Hypothesis means testable assumption, a statement of particular outcomes or a predicted answer to a research problem or a hypothetical/academic statement of the relationship between two or more variable being examined or a assumption from a claim. It can be described as the predicted answer to the research objectives and questions. It is a notion or guess concerning what the student researcher believes the results of a research will be like. Basically, it is a unconfirmed assumption, speculation, guess, possibility, explanation, deduction or prediction about some observable facts that can be tested in a realistic time limit. It is always stated in declarative manner to connect/link the study variables to each other in general or specifically. Even as the hypothesis predicts what the researchers anticipate to see, the aim of research work is to find out whether this prediction is right or wrong.
Types of Hypothesis
There are two types of hypotheses. They are Null and Alternative. Null hypothesis is declared and stated in a negative manner e.g. there is no significant relationship between Corporate Social Responsibility and Financial Performance of Firms. It is usually represented with Ho. The Alternative hypothesis is declared and stated in a positive or affirmative manner e.g. there is significant relationship between Corporate Social Responsibility and Financial Performance of Firms. It is usually represented with Hi. It should be noted at this junction that it is not in all research topics that require hypotheses formation; research questions only is enough for some studies.
Sources of Hypotheses
Hypotheses can be formulated from the:
- Research variables, objectives
- Research questions
- Results of other studies
- Through personal or practical innovative effort
- Logical way of thinking and inquisitiveness
- Problems that needs a solution
- Opinions from students or colleagues or professionals in the subject
It is very important that your hypotheses are testable. Yes!
A testable hypothesis declares a statement regarding a Supposed or theoretical relationship between two or more measurable and observable variables. Therefore, a testable hypothesis is supposed to state or mean that the variables can be measured or observed. It should also indicate the relationship among the variables, e.g. improved firm’s financial performances are recorded after spending largely on Corporate Social responsibility.
Example of non-testable hypotheses include, “getting educated is good“, “children are the leaders of tomorrow”.
Characteristics of Good Hypotheses
A good research hypothesis should have the characteristics listed below:
- It should offer a particular and sufficient answer to a research problem that has limited scope
- It should be related to the research topic.
- It should include independent and dependent variables. It must be structured in a manner that the anticipated relationship between independent and dependent variables are clearly stated.
- Hypotheses must be observable and measurable. Selection of hypotheses that is not testable will lead to difficulties during the course of the project work
- It must be properly, clearly and accurately stated.
- It must state predicted relationships between the research variables. Hypotheses should be suitable as a foundation for the research.
- It should have limited scope. Student researcher must not necessarily formulate hypotheses that have global significance. They should rather propose hypotheses that are fairly simple to test, and nevertheless are greatly significant.
- It must be reliable with most facts. Any hypothesis formulated for research should be in line with the substantial association of established facts.
- The formulated hypotheses should be adaptable to testing within a realistic time frame.
- Hypotheses must be formulated with consideration to ethical standards.
Step by step guide to an awesome hypotheses testing
- Formulate your Hypothesis: The first step is to state a Null Hypotheses (Ho) which can either be accepted or rejected.
- Select an appropriate level of significance. This indicates the level at which the null hypothesis is accepted or rejected. It is traditional among researchers across the world to accept or reject hypothesis at 0.05 or 0.01 level of significance. This indicates that the researcher is allowed some error margin in the research findings. For example 0.05 level of significance means that the researcher allows 5% error limit and he is 95% confident of whatsoever finding or conclusion is drawn from the research while for 0.01 level of significance the researcher allows only 1% error margin and he is 99% confident of the result or conclusion drawn from his study. 0.05 level of confidence is commonly used by undergraduates and postgraduates in education, social sciences and management sciences. 0.01 level of significance is commonly used in high tech research mostly in health related research or drug testing experimental studies
- Choose suitable statistical techniques. There are many methods from which a researcher can select the most suitable to test the formulated hypothesis.
The following things are to be noted when choosing an appropriate statistical technique:
When your research involves larger sample (more than 30) and you are testing hypothesis for a significant difference between two variables, the z – test indicating normal distribution is utilized and for smaller studies (less than 30), t – test is used.
When the hypothesis is formulated to test for relationship between two variables, the Pearson product moment correlation co-efficient is employed. Studies where there is equal distribution of non-parametric variables, the Spearman rank order correlation coefficient is employed.
When the hypothesis is formulated to test the differences or relationship between one dependent variable and two or more independent variables, the analysis of variance (ANOVA) or analysis of covariance (ANCOVA) can be used.
Multivariate Analysis of Covariance (MANCOVA) is used when there are two or more independent variables against two or more dependent variables.
Chi-square test is applied to data derived from normal scale of measurement, that is, data obtained in form of frequency counts. It is used for testing hypothesis concerning the difference between sample frequency observed within certain categories and those expected with the categories and those expected with the categories. Chi-square can only indicate whether or not a set of observed frequencies differ significantly from the corresponding set of expected frequencies.