Content validity

Content validity (german content validity ) refers to multivariate statistics one aspect of construct validity and is present when the measurements of a construct whose content in all its aspects fully grasp. Content validity closes the gap between a mentally - theoretical construct and its measurement by an existing indicators of scale.

Example

A teacher wants to measure the intelligence of their students using a test. Intelligence is thus in this case the construct is to be measured, the test scale. To this end, it provides three computational tasks to each a measure of intelligence. Obviously with such a test is no content validity because the computational skill is only one aspect of intelligence. To increase the content validity, a definition which instead would be first necessary to what intelligence really is. Through discussions with experts (such as intelligence researchers ) and a search of the literature on intelligence could now find indicators for the different aspects of intelligence. Without the balance of the aspects would be a discrepancy between the construct and measurement scale.

Statement

Content validity is only one component to determine the construct validity of a construct. Other modules include discriminant validity, and nomological validity Konvergenzvalidität.

Content validity can be typically not determine objectively with a statistical characteristic. John G. Wacker (2004) highlights the importance of formal conceptual definitions as the most important step before any traditional statistical validity test is performed. A construct thus must - be defined - by those skilled in approximately based on a literature search or on the basis of interviews. Based on the definition can be - again on the basis of literature and those skilled - identify possible indicators of the construct. There are various methods by which it can be determined whether each individual indicator or indicators jointly capture the content of the construct in all its aspects, in full or whether a unilateral deviation from the construct is present through issues not addressed.

Lawshe procedure

One known method of estimation of the content validity dates of Lawshe (1975). This refers to the extent to which a group of experts ( judges ) is a general consensus as to whether the knowledge measured by an indicator "essential" is " useful but not essential" or " not necessary " for the measurement of the construct. The criterion for the content validity is that at least half of the jurors must agree that the indicator is classified as "essential".

Moore Benbasat process

Another well-known subjective method that captures alongside other aspects of content validity, construct validity, by Moore and Benbasat (1991 ) was developed. Here organize jurors on index cards noted indicators on the one hand to dial in yourself and calling himself categories ( thus to constructs ) and secondly into predetermined categories. Cohen's kappa and the indicator classification rate ( engl. item- placement ratio) are thereby used to determine the judgments interrater agreement. An extension of the method may be to ask the jurors in each case, be able to identify missing aspects of the construct and formulate indicators to cover these aspects.

Criticism

The mere consideration of Konvergenzvalidität and discriminant validity for the detection of construct validity is criticized especially by John R. Rossiter, by citing that the construct must be obtained independently of other constructs. He emphasizes the importance of content validity and sets them equal even with construct validity. Thus, measures to improve the discriminant and Konvergenzvalidität can cause indicators are removed and statistically measurable properties of the measurement models be improved by it, but to remove the measurement models simultaneously by the semantic content of their constructs.

The overall conclusion is that measures in the past to define a construct, and in particular to improve the content validity is often not the necessary attention was paid while to improve purely objective statistical quality criteria such as Cronbach's alpha or the goodness of fit of a structural equation model often at the expense of content validity rashly indicators have been deleted. Objective and subjective criteria to ensure the construct validity must go hand in hand instead. In particular, the content validity over and be kept constantly in mind has to, as well as a project carried out at the beginning of the scale development process procedure as that of Moore and Benbasat not prevent one taking place on the end frivolous Delete ( "scale purification" ) of indicators in the course of Examining the Inhaltsvaldität destroyed Konvergenzvalidität and discriminant validity again. If indicators due to other validity tests, or reliability ( eg, Cronbach's alpha ) must be deleted, must remain sufficiently many indicators for each of the substantive aspect of a construct. The remaining indicators must continue to measure in their interaction, the construct excellent. The developer of a scale is therefore often no other choice than both at the beginning and at the end of the scale development process to verify the content validity.

Swell

• Multivariate Statistics
• Descriptive Statistics