Articles published on this site are based on the references made by the editors. These articles are meant only for general information and are not suggested as replacement to standard references. Any inaccurate information, if found, may be communicated to the editor..
An instrument in a research study is a device used to measure the concept of interest in a research project.
It is used to measure a concept of interest.
An ideal measuring instrument is one which results in measures that are relevant, accurate, objective, sensitive and efficient.
Measures which are Physical and physiological have higher chance of success in attaining these goals than measures that are psychological and behavioral.
Instruments can be:
Validity and reliability are two statistical properties used to evaluate the quality of research instruments (Anastasi, 1986).
It is important that assessment techniques possess both validity and reliability.
Validity in relation to research is a judgment regarding the degree to which the components of the research reflect the theory, concept, or variable under study (Streiner& Norman. 1996).
The validity of the instrument used and validity of the research design as whole are important criteria in evaluating the worth of the results of the results conducted.
Internal validity refers to the likelihood that experimental manipulation indeed was responsible for the differences observed.
External validity refers to the extent to which the results of the study can be generalized to the larger population (Polit & Hungler, 1999).
Four types of validity are used to judge the accuracy of an instrument:
(1) content validity
(2) predictive validity
(3) concurrent validity, and
(4) construct validity.
the extent to which different items in the assessment measure the trait or phenomenon they were meant to.
High level of content validity indicates that test items accurately reflect the trait being measured.
A questionnaire to assess anxiety, for example, would be high in content validity if it included questions about known symptoms of anxiety such as muscle tension and a rapid pulse rate.
the ability of an assessment measure to predict someone’s future behaviour in related but different, situation.
An assessment measure with high predictive validity is capable of making accurate predictions of future behaviour.
Low predictive validity means that a measure is of little use in predicting a particular behaviour.
reflects how well different measures of the same trait agree with another.
If a test possesses high degree of concurrent validity, then it can be expected to give results very similar to other measures of same characteristic.
the extent to which a theoretical construct such as a personality trait can be empirically defined.
Reliability of an instrument reflects its stability and consistency within a given context.
Reliability is the consistency of measurement over time, whether it provides the same results on repeated trails.
It is defined as a characteristic of an instrument that reflects the degree to which the instrument provokes consistent responses.
For example, a scale developed to measure intelligence might not be reliable for measurement of personality.
Three characteristics of reliability are commonly evaluated:
internal consistency, and
Test-retest reliability or stability
refers to degree to which research participants’ response change overtime.
Test-retest method is used to test stability of the tool.
In this method an instrument is given to the same individuals on two occasions within relatively short duration of time.
A correlation coefficient is calculated to determine how closely the participants’ responses on the second occasion matched their responses on the first occasion.
Half-split reliability or internal consistency
refers to a measure of reliability that is frequently used with scales designed to assess psychosocial characteristics.
Instruments can be assessed for internal consistency using half-split technique (i.e. answers to one half of the items are compared with answers to the other half of the items) or by calculating the alpha coefficient or using Kuder-Richardson formula.
In the case alpha coefficient and Kuder-Richardson formula, a coefficient that ranges from 0 to 1.00 usually results.
Interrater reliability or the notion of equivalence
It is applicable when different observers are using the same instrument to collect data at the same time.
A coefficient can be calculated or other statistical or nonstatistical procedure can be used to see the correlation of values.