Essay: What are the hypothesis tests and their role?

Essay details:

  • Subject area(s): Science essays
  • Reading time: 8 minutes
  • Price: Free download
  • Published: June 27, 2021*
  • File format: Text
  • Words: 248 (approx)
  • Number of pages: 1 (approx)
  • What are the hypothesis tests and their role?
    0.0 rating based on 12,345 ratings
    Overall rating: 0 out of 5 based on 0 reviews.

Text preview of this essay:

This page of the essay has 248 words. Download the full version above.

What are the hypothesis and its role?

One main step for research are the hypothesis tests. Hypotheses are statements that assign variables to cases. A hypothesis performs a number of essential functions. The most important is that it accompanies the guidance of the study. A common problem that appears within the research is the accumulation of interesting information. If the researcher does not manage the strong desire to include extra elements, a study can be weakened by not that important concerns that do not have an answer for the predominant question that is posed. The advantage of the hypothesis is that, if the researchers take it seriously, it reduces what shall be studied and what shall no longer. It differentiates the related and not related facts and in addition, it proposes which is the most right and applicable research method. The ultimate role of the hypothesis is to support a framework as a way to organize the conclusion of the research.

The logic behind the hypothesis testing

In classical tests of significance, there are two kinds of hypothesis used. The first of them is the null hypothesis which says that there is no difference between the parameter and the statistic being in comparison to it. The second one is the alternative hypothesis, which is the opposite of the null hypothesis. It may appear in a number of varieties which rely on the objective of the researcher. The types may be “not the same” (≠), “higher than” (>), or “less than” (<).

If we reject the null hypothesis (finding the statistically important difference), we accept the alternative hypothesis.

Tests of significance: types of tests.

After the assessment of the main types and their assumptions, we will have to choose an appropriate test. There are the two basic classes of significance tests: parametric and nonparametric tests.

Parametric tests

The parametric tests are more effective than the nonparametric tests given that the information that they use are borrowed from interval and ratio measurements. The parametric tests have a few assumptions:

  • The observation ought to be independent
  • The observations have to be taken from normally distributed populations
  • The populations should have equal variance
  • The measurement scales must be an interval in order that the arithmetic operations can be used with them.

Parametric tests place different importance on the assumptions. Some tests are fairly powerful and remain successful despite the infringements. Other tests, consider that if they depart the linearity or equality of variance it is going to reason a danger to the validity of the results.

Nonparametric tests

Nonparametric tests are used to test the hypotheses with nominal and ordinal data. These tests have less strict assumptions. They do not identify normally distributed populations or the sameness of variance. Some tests demand the independence of cases. Other tests are clearly made for situations with related cases. Nonparametric tests are the only ones which can be used with nominal data. In addition, they are the only ones that can technically be used correctly with ordinal data, despite the fact that parametric tests are usually utilized in this case.
Nonparametric tests can also be in use for interval and ratio data, despite the fact that they trail some of the existing information. They are easy for understanding and using. It is true that parametric tests are extra effective when it is proper to use them, but even in such cases, nonparametric tests achieve 95 percent of efficiency. Because of this, if the parametric test has a sample of 95, it will have the same statistical testing power as a nonparametric test with a sample of 100.

To sum up, parametric and nonparametric exams are applicable in one-of-a-kind stipulations and circumstances. Parametric tests use interval and ratio data and they are ideally used when their assumptions can be met while nonparametric tests do not need strict assumptions about population distributions and they are in use with weaker nominal and ordinal measures.
The most frequently used tests and the importance of the nonparametric tests in their usage

One-Sample tests

We use one-sample tests when we have a single sample and we wish to test the hypothesis that it derives from a defined population. A number of nonparametric tests can be put-upon in a one-sample situation, but we must not forget the measurement scale and some other stipulations. When the measurement scale is nominal, or in other words classificatory only, we can use each binomial test and the chi-square (ꭓ2) one-sample test. The binomial test can be used if the population is viewed as only two classes (for example, female and male). The researchers opt for to use binomial tests when the size of the sample is small and the ꭓ2 test is not in a use.

