Aim:-To test the effect of drug A in reducing anxiety by comparing two groups using t test.

Basic concepts

A t-test is a statistical method used to assess whether the difference in responses between two
groups is statistically significant. It encompasses any statistical hypothesis test where the test
statistic conforms to a Student's t-distribution under the null hypothesis. This test is employed to
ascertain if the observed variance in sample means indicates a genuine and meaningful difference
in the populations from which the samples were drawn, or if it arises due to random variation. By
relying on the t-distribution, the t-test calculates a test statistic (t-value) which is then compared
against critical values to determine statistical significance. The t-test assumes data normality or,
with larger sample sizes, a resilience to deviations from normality.

As stated by the American Psychological Association (APA), the t-test is a statistical method
utilized to assess hypotheses concerning the mean of a sample derived from a population,
assuming a normal distribution and unknown population variances. It is adaptable for both
independent and dependent groups, facilitating comparisons of means like pre- and post-training
test scores. Commonly referred to as Student's t-test, it was formulated by William Sealy Gosset,
an English scientist who published his findings under the alias "Student" in 1908. Gosset also
introduced the t distribution, comprising a range of curves shaped by degrees of freedom, which
represent the number of independent observations minus one. With increasing sample sizes, and
thus degrees of freedom, the t distribution gradually approximates the bell-shaped curve of the
standard normal distribution.

Characteristics of t-test-
Here are key characteristics of the t-test:
1. When samples are large, the values of standard error of the sample mean (sx̄) will be close to
that of population standard error (σx̄), and the distribution of t will be very nearly normal (with
an infinitely large sample, t = z). On the other hand, when sample size is small, the values of
standard error of the sample mean (sx̄) will vary considerably from population standard error
(σx̄). The distribution of t will then also depart considerably from that of the normally distributed
z.
2. Student’s distribution of t is not a single distribution, but rather a family of distributions like
standard deviation. They differ in their degree of approximation to the normal curve. The exact
shape of a particular member of the family depends on the sample size or, more accurately, on
the number of degrees of freedom (df), a quantity closely related to sample size.
3. When standard deviation is not known and the sample size is small (less than or equal to 30),
we use t distribution for the sampling distribution of mean.
4. Student’s t-distribution is symmetrical like a normal curve and is unimodal, meaning it has a
single mode.
5. The t-test shape or distribution depends on sample size. The top shape/peak of the t-test is
platykurtic.
6. Like normal distribution, the range of t-test or distribution is also -infinity to +infinity.
7. It also has “zero” as their centre point.
8. When the sample size changes, the degrees of freedom also change in a t-distribution.
9. At times, we can use t-test with a larger sample but the condition is that the population
variance is unknown.
10. It is a continuous probability distribution just like the z-score in which the estimated standard
deviation rather than the true standard deviation is considered.
11. The t-test estimates the true difference between two group means using the ratio of the
difference in group means over the pooled standard error of both groups.
12. A t-test can be calculated manually using a formula or using a statistical analysis software.

Hypothesis-

A hypothesis is a preliminary and testable statement or prediction formulated by researchers

based on their observations or inquiries into a phenomenon. It provides a foundation for
scientific inquiries, directing researchers in crafting experiments or studies to gather data.
Initially, a hypothesis is constructed or formulated, and then it can be either validated or
invalidated through testing. This testing can involve assessing a single population mean or
comparing means between two independent groups. Hypothesis testing involves making
statistical decisions about population parameters based on sample data.

There are two main types of hypotheses in statistical hypothesis testing:

1. Null Hypothesis (H0) - The null hypothesis represents a statement regarding a population
parameter that researchers assess through indirect and implicit testing. It posits no discernible
relationship between two variables. A null hypothesis is non-directional and symbolized by "H0".
It asserts the absence of an effect within the population, typically represented as "="
mathematically. For example, there exists no difference in job satisfaction between MNC and
Govt

H0 : μ = μ0
2. Alternative Hypothesis (H₁ or HA): An alternative hypothesis refers to a hypothesis concerning
a population parameter that opposes the null hypothesis, where testing is straightforward and
clear, indicating sample statistics. It represents a statement where there exists statistical
significance between two variables. An alternative hypothesis can be either directional or
non-directional and is symbolized by "HA". The alternative hypothesis discusses alterations in
opinions or actions when it is acknowledged. Mathematically, it is denoted as "≠, < & >". For
example, there exists difference in job satisfaction between MNC and Govt
 HA : μ ≠ μ0

In the process of hypothesis testing, when researchers reject the null hypothesis, it implies that
there is evidence to support the alternative hypothesis. The rejection of the null hypothesis is an
indication that the observed data provides sufficient evidence to conclude that there is a
significant effect, difference, or relationship in the population—supporting the claims made in
the alternative hypothesis. Also, failing to reject the null hypothesis does not prove the null
hypothesis true; it simply means there is not enough evidence to reject it based on the given data.
The process of hypothesis testing is designed to make informed and evidence-based decisions
about the characteristics of the underlying population being studied.

