VALIDITY AND RELIABILITY

VALIDITY AND RELIABILITY

Validity and reliability concepts are used in the evaluation of the quality of research. They show how well a test or technique measures things. Validity deals with the accuracy of a measure, while reliability deals with the consistency of a rate. These are two related concepts in research; a test can be reliable and not valid, but a test is not valid unless it is reliable. The consistency of the results of a measure makes it reliable, but for it to be valid, it has to measure the required output. Validity is challenging to access compared to reliability. However, it is estimated through comparison of the results obtained to other relevant data, while reliability is determined by a comparison of several versions of like measurements. These two aspects are essential in research to ensure results are compelling.

The degree to which the outcome measures is what is intended is validity (Haradhan, 2017). It is assessed by confirming the correspondence of results to established theories and other counts of the concept, comparing the output obtained to other r relevant data. For a result to be considered valid, the results got to conform to characteristics, real properties, and variations of the physical world. There are several types of validity, and they can be evaluated through statistical methods and expert judgments. The examples of validity include criterion, construct, content internal, and external validity.

Criterion validity is the extent to which a test can precisely predict certain criterion variables. They are predictive and concurrent validity; predictive validity involves measure that predicts the future performance of another while concurrent validity requires administered steps at a similar time. It assesses the extent the results measured correspond to other valid measures of comparable concepts (Taherdoost, 2016). Correlated coefficients measure it through statistical computations. An example is a survey involving voters of a region to measure their political opinions. The criterion validity of the review is high when the later outcome of an election is accurately predicted by the results obtained from the voters of that region.

Content validity is another type that involves the actual content of a test that should examine all aspects that describe the objective adequately (Chiang, Jhangiani, & Price, 2015). It is more of a qualitative measurement. It assesses the extent to which all aspects of the concept measure are covered (Middleton, 2020). A test aims to measure the reading, writing, and speaking components of students in a Spanish class but does not involve the listening component. The analysis lacks content validity in measuring the general ability of students in the mastery of the Spanish language because listening components have been exempted from the test. In language, ability listening is an essential aspect.

The third type is construct validity, which is the extent to which the intended construct, skill, ability, or attribute that exists in the human brain based on established theories; is measured by a test. Its components are convergent and discriminant validity. Convergent validity relates to how a measurement concurs with others that measure a cognate construct. In contrast, discriminant validity refers to the differences in the analysis of two concepts that do not correlate (Taherdoost, 2016). It assesses how a measure adheres to the existing knowledge and theory of a measured idea. Traits assumed to be related, and those related to self-esteem can be measured in a self-esteem questionnaire. High construct validity could be associated with a strong correlation between associated traits and self-esteem.

Internal and external validity are considered in the assessment of a cause and effect relationship validity. Internal validity is the degree to which the research study design and the yielded data enable one to make accurate conclusions of the cause and effect (Haradhan, 2017). Enhance internal validity, double-blind experiment, controlled laboratory study, and unobtrusive measures are required. Studying how humor affects sales in television commercials, two adverts are compared, one with a sense of humor and one without a sense of humor. The time the advert with a sense of humor was aired, there were more sales hence the validity of humor. External validity scrutinizes research’s implications and attempts making it realistic. In enhancement, a sample that is representative, a real-life setting, and replication of a different context is essential.

When research is done several times under similar conditions, the extent of the results’ productivity is termed as reliability. It is assessed by looking at the consistency of results from different observers at given timelines and also parts of the tests; compare various versions of a similar measurement. For a result to be considered reliable, the same effect has to be obtained regularly using the same method and similar circumstances (Taherdoost, 2016). If the reliability coefficient is high, for example, r = 0.98, we can suggest that both instruments are relatively free of measurement errors. If the ratios yield above 0.7, are considered acceptable, and coefficients yield above 0.8 are considered very well (Haradhan,2017). There are various types of reliability, and they can be estimated through given statistical methods. They include interrater, internal consistency, and test-retest reliability.

