Validity and reliability
Introduction to validity and reliability
In research the concern of an investigator is how to minimize possible errors and bias by maximizing the reliability and validity of data. This then requires that the tool for the collection of data is valid and reliable. This section explains the technical meaning of these two concepts. The reliability of the findings, conclusions and recommendations arrived at in any study largely depends on the validity, and reliability of the methods and instruments used in data collection.
Definition of validity
This refers to the extent to which a measurement does what it supposed to do. Data need not only to be reliable but also true and accurate. If a measurement is valid, it is also reliable. But if is reliable, it may or may not be valid. Validity of an instrument is easy to determine if one is dealing with information that can be quantified. For example in growth monitoring, the height and weight of a child is easy to determine. However, when handling qualitative information such as feelings, likes, dislikes, opinions etc, validity is more difficult to determine.
Defination of reliability
Reliability refers to the consistence, stability, or dependability of the data. Whenever an investigator measures a variable, he or she wants to be sure that the measurement provides dependable and consistent results. A reliable measurement is one that if repeated a second time will give the same results as it did the first time. If the results are different, then the measurement is unreliable. It is easier to determine reliability when dealing with information that can be quantified. For example it is easier to determine the reliability of an instrument used to measure the performance of mathematics in a form 2 class, than an instrument used get their music ability. Reliability of an instrument is increased by identifying the precise data needed and repeated use of the instrument in field testing.
In surveys, reliability problems commonly result when the respondents do not understand the question, are asked about something they do not clearly recall, or are asked about something of little relevance to them. Data obtained for instance from educational statistics or records can be unreliable if educational administrators fail to record information or make frequent errors in entering the data.
Types of validity
- Content validity
It deals with representativeness of the adequacy of the content of an instrument. It seeks to answer whether the content of the measure is representative. The instrument must show that it comprehensively covers the items it claims to cover. This can be done by ensuring elements of the wider issue under investigation and the items used have taken care of depth and breadth.
- Predictive validity
This involves predicting by means of assessment or technique performance on some other criterion. An assessment for example can be used as a predictor when it is used to place children in groups. Predictive validity therefore is achieved if the data acquired at the first round of research correlates highly with that acquired later.
- Concurrent validity
This is when data gathered from one instrument must correlate highly with data gathered from using another instrument.