In statistical significance testing, both one-tailed and two-tailed tests serve as alternative methods
to assess the statistical significance of a parameter derived from a dataset, based on a test
statistic.
One-and two-tailed tests are used to determine the association between statistical variables. A
one-tailed test is used to determine the relationship between variables in a single direction (left or
right). A two-tailed test is used to determine whether the relationships between variables are in
any direction.
1. One-Tailed Test: A one-tailed test is based on a uni-directional hypothesis where the area of
rejection is on only one side of the sampling distribution. It determines whether a particular
population parameter is larger or smaller than the predefined parameter. It uses one single critical
value to test the data.
Left Tailed: In a left-tailed test, the alternative hypothesis suggests that the population parameter
being tested is significantly less than the hypothesized value. HA : μ < μ0

Right Tailed: In a right-tailed test, the alternative hypothesis suggests that the population
parameter being tested is significantly greater than the hypothesized value. HA : μ > μ0

2. Two-Tailed Test: A two-tailed test is also called a nondirectional hypothesis. For checking
whether the sample is greater or less than a range of values, we use the two-tailed. It is used for
null hypothesis testing. HA : μ ≠ μ0

One-Tailed Test Two-Tailed Test

A test of any statistical hypothesis, where the A test of a statistical hypothesis, where the alternative
alternative hypothesis is one-tailed either right-tailed hypothesis is two-tailed.
or left-tailed.
When the alternative hypothesis specifies a direction If no direction is given then we will use a two-tailed
then we use a one-tailed test. test.

Here, the Entire level of significance (α) i.e. 5% has It splits the level of significance (α) into half.
either in the left tail or right tail.

Rejection region is either from the left side or right Rejection region is from both sides i.e. left and right
side of the sampling distribution. of the sampling distribution.

It checks the relation between the variables in a single It checks the relation between the variables in any
direction. direction.

It is used to check whether one mean is different from It is used to check whether the two mean different
another mean or not. from one another or not.

For one-tailed, we use either > or < sign for the For two-tailed, we use ≠ sign for the alternative
alternative hypothesis. hypothesis.

Critical region lies entirely on either the right side or Critical region is given by the portion of the area lying
left side of the sampling distribution. in both the tails of the probability curve of the test
statistic.

Degrees of Freedom-

Degrees of freedom, often represented by df, the degrees of freedom in statistics indicate the
number of independent values that can vary in an analysis without breaking any constraints. It is
an essential idea that appears in many contexts throughout statistics including hypothesis tests,
probability distributions, and linear regression. It’s calculated as the sample size minus the
number of restrictions. t-tests are hypothesis tests for the mean and use the t-distribution to
determine statistical significance. A one-sample t test is used to examine whether the difference
between the sample mean and the null hypothesis value is statistically significant. We know that
when estimating the mean from a sample, there are n - 1 degrees of freedom, where n is the
sample size. As a result, for a 1-sample t-test, use n - 1 to determine degrees of freedom and
N - 2 is used to determine degrees of freedom for a two-sample t-test because there are now two
parameters to estimate.
Level of Significance-

The significance level, also known as alpha or α, is a measure of the strength of the evidence that
must be present in your sample before you will reject the null hypothesis and conclude that the
effect is statistically significant. The researcher determines the significance level before
conducting the experiment. The significance level represents the likelihood of rejecting the null
hypothesis when it is true. For example, a significance level of 0.05 represents a 5% chance of
determining that there is a difference when there is none. Lower significance levels suggest that
you need more evidence before rejecting the null hypothesis. Use significance levels during
hypothesis testing to help you decide which hypothesis the data supports.

These three numbers - 0.05, 0.01, and 0.001 - are the most commonly used levels of significance,
thus we usually choose one of the following:

1. Alpha = 0.05

2. Alpha = 0.01

3. Alpha = 0.001

Compare your p-value and significance level. If the p-value falls below your significance level,
you can reject the null hypothesis and conclude that the effect is statistically significant. In other
words, the evidence in your sample is robust enough to reject the null hypothesis at the
population level. A two sample t-test is used to test whether or not the means of two populations
are equal.

