ) the purpose of the test is to determine how strong the
evidence is against the null hypothesis. The null hypothesis is frequently
stated as no difference. The alternative hypothesis denoted by ( H a ) is when
( H 0 ) is not true. The alternative is one-sided if a parameter is lower or
higher than the null hypothesis value. If it specifies that the parameter differs
from the null value, then it is two-sided (it could be either smaller or larger).
The significant level (α) is therefore found at both ends of the curve. Half of
(α) is on the upper end, while the other half is on the lower end. As a result,
there is a low and a high cutoff value. The p-value is calculated by
multiplying the area to the right of z by 2 in two-sided scenarios. The p-value
may only be doubled after that. “A t-test is an inferential statistic that is used
to see if there is a significant difference in the means of two groups that are
connected in some way” (Maüll Miquel [7]). The two-sample t-test can be
used to determine whether the means of two groups are equal. The t-test
is a parametric test: it is founded on an assumption that the underlying
population from which the samples are drawn is nearly normally distributed.
 The Use of Two-sample t-test in the Real Data 15
It is reasonably robust to failures in this assumption but should be treated
with caution as the true distribution deviates from normal. The test considers
means of two samples and tests the null hypothesis that the two samples are
drawn from populations with the same mean [5, 6]. Variants are provided
based on what is known about the underlying populations’ variation. The t-
test is a more cautious form the z-test, which relies on the Central Limit
Theorem’s confirmation that the sampling distribution of the mean is normal
for large samples. The t-test can be used with fewer samples to compensate
for the distribution distortion caused [5, 6].

2. Literature Review

In 1908, William Gosset, an Englishman publishing under the

pseudonym student, developed the t-test. The t-test is a parametric test: it is
founded on an assumption that the underlying population from which the
samples are drawn is nearly normally distributed. The test is considered to
be one of the most commonly used statistical methods. This method was
developed and is accredited to Gosset, Karl Pearson, in the 19th century
(Edwards and Fisher [2]). The method was further developed into the “two-
sample test” (Snedecor and Cochran [10]). The key new statistical challenge
for Gosset was that with such tiny samples, it was uncertain how well the
sample standard deviation, s, represented the population standard deviation
σ. This made it exceedingly difficult to detect if the two barleys differed
much. A common application of the two-sample t-test is to test whether
a new process or treatment is superior to a current process or treatment.
Because comparing the means of two samples is such a typical experimental
design, the student’s t-test for two samples is mathematically equivalent to a
one-way with two categories. McDonald [8] mentioned that “for the two-
sample t-test, we need two variables. One variable defines the two groups.
The second variable is the measurement of interest”. Peck et al. [9]
mentioned that two samples are deemed independent if the individuals or
objects that make up one sample have no bearing on the individuals or
subjects in the other sample. “The two-sample t-test should be used to
compare the mean values of two samples in this scenario”. He, also,
16 Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed
mentioned that “if the observations in the first sample are coupled with some
particular observations in the other sample, the samples are considered to
be paired”. The t-test is a useful statistical tool with moderate power for
determining if there is a significant difference between two groups. The
point at which t becomes significant is also influenced by sample size; the
bigger the n, the lower the t necessary to become significant. Excessive
repeating of the t test on the same dataset is a typical mistake in the usage of
the t-test (Foster and Gerald [3]).

3. Derivation of the Two-sample t-test

Let X 1, 1 , X 1, 2 , ..., X 1, n1 and X 2, 1 , X 2, 2 , ..., X 2, n 2 be two samples

from two independent normal distributions with means µ1 , µ 2 and common

variance σ2 . The unbiased estimators for the population means µ1 , µ 2 and

the variance σ 2 , respectively, are:

nj
xj =
1
nj  xi, j , (1)
i =1

nj

 ( xi, j − x j )2 ,
1
s 2j = j = 1, 2. (2)
nj −1
i =1

The maximum likelihood estimators can be written as (Heckert et al.

[4]):

(n1 − 1) s12 + (n2 − 1) s22

µˆ 1 = x1, µˆ 2 = x2 , σˆ 2 = . (3)
n1 + n2 − 2

This yields the test statistic

x1 − x2
t ( x1, x2 ) = (4)
 1 + 1  (σˆ 2 )
n 
 1 n2 
 The Use of Two-sample t-test in the Real Data 17
or, simply,

x1 − x2
t = (5)
σˆ 2 σˆ 2
+
n1 n2

which has a t-distribution with ( n1 + n2 − 2 ) degrees of freedom (Chang and

Pal [1]).

4. Application of the t-test

In this section, we present descriptive statistics for the data of

this research, which is the GPA score of the students for three grades
from Mathematics Education Department, Faculty of Education, Tishk
International University, the second grade has nine students, seven female
and two males, the third grade has sixteen students, ten of them females
and six of them are males, and the fourth grade has nineteen students,
ten females and nine males. A comparison between the students of the
department is done to see whether there is a significant difference between
the GPAs for the grades, and the gender. The two-sample t-test is used to see
this significance. The data is given as in Table 1:

Table 1. The GPA scores of the students for the three grades
Grade 2 3.53 2.98 2.77 2.59 2.44 2.13 2.05 1.95 1.30
Grade 3 3.92 3.70 3.62 3.60 3.03 2.98 2.94 2.17 2.84
2.72 2.70 2.66 2.60 2.53 2.44 2.32
Grade 4 3.76 3.55 3.44 3.09 3.01 2.89 2.87 2.83 2.80
2.78 2.76 2.67 2.58 2.31 2.30 2.04 1.81 1.23

The descriptive of the above data is given in Table 2:

Table 2. Descriptive of the scores of the students according to grades

Grades N Mean SE mean Minimum Maximum Median
Grade 2 9 2.416 0.217 1.300 3.530 2.440
Grade 3 16 2.923 0.131 2.170 3.920 2.780
Grade 4 18 2.724 0.138 1.230 3.760 2.800
18 Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed

Figure 1. The percentage of students according to grades.

