Validity and Reliability
How Valid and Reliable is your Questionnaire? This is one question every Research Student must answer one day in front of their supervisors and perhaps in front of the defense jury. Asking the right questions, applying the right research/measurement instruments to achieve research objectives and meeting the right respondents are very crucial when conducting a survey.
A Reliability and Validity check is something you must perform to achieve an accurate and unbiased result from a study.
Now the next question is How do they go about? I guess you really want to find out. The next few paragraphs will answer that question.
But before I go into that, you need to understand some basics about Validity and Reliability of a questionnaire.
It is the degree at which a questionnaire measures what it is meant to measure. The validity check revolves around these three outlines:
- Structure of the questionnaire
- Goal of the questionnaire and
- The target population
The following below are various forms of validity checks you can adopt in your research writing process
1. Content Validity
This is a logical method of validity check to determine if the whole content of the behaviour section is represented in the questionnaire. We determine the similarities and differences between the task and content of the behaviour. For example, let’s say you want to test knowledge on Cameroonian geography, it won’t be appropriate for the questionnaire to be restricted to a particular region.
2. Face Validity
Fundamentally, face validity means the extent to which a test seems to measure what it intends to measure.
3. Predictive Validity
This is an empirical validity check. when your expectation from a future performance is focused on the scores obtained recently by the measure, you have to correlate the scores obtained with the performance. The previous performance is the criterion, while the current score is the prediction.
4. Concurrent Validity
As the name implies, this validity check is the degree to which the scores on a questionnaire correlates with the test on another already established questionnaire distributed during the same period, or another valid test available at the same time.
For instance, a new simple questionnaire is to replace the old ones that are difficult to handle, but still relevant. Measurements are obtained from the new and old questionnaire at the same time. Predictive and concurrent are similar but differs in the sense that there is no time difference in concurrent validation.
5. Construct Validity
This is the extent to which a questionnaire measures a proposed hypothesis. Psychologists often measures theoretical attributes. The process of validating the outcome of the hypothesis as shown by the test is construct validation.
Reliability means “yielding the same,” in other terms, the word “reliable” mean something is dependable and that it will give the same outcome every time.
Basically, reliability is the extent to which a questionnaire continues to measure whatever it is meant to measure without error. A good research study needs a dependable measurement instrument. A questionnaire needs to be reliable to the point at which they are repeated and any random factor that seems to make outcomes differ under different occasions is a source of measurement error.
The following are reliability checks you can apply:
1. Test-Retest Reliability
This the extent to which outcomes obtained from a questionnaire are consistent over time. It shows the difference in the scores from various testing sessions.
2. Equivalent Forms Reliability
Two questionnaires that are very similar except for the real questions included, is used when there is probability that the respondents will remember their responses during the initial session. The two scores obtained are compared to get a coefficient. This coefficient is referred to as Coefficient of Stability or Coefficients of Equivalent.
3. Split-Half Reliability
Unlike equivalent forms it requires just a single administration. Split half reliability is applicable when the questionnaire is bulky. The Odd-Even method is used to divide the questionnaire into two. Since extensive questionnaire are likely to be more reliable and split- half reliability only represent a percentage of the actual questionnaire, a correction formula must be applied to the coefficient.
4. Standard Error of Measurement
This can also be used to determine the reliability of a test. It is the rough calculation of the occurrence of errors of a specific sample size.
5. Rationale Equivalence Reliability
This check is not determined by correlation but calculates internal consistency by ascertaining how all questions in a questionnaire relates to all other items and to the questionnaire as a whole.
NB: If you need to tests your research instrument (questionnaire) for validity and reliability, kindly send us a message. We will Help you, its free