Assumptions of t-test-

The assumptions of a t-test depend on the specific type of t-test being conducted. However, there
are some common assumptions that apply to all types of t-tests:

1. The initial assumption pertinent to t-tests revolves around the scale of measurement. It
presupposes that the data collected adheres to a continuous or ordinal scale, such as IQ test
scores.
2. The second assumption involves a simple random sample, indicating that the data is derived
from a representative, randomly chosen subset of the overall population.
3. The third assumption entails that when plotted, the data exhibits a normal distribution
characterized by a bell-shaped curve. Assuming a normal distribution allows for the specification
of a probability level (alpha level, significance level, p) as a criterion for acceptance, often set at
5%.
4. The fourth assumption emphasizes the use of a reasonably large sample size. A larger sample
size implies that the distribution of outcomes should approximate a normal bell-shaped curve.
5. The fifth assumption pertains to the homogeneity of variance. Homogeneous variance, or
equal variance, is present when the standard deviations of samples are roughly equivalent.

Uses of t-test-

Uses of t-testThe uses of t-test in inferential statistics are as follows:

1. The calculations of a confidence interval for a sample mean.
2. To test whether a sample mean is different from a hypothesized value.
3. A t-test is an inferential statistic used to determine if there is a statistically significant
difference between the means of two variables.
4. The t-test is a test used for hypothesis testing in statistics.
5. Calculating a t-test requires three fundamental data values including the difference between
the mean values from each data set, the standard deviation of each group, and the number of data
values.
6. A t-test may be used to evaluate whether a single group differs from a known value (a
one-sample t-test), whether two groups differ from each other (an independent two-sample
t-test), or whether there is a significant difference in paired measurements (a paired, or dependent
samples t-test.

Additionally, t-tests find applications across diverse fields, serving in various capacities.

1. Psychology: t-tests are used to analyze the results of experiments comparing two groups,
such as before-and-after measurements or control and experimental groups. t-tests can be
applied to compare means between groups of individuals with different personality traits
or characteristics.
2. Education: t-tests are used to assess the effectiveness of teaching methods or
interventions by comparing the performance of students in different groups.
3. Quality Control: t-tests are employed to compare the means of product quality
measurements before and after process changes or between different production lines.
4. Social Sciences: t-tests can be used to compare means between groups in studies related
to social behavior, attitudes, or demographics.
5. Market Research: t-tests are used to compare means in studies assessing consumer
preferences or responses to marketing stimuli.
6. Biomedical Research: t-tests are used to compare the means of a treatment group and a
control group to assess the effectiveness of a new medical treatment. t-tests help compare
the means of groups receiving different drug dosages or placebo to determine the impact
of a drug.
 Each type of t-test has its own assumptions and requirements regarding the data and the
research design, and it is important to choose the appropriate test based on the characteristics of
the data and the research question at hand. Some of the widely used t-test types are as follows:

1. One-Sample t-test: While performing this test, the mean or average of one group is
compared against the set average, which is either the theoretical value or means of the
population. For example, a teacher wishes to figure out the average height of the students
of class 5 and compare the same against a set value of more than 45 kgs.The teacher first
randomly selects a group of students and records individual weights to achieve this. Next,
she finds out the mean weight for that group and checks if it meets the standard set value
of 45+.
2. Two-Sample t-test: This is the test conducted when samples from two different groups,
species, or populations are studied and compared. It is also known as an independent
t-test. For example, if a teacher wants to compare the height of male students and female
students in class 5, she would use the independent two-sample test.
3. Paired Sample t-test: This hypothesis testing is conducted when two groups belong to the
same population or group. The groups are studied either at two different times or under
two varied conditions.
4. Equal Variance t-test: This test is conducted when the sample size in each group or
population is the same or the variance of the two data sets is similar. It is also referred to
as pooled t-test.
5. Unequal Variance t-test: The unequal variance testing is used when the variance and the
number of samples in each group are different.

While the t-test is a valuable statistical tool for comparing means, it has some limitations that
researchers should be aware of. Here are some common limitations of the t-test:

1. Limited to Pairwise Comparisons: t-tests are suitable for comparing only two groups at a time.
When dealing with more than two groups, conducting multiple t-tests can increase the risk of
Type I errors.