Figure 2. The percentage of students according to grades and gender.

 The Use of Two-sample t-test in the Real Data 19

Figure 3. The percentage of students for all the grades according to gender.

Now, we will apply the two-sample t-test to see whether there is a

significant difference between the GPA scores of the student according to
their grades and according to their gender.

Two-sample t-test between grades 3 and 4

The null hypothesis is that there is no significant difference between the

two grades 3 and 4, i.e., H 0 : µ3 = µ 4 .

Regarding to equations (1), (2), (3) and (5), we have the following table:

Table 3. Two-sample t-test for grades 3 and 4

Grade N Mean Variance Pooled variance t-value p-value
4th 19 2.724 0.36 0.324 1.03 0.310
3rd 16 2.923 0.28

From the above table, we can see that there is no significant difference
between the GPAs of the two grades. Therefore, we accept the null
hypothesis.
20 Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed
Two-sample t-test between grades 2 and 3

The null hypothesis is that there is no significant difference between the

two grades 2 and 3, i.e., H 0 : µ 2 = µ3 .

Regarding to equations (1), (2), (3) and (5), we have the following table:

Table 4. Two-sample t-test for grades 2 and 3

Grade N Mean Variance Pooled variance t-value p-value
2nd 9 2.416 0.424 0.327 2.135 0.0427
3rd 16 2.923 0.276

From the above table, we can see that there is a significant difference
between the GPAs of the two grades. Therefore, we reject the null
hypothesis.

Two-sample t-test between grades 2 and 4

The null hypothesis is that there is no significant difference between the

two grades 2 and 4, i.e., H 0 : µ 2 = µ 4 .

Regarding to equations (1), (2), (3) and (5), we have the following table:

Table 5. Two-sample t-test for grades 2 and 4

Grade N Mean Variance Pooled variance t-value p-value
2nd 9 2.416 0.424 0.382 1.23 0.229
4th 19 2.724 0.36

From the above table, we can see that there is no significant difference
between the GPAs of the two grades. That is, we accept the null hypothesis.

Two samples t-test between females and males

The null hypothesis is that there is no significant difference between

females and males, i.e., H 0 : µ f = µ m .
 The Use of Two-sample t-test in the Real Data 21
Regarding to equations (1), (2), (3) and (5), we have the following table:

Table 6. Two samples t-test for the females and males

Grade N Mean Variance Pooled variance t-value p-value
Male 17 2.803 0.246 0.368 0.607 0.547
Female 27 2.689 0.443

From the above table, we can see that there is no significant difference
between the GPAs of the two genders. That is, we accept the null hypothesis.

5. Conclusions

According to the GPA scores, there is no significant difference between

grades 3 and 4, there is no significant difference between grades 2 and 4.
There is a significant difference between grades 2 and 3. According to the
GPA scores, there is no significant difference between males and females for
the three grades altogether. The maximum GPA score is for grade 3, and the
minimum GPA score is for grade 2.

References

[1] C.-H. Chang and N. Pal, A revisit to the Behrens-Fisher problem: comparison of
five test methods, Comm. Statist. Simulation Comput. 37(6) (2008), 1064-1085.
[2] A. W. F. Edwards and R. A. Fisher, Statistical methods for research workers,
Landmark Writings in Western Mathematics: Case Studies, I. Grattan-Guinness,
ed., Elsevier, Amsterdam, 2005, pp. 1640-1940.
doi: 10.1016/B978-044450871-3/50148-0.
[3] S. Foster and K. Gerald, Review of the two sample t tests, Nurse Anesthesia
1(1) (1990), 38-40.
[4] N. A. Heckert, J. J. Filliben, C. M. Croarkin, B. Hembree, W. F. Guthrie, P.
Tobias and J. Prinz, Handbook 151: NIST/SEMATECH e-Handbook of Statistical
Methods, 2002.
[5] F. Bacchus, AIPS 2000 planning competition: The Fifth International Conference
on Artificial Intelligence Planning and Scheduling Systems, AI Magazine
22(3) (2001), 47-56.
 22 Mowafaq Muhammed Al-Kassab and Aveen Hameed Majeed
[6] F. Bacchus and M. Ady, Planning with resources and concurrency: a forward
chaining approach, IJCAI, Vol. 1, 2001, pp. 417-424.
[7] A. Maüll Miquel, Functional stability of activated graphene-based electrodes after
sterilization with ethylene oxide, Bachelor’s thesis UPF, 2021.
[8] J. H. McDonald, Handbook of Biological Statistics, Vol. 2, Sparky House
Publishing, Baltimore, MD, 2009.
[9] R. Peck, C. Olsen and J. Devore, Introduction to Statistics and Data Analysis,
Cengage Learning, Boston, 2012.
[10] G. Snedecor and W. Cochran, Arc sine transformation for proportions, Statistical
Methods, 8th ed., Iowa State University Press, Ames, 1989, pp. 289-290.
[11] M. Xu, D. Fralick, J. Z. Zheng, B. Wang, X. M. Tu and C. Feng, The differences
and similarities between two-sample t-test and paired t-test, Shanghai Archives of
Psychiatry 29(3) (2017), 184-188.
[12] K. H. Yim, F. S. Nahm, K. A. Han and S. Y. Park, Analysis of statistical methods
and errors in the articles published in the Korean Journal of Pain, The Korean
Journal of Pain 23(1) (2010), 35-41.

View publication stats