We live in a day and age where information can be mathematically determined with the aid of statistics. However, the study of statistics, as it seems, isn’t simply that of facts and numbers.

Statistical inference consists of the use of statistics to create decisions concerning the parameters of a population, based on random sampling. The implementation of statistical inference involves hypothesis testing and talks about how this procedure is employed by statisticians to just accept or reject the assumption of a population parameter. Under this methodology, the topic of T-Tests and its various types, that is, one sample T-Test, independent T-Test and paired T-Test, comes about.

**Paired T-Test vs Unpaired T-Test**

The difference between the two statistical terms Paired T-test and Unpaired T-test is that in Paired T-Tests, you compare the differences between the paired measurements that have been deliberately matched whereas, in Unpaired T-Tests, you measure the difference between the means of two samples that do not have a natural pairing.

## Comparison Table Between Paired T-Test and Unpaired T-Test (in Tabular Form)

Parameter of comparison | Paired T-Test | Unpaired T-Test |
---|---|---|

Meaning | Paired T-Test, also known as repeated samples T-Test, determines the distinction between the two means of the same subject. | Unpaired T-Tests, also known as independent T-Tests or student’s T-Test, is determining the two means groups of different/unrelated subjects. |

Homogeneity of variances | Under Paired T-Test the variance of the two mean groups are not equal. | Under Unpaired T-Test the variance of the two mean groups are equal. |

Effects/impacts | Paired T-tests deal with very minor errors since the test is done only between two similar groups. | Unpaired T-Tests have slightly more errors in comparison with paired T-Tests since the experimenter would be affected by variations between two different subjects. |

Outcome | Paired T-Tests need not collect massive amounts of sample data for comparison this successively saves money and time. | Since Unpaired T-Tests have to compare the means of two independent subjects this lands up being a slightly more costlier and time-consuming process. |

## What is Paired T-Test?

A Paired T-Test, additionally referred to as correlated pair t-test/paired sample t-test/dependent t-test, is a statistical procedure that runs a test on dependent variables. A paired test is done on similar subjects before the allocation of data and two tests are done before and after a treatment. For instance, improvement of scholars seen in an English class test conducted at the beginning of the year and the end of the year, the before and after impact of a medication on the same group of individuals, etc.^{}

**Hypothesis:**

The two hypotheses under paired t-test.

- The null hypothesis (H
_{0}): no significant difference between specified populations, H_{0: }μ_{1 = }μ_{2} - The alternative hypothesis (H
_{1}): there is a statistically significant difference between the two population means caused by rejecting the null hypothesis. H_{1}: μ_{ 1}≠ μ_{2}

**Assumptions:**

The paired sample t-test makes the subsequent assumptions:

- The differences between the similar pairs follow a normal probability distribution.
- The observations ought to be sampled independently and identically distributed.
- A Paired t-test is measured on a gradual level with the help of ratios or intervals. Since T-Tests are based on a normal distribution, the data needs to be continuous and not discrete
- The independent variables should comprise of two dependent/similar groups.

## What is Unpaired T-Test?

An unpaired t-test, also known as an independent sample t-test/two-sample t-test, is a statistical method that determines whether or not there is a significant distinction between the means of two unrelated independent groups. For example: when you want to compare the average sleep cycle of individuals grouped by gender: male and female groups.

**Hypothesis for the independent t-test:**

The null hypothesis for the independent t-test is that the population means from the two different groups are equal:

H_{0: } μ_{1}= μ_{2}

Alternative hypothesis is accepted once the null hypothesis is rejected, which means that the population means are not equal

H_{1: } μ_{1}** **≠ μ_{2}

To reject or accept the null hypothesis, a significance level is critical. This particular value is 0.05.

**Assumptions:**

- The first assumption concerns the scale of measurement- the data collected ought to follow a continuous or ordinal scale.
- The data should be collected from a randomly selected portion of the total population.
- The data should result in a normal, bell-shaped distribution curve. The significance level can be specified when a normal distribution is assumed.
- A massive sample size ought to be used.
- The variance and standard deviations should be equal for the dependent variables.

**Main Differences Between Paired T-Test and Unpaired T-Test**

*Paired T-Tests*means comparing the difference between the two mean groups of dependent subjects. For example: the IQ of 5 students before and after training.*Unpaired T-Tests*refers to calculating the differences between the two mean groups of independent subjects. For example: Mean of 100 students Group 1- 50 girls and group 2- 50 boys- The variance of
*Paired T-Tests*is said to be equal. Since the variance is equal standard deviation also is equal for the two mean groups. The variance of*Unpaired T-Tests*is assumed to be unequal and because of this, the standard deviation is also unequal. *Paired T-Tests*has less random errors since Paired T-Tests mainly deal with finding the variations between two mean groups of similar subjects the experimenter need not focus on the individual differences. Under*Unpaired T-Tests*the experimenter’s observation will be affected by the individual differences of the two groups of unrelated subjects and this would in turn end in a random error.*Paired T-Tests*saves heaps of time and money for the experimenter as he need not find large amounts of sample data to calculate the two similar mean groups.*Unpaired T-Tests*are slightly costlier and time-consuming process as the experimenter would have to find a lot of data to analyze the two independent mean groups.

## Conclusion

Every day, individuals find themselves analyzing new ideas, attain quick methods to complete the allocated work, or find a straightforward, not so sophisticated approach, to try to do what they are doing best. The essential question is whether or not the new idea is notably better than what they originally had in mind. These new ideas individuals tend to come up with is commonly called a hypothesis. Testing these ideas to resolve whether or not one would work better than the other is called hypothesis testing. It is the art of making decisions using data.

The above piecework gives an overview of the two statistical terms- Paired T-Tests and Unpaired T-Tests. It helps us go into a detailed study about the concept of Unpaired T-Tests and pops the question of how advantageous it is in matters of deciding the probability of a value in a sample and whether or not the benefits outgrow the drawbacks when it comes to choosing this technique of calculation.

It also provides us a run-through of the concept of Paired T-Tests and shows us the various fields and examples where Paired T-Tests are suitably used, assumptions that are necessary to follow before-hand and the formula that could be used for the calculation to ensure the significance of the distinction between means of measures taken twice from the same subject.