2. Cumulative Type I Error: Without adjusting the significance level, conducting multiple t-tests
can inflate the overall Type I error rate, especially in the context of multiple comparisons.

3. Complexity with Multiple Groups: Handling and interpreting results from multiple t-tests can
become burdensome, particularly when dealing with numerous groups.

4. Assumption of Normality: t-tests assume that data follow a normal distribution; thus, results
may be invalid if this assumption is violated.
5. Sensitivity to Outliers: t-tests can be influenced by outliers, whereby extreme values
disproportionately affect the outcomes.

6. Equal Variance Assumption: The standard t-test assumes equal variances between groups.
However, if this assumption is violated, alternative tests like Welch's t-test may be more
appropriate.

7. Parametric Assumption: t-tests are parametric tests, relying on assumptions about the data
distribution. Non-parametric alternatives may be preferable if these assumptions are not met.

8. Sample Size Impact: Reliable results from t-tests often necessitate larger sample sizes,
particularly when data deviate from normality. Small sample sizes may not yield dependable
outcomes.

Case Study

A researcher wants to test whether Drug ‘A’ reduces anxiety. For this purpose, he used two
groups of individuals- experimental & control. He made use of an anxiety scale for the
assessment of anxiety among the subjects of the group. The experiment group was given the
Drug ‘A’ while no such thing was given to the control group. The results were pooled.

Experimental Group (X) Control Group (Y)

77 87
88 77
77 71
90 70
68 63
74 50
62 58
93 63
82 76
79 65

Test the null hypothesis at 0.05 level of significance.

At the start of the research study, we formulated both the null hypothesis and the research
(alternative) hypothesis.

Population 1- People who received the drug

Population 2 - People who didn't received the drug

Null hypothesis- There is no difference in anxiety between 2 groups

H0 : ux = uy

Alternative hypothesis- There will be a difference in anxiety between two groups

HA; ux ≠ uy

Solution-

Experimental Group Control Group X2 Y2

(X) (Y)

77 87 5929 7569
88 77 7744 5929
77 71 5929 5041
90 70 8100 4900
68 63 4624 3969
74 50 5476 2500
62 58 3844 3364
93 63 8649 3969
82 76 6724 5776
79 65 6241 4225

ΣX=790 ΣY=680 ΣX2 =63260 ΣY2 =47242

n= 10

α= 0.05
 X = ΣX
一
n

790
= 一一 = 79
10
—
Y = ΣY
一
n

680
= 一一 = 68
10

SSx = ΣX2-(ΣX)2 SSy= Σ Y – (Σ Y)2

——
n
一一
=63260 – (790)2 n
———
= 47242 – (680)2
10 一一
10
= 63260– 62410 =47242–46240
= 850 =1002

𝑆𝑆𝑥
𝑆𝑥 = 𝑛−1

(𝑆𝑥 is standard Deviation of experimental group)

850
𝑆𝑥 = 10−1

850
𝑆𝑥 = 9

𝑆𝑥 = 94. 44
𝑆𝑥 = 9. 71

𝑆𝑆𝑦
𝑆𝑦 = 𝑛−1

(𝑆𝑦 is standard Deviation of control group)

 1002
𝑆𝑦 = 10−1

1002
𝑆𝑦 = 9

𝑆𝑦 = 111. 33
𝑆𝑦 = 10. 54

(𝑋 − 𝑌) −(𝝁𝑥−𝝁𝑦)ℎ𝑦𝑝𝑜𝑡ℎ𝑒𝑠𝑖𝑠
t=
𝑆𝑆𝑥+𝑆𝑆𝑦

𝑛(𝑛−1)

(79−68)−(0)
= 850+1002
10(10−1)

11
= 1852
10(9)

11
= 1852
90

11
=
20.58

11
= 4.53

= 土2.42
Level of significance (α) = 0.05

0.025 0.025

df=(nx-1) + (ny-1)
=(10-1)+(10-1)
= 9+9
= 18

t-critical value = 土2.101

t obt. = +2.42
t crit. = 土2.101
Df = 18
Hypothesis is in 2 directions. Null hypothesis is rejected and alternative hypothesis is accepted.

Effect size
𝑋−𝑌
Cohens’ d = 𝑆

(X= Mean of experimental group)

(Y= Mean of control group)

79−68
d= 10.54
= 1. 04

So as effect size is large

Therefore, a large effect size indicates greater practical significance in this study, suggesting that
the drug aids in reducing anxiety