Validity and Reliability Types of Validity According

Validity and Reliability

Types of Validity

According to Trochim (2007), there are six different types of construct validity. Construct validity pertains to the accurate reflection of the operationalization's construct within the conclusion. Of the six types, Trochim (2007) divides them up to translation validity -- which depends on the definition of the construct and the checks against it -- and criterion-related validity -- which applies one's theory of construct and examines the proper behavior of the operationalization. The two translation validity types are thus: face validity, which views an operationalization solely on its face value; and content validity, which juxtaposes operationalization with the content domain of the construct. The four criterion-related validity types are thus: predictive validity, which uses prediction to evaluate operationalization; concurrent validity, where operationalization is gauged in its being able to distinguish between two similar groups; convergent validity, a gauging of how similar an operationalization is to other operationalizations; and discriminant validity, a gauging of how dissimilar an operationalization is to other operationalizations.

To view these validity types through examples, one can find that the apprehension of a perpetrator in the criminal justice system (cops and robbers, to plainly speak). Through face validity, a cop merely observes suspicious conducts of individuals prowling around at night and acts upon the conclusion that such conducts should be further investigated in case of future. This can be further realized in content validity, where certain criteria can further validate the cop's reasons for continuing to watch and maybe follow the prowling individuals. Predictive validity would entail the cop's experience in such matters, and his or her prediction of how likely it is for the shady individuals to do wrong. In concurrent validity, the cop should be able to distinguish between this shady event as a minor theft or a large-scale robbery. Through convergent validity, the cop would find similarities within the typical characteristics of thieves and apply this connection to the group of individuals. Finally, through discriminant validity, the cop would find little similarities within the characteristics of thievery and another type of crime when it comes to the group of individuals.

Types of Reliability

In Trochim's (2007) text, reliability relies on the measurement of one's consistency in particular matters; when a result is the same after a repeated amount of observations, then a subject is reliable. There are four types of reliability measurements when it comes to research: inter-rater or inter-observer reliability, which measures the consistency of ratings through the use of observers or raters; test-retest reliability, which repeats the collecting of measurements at a different time in order to assess consistency; parallel-forms reliability, a method that assesses consistency through the results collected from two tests constructed from the same content domain; and internal consistency reliability, a method that assesses consistency through the results collected from different items on the same test.

These types of reliabilities can be further discussed within research examples that, once again, pertain to the criminal justice system. Here, one will talk about witness reliability. The reliability of a witness is in question when two different witnesses have two completely different accounts of a particular event. In order to deliver a solid testimony, a lawyer attempts to calibrate witness accounts in order to increase reliability within the inter-observer reliability type. The test-retest reliability measure enables a cross-examiner to confirm that the witness' testimony remains the same, even when questioned after a given period of time has elapsed. In cross-examinations, the parallel-forms reliability can be measured through the use of different -- though similar -- tactics of questioning, where one set of questioning might be more effective than another. In a more specific manner, internal consistency reliability focuses on similar types of questions within a particular set of questioning, and whichever gives a more positive result is the one used for further research.

Random and Systematic Error

Trochim (2007) defines random error as a result of measurements wherein factors are randomly affected. An example given is the random mood experienced by a particular individual on a measurement day. Mood is affected by many random things, and can produce errors in a measurement when it comes to human results. Systematic error, as defined by Trochim (2007) faces a bias in the collected measurements, as it is caused by an element that affects the entire experiment. This type of error notes the example of a loud event happening outside of a classroom filled with test takers; the loudness of the event distracts the test-takers and thus diminishes their test-taking potential. Both types of errors, however, can be remedied through certain careful methods.