| Introduction: | | | | the validity of a study does not depend on the |
| The conclusions and propositions of any social | | | | use of good study design or even use of good |
| science research must be reliable and valid, | | | | samples in the study. Validity measure is divided |
| reliability and validity are related but are different | | | | into four including conclusion validity, internal |
| in their perception, this paper focuses on the | | | | validity, construction validity and finally external |
| definitions and meaning of validity and reliability and | | | | validity. |
| also shows the relationships that exists between | | | | Conclusion validity: |
| both concepts. In any research we can have valid | | | | Conclusion validity involves comparing the results |
| and reliable results, reliable but invalid results, valid | | | | of the study and the conclusion, for example if in |
| and non reliable results and finally invalid and | | | | a study we want to establish the relationship that |
| unreliable results. | | | | exits between variable A and B and that A |
| Reliability: | | | | causes B then if we conclude that A causes B |
| Reliability means that the results of a study are | | | | then the results are valid. |
| of quality in measuring relationships between | | | | Internal validity: |
| variables, when measures are reliable it means | | | | In this form of validity we establish whether the |
| that if another study was undertaken the same | | | | relationship between variables is causal, example if |
| results will be obtained, this means that the | | | | we establish that A and B are related in a study |
| measure of the study is consistent and repeatable | | | | we test whether A causes B and this is achieved |
| if other studies of the same type are undertaken. | | | | through tests. |
| There are various reliability measures which | | | | Construction validity: |
| include observer reliability, test reliability, parallel | | | | This validation method tend to establish whether |
| reliability and consistent reliability: | | | | the relationship being established is credible, this is |
| Observer reliability: | | | | done through determining whether the measures |
| This involves the efforts to establish where there | | | | that were intended to be measured were |
| observations are consistent with the others, | | | | measured, this can only be achieved if the study |
| example a study involving the interview of two | | | | construction was well prepared for the collection |
| individuals then there is a need to compare the | | | | of observations. |
| two observations and establish whether the | | | | External validity: |
| results from both observations are consistent and | | | | This tends to establish that a study is valid if |
| therefore reliable. | | | | same results can be achieved using another |
| Test reliability: | | | | sample or if the study was undertaken at |
| In test reliability we try to establish the correlation | | | | different area or different time period, this is |
| between studies undertaken in two time | | | | what is referred to as external validity. Therefore |
| periods, it is evident from research that given the | | | | All the above validity methods will help establish |
| same sample and undertake the study tafter a | | | | whether a research proposition or inference is |
| short time period then the higher the correlation | | | | valid or not valid.validity and reliability:in social |
| but if we took the same sample and the have a | | | | research we can have both valid and reliable |
| longer time period then the lower the correlation. | | | | results, we can also have reliable but invalid |
| Parallel reliability: | | | | results, we can also have valid and non reliable |
| This involves testing reliability using many set of | | | | results and finally we can have invalid and |
| questions that tend to yield same results, in a | | | | unreliable results. |
| study questions can be split into two where the | | | | |
| second part tends to establish the same results | | | | D |
| as the first part, this way the similarities will help | | | | |
| establish whether the results are consistent and | | | | C |
| reliable. | | | | |
| Consistency reliability: | | | | B |
| This involves establishing a single measure of | | | | |
| consistence on a group of participants at the | | | | Awe can discuss the diagram as follows: |
| same time and try to establish whether the | | | | Diagram A shows reliability but results are not |
| results are consistent, this helps in determining | | | | valid, in such a research study we aim at |
| whether a study undertaken on the group will be | | | | measuring a certain variable but we end up having |
| consistent. | | | | results that are reliable but not valid. |
| All the above methods are used in establishing | | | | Diagram B shows results that are valid but not |
| whether the results of a study are reliable or not, | | | | reliable, the results are valid because they are |
| however the methods have their strength and | | | | distributed but they are unreliable because results |
| weakness but they are important in determining | | | | are not consistent. |
| reliability. | | | | Diagram C shows a situation in which we do not |
| Validity: | | | | have valid and still the results are unreliable, finally |
| The validity can be defined as the best | | | | in the last diagram D shows reliable and invalid |
| approximation of truth of an inference, however | | | | results. |