Chi-Square Test

This is probably the most used nonparametric tests of significance. It is vitally valuable in tests which include nominal data, however, it can be also applied in situations with higher scales. The typical cases that it takes part of being the cases where the objects are separated in two or more nominal categories (for instance, “yes – no”, or “A, B, C, or D”).

The use of this technique makes it possible for us to test for important differences between the observed distributions of data for each of the categories and the expected distribution keeping with the null hypothesis. It must be calculated not in percentages, but with actual counts.
Chi-square tests are priceless in one-sample analysis, two independent samples, or k independent samples. Within the one-sample analysis, we use a null hypothesis established on the expected frequency of objects in the different categories. The components during which the ꭓ2 is calculated is:

In Which:

O = Observed frequency
E = Expected frequency
∑ = Summation
X2 = Chi Square value

There is a different distribution for each number of degrees of freedom:

df = k – 1, where k is the number of categories in the classification. The procedure is:

Defining the null hypothesis and the alternative hypothesis: Ho:Oi=Ei; H1: Oi≠Ei

Statistical test: use one-sample X2

Significance level

Calculated value: with the formula of X2 and outline df.

Critical test value

Decision

Two Independent Samples Tests

In business researches it is normally required to use two independent sample tests. Here the chi-square (ꭓ2) test is very useful especially in situations which require a test of differences between samples. It can be used with both nominal and ordinal measurements. When there is a reduction of a parametric data into categories they are most likely handled by the chi – square test although it loses information.
We need the chi-square test to work properly. So, we ought to take the data from random samples of multinomial distributions, and the expected frequencies should not be too small.

The formula is:

X2=∑∑ [(Oij – Eij)2/Eij],

In which:

Oij – observed number of cases categorized in the ith cell

Eij – expected number of cases under H0 to be categorized in the ith cell

The testing approach is:

Defining the null hypothesis and the alternative one

Statistical test X2

Significance level and df

Calculated value

Critical test value

Decision

Two Related Samples Tests

They are used in situations where people, objects or events, for example, are nearly corresponding. Here, we can apply each of the parametric and the nonparametric tests.

Nonparametric Tests: The McNemar test is appropriate to be used with nominal and ordinal data and it is also beneficial with before-after computation of these subjects.

The formula is:

X2 = [(|A – D| – 1)2 / A + D] with df = 1

The testing procedure is:

Defining the null and alternative hypothesis

Statistical test

Significance level

Calculated value

Critical test value

Decision

K Independent Samples Tests

Many of the researchers decide to use k independent sample tests in management and economic researches when there are over three samples that are concerned. They’re also quite interested in seeking out the place the samples might come from – if they come from the same or identical populations. The analysis of variance and the F test are utilized in situations where the data are measured on an interval-ratio scale and the researchers can meet the wanted assumptions. A nonparametric test must be chosen when introductory analysis shows that the assumptions cannot be met or if the data have been measured on a nominal or ordinal scale.

The samples are approved to be independent. We take this as a condition of a totally randomized experiment when the subjects are set to specific treatment groups. The requirement of the evaluation of more than two independent sample means is also common.
The chi-square tests are very proper when there are k independent samples for which the nominal data have been collected. It can be utilized to classify the data at higher measurement levels, but after we make it smaller, the metric information is lost. The k-sample chi-square test is calculated within the same method as the two independent sample cases.

When we have an ordinal scale collected data or interval data that does not meet the F-test assumptions or for any reason is not suitable for a parametric test the researchers use the Kruskal-Wallis test. It is a one-way analysis of variance with the aid of ranks. It accepts an underlying continuous distribution as well as random selection and independence of samples.

With this test we rank all observations from the smallest to the largest.

K Related Samples Case

This test is used in situations where there are more than two levels of the grouping component, observations are matched or they are measured more than once, the data are at least interval. In experimental designs, it\’s a just right notion to measure the subjects several times. These multiple measurements are named trails.

The Cochran Q test (also a nonparametric test) is preferably used when the k related samples are measured on a nominal scale. It checks the hypothesis that the proportion of cases in one category is equal for various associated categories.

The Friedman two-way analysis of variance can also be preferable to use when the data are ordinal. It examines the matched samples, ranking every one of the cases and calculating the mean rank for each variable for all cases.