Interrater reliability gauges the reliability of unlike individuals. It assesses the consistency of a measure from different observers (Haradhan, 2017). The existence of multiple observers is an assumption, which is not always the case. It is often used in the testing of people’s similarity in scoring items and alike categorizing items by people. For resembling, student project examiners submitted conflicting results based on an assessment criteria checklist. The assessment checklist has low interrater reliability because it is too subjective. The other reliability is internal consistency, which rates dissimilar queries comparing their ability to give appropriate and uniform results. It involves average inter-item, average item-total Cronbach’s alpha, and split-half correlations.

The same contrast tested from pairs of questions trough the calculation of the mean correlation is compared through the average inter-item correlation. Calculated total score and averages for average inter-item correlations are the average item-total correlation. Split-half correlation divides into two tests items that measure the same contrast applies to comparable groups of people calculating the correlation between the two scores. Chronbach’s alpha helps to obtain an average of the possible split-half correlations (Chiang, Jhangiani, & Price, 2015). The calculation equation is: –

a=

Where a is Chronbach’s alpha

N is the number of components

r-bar is the average of all Pearson correlation coefficients

it assesses consistency in the measurement; did you obtain matching results from unlike portions of a test intended to measure the same things? To measure self-esteem, you design a questionnaire and slit the results into two halves with a strong correlation. The relation between the two results indicates their internal consistency.

Test-retest is the third type of reliability that analyzes reliability across time (Haradhan, 2017). It varies with interruptions, mood, time of the day, and the response to a test. It assesses the consistency of a measure across time; do you obtain resembling output when you redo the measurement? A group of people fills a questionnaire to measure personality traits. If the refill the same poll after a given span and the corresponding answers are offered, then there is a high test-retest reliability.

Validity and reliability of results require the creation of robust research design, selection of appropriate samples and methods, and conduction of careful and consistent research. To do so, one has to ensure validity and reliability. In ensure validity, appropriate measurement methods have to be considered and selection of subjects with the proper sampling method. Validity can be achieved by using high-quality measurement techniques thoroughly researched based on available information that measures the exact output (Middleton, 2020). For example, the use of standardized questionnaires for personality traits and not individually developed ones. For designed inquiries, the questions should be precisely and carefully worded. In sampling enough, participants representing the available population would help in obtaining data that is not biased. Reliability can e ensured by the application of consistent methods and standardization of research conditions. The same steps and way should be done for all measurements, especially in the involvement of multiple researchers. Terms of data collection should be kept constant to avoid errors and variations in collected data due to internal and external factors.

Reliability and validity are necessary measurements that should be considered in the analysis of research to ensure the attainment of goals and objectives (Middleton, 2020). The variability and reliability of a measure are maybe by scrutiny of results across various studies and not by one study; it is a continuous process (Chiang, Jhangiani, & Price, 2015). Variability and reliability have been essential concepts in research.

References

Chiang, I.-C. A., Jhangiani, R. S., & Price, P. C. (2015, October 13). Reliability and Validity of Measurement. Retrieved February 13, 2020, from https://opentextbc.ca/researchmethods/chapter/reliability-and-validity-of-measurement/

Haradhan, M. (1AD). Two Criteria for Good Measurements in Research: Validity and Reliability. Annals of Spiru Haret University, 17(3), 58–82. Retrieved from https://www.citationmachine.net/apa/cite-a-journal/search?utf8=✓&q=https://mpra.ub.uni-muenchen.de/83458/1/MPRA_paper_83458.pdf

Middleton, F. (2020, January 13). Reliability vs. Validity in Research: Differences, Types, and Examples. Retrieved February 13, 2020, from https://www.scribbr.com/methodology/reliability-vs-validity/

Taherdoost, H. (2016). Validity and Reliability of the Research Instrument; How to Test the Validation of a Questionnaire/Survey in a Research. SSRN Electronic Journal, Vol. 5, No. 3, 28–36. doi: 10.2139/ssrn.3205040