Other nonparametric significance tests

One-Sample Case

Kolmogorov-Smirnov Test

The Kolmogorov – Smirnov test is suitable to use when the data are ordinal and the research needs to compare the observed sample distribution with the theoretical distribution. In this case, the Kolmogorov-Smirnov test (KS) is more applicable than the chi-square test. It can be used for small samples when other tests like Chi-square test, for instance, cannot be used. The theoretical distribution shows the researcher’s expectations under the null hypothesis H0. We determine the D (maximum deviation) which is the point of the great divergence, or in other words, the greatest variation, between the observed and theoretical distribution. It is calculated as follows:

D=maximum |F0 (X) – FT (X) |

In which:

F0 (X) – The observed cumulative frequency distribution of a random sample of n observations. X is any possible score, F0 (X) = k/n, where k=the number of observations which are equal to or less than X

FT (X) – The theoretical frequency distribution under H0.

The process of testing is:

Defining the null and alternative hypothesis

Statistical test

Significance level

Calculated value

Critical test value

Decision

Two-Sample Case

Sign Test

It is used when the only information is the identification of the pair member that is greater or smaller or has probably most of the characteristics. This test is based on the binomial widening and it is good to use with small samples.

Wilcoxon Matched-Pairs Test

This test is mainly used when the researchers can determine the direction and magnitude of differentiation between pairs that are carefully matched. The way of calculation is also simple. We have to find the difference score (DI) between the pairs, and after that rank the differences from smallest to largest without attention to signs. Then the actual sign is added to the rank values, and the T test statistic is calculated (T is the sum of ranks with less common sign).

The formula for this particular test is:

“z= T-μ/σ”

In which:

µ – mean

“σ” – Standard deviation

The testing process is:

Defining the null and alternative hypothesis

Statistical test

Significance level

Calculated value

Critical test value

Decision

Kolmogorov-Smirnov Two-Sample Test

It is used by the researcher when he or she has two independent samples of ordinal data. This test is plagued by the agreement between two cumulative distributions, and each of them represents sample values.

We reject the Ho when the cumulative distribution shows sufficiently large maximum deviation D. To protect the maximum deviation, we should use as many intervals as possible for the reason that we do not need to show the maximum cumulative difference.

The formula for Two-Sample KS test is:

D= maximum |FN1 (x) – FN2 (X) | (two-tailed test)

D= maximum |FN1 (x) – FN2 (X) | (one-tailed test)

The testing process is:

Defining the null and alternative hypothesis

Statistical test

Significance level

Calculated value

Critical test value

Decision

Mann-Whitney Test

This test is used with two independent samples and if the data are ordinal. When the greater of the two samples is twenty or smaller, it is interpreted by special tables U. However, when the larger sample is bigger than twenty, we use a normal curve approximation.
We are able to compute U statistics in two exclusive methods. There are the formulas:

U=n1n2 + [n1 (n1+1) /2] -R1 or U=n1n2 + [n2 (n2+1) /2] -R2

In which:

n1 – number in sample 1

n2 – number in sample 2

R1 – sum of ranks in sample 1

For testing purposes we have to use smaller U.

The testing process is:

Defining the null and alternative hypothesis

Statistical test

Significance level

Calculated value

Critical test value

Decision

Other nonparametric tests

Fisher exact probability test

When the measurement is only nominal

The median and Wald-Wolfowitz runs test

When the data are at least ordinal

About Essay Sauce

Essay Sauce is the free student essay website for college and university students. We've got thousands of real essay examples for you to use as inspiration for your own work, all free to access and download.

...(download the rest of the essay above)

About this essay:

If you use part of this page in your own work, you need to provide a citation, as follows:

Essay Sauce, What are the hypothesis tests and their role?. Available from:<https://www.essaysauce.com/science-essays/what-are-the-hypothesis-tests-and-their-role/> [Accessed 28-01-22].

These Science essays have been submitted to us by students in order to help you with your studies.

* This essay may have been previously published on Essay.uk.com at an earlier date.

Review this essay:

Please note that the above text is only a preview of this essay.

Name
Email
Rating
Review Content

Latest reviews: