One theme that has proved difficult to understand and implement studies and thesis work is to define and test the validity and reliability of research tools. And it becomes difficult, being that there are few sources that develop in depth the identity of these elements and how they are documented.
considerations before
Important is to know that the validity and reliability is measured directly in research tools that apply , not to the studies. Therefore, we seek a questionnaire or interview guide is valid and reliable. If the instruments have these two characteristics, namely the study is valid and reliable by inheritance.
will have to start from the beginning, defining each element and explain in practical terms.
Reliability
is traditionally defined as the degree of consistency of data provides a tool when working on an equal footing .
Measure the reliability of an instrument for reasons statistical representativeness. The premise of sampling indicates that it can retrieve data from a small number of elements, and these data are very similar to the population curve as a margin of error. This representation is reached through certain sample size calculations and determination of appropriate sampling model (how to identify and select the respondent), and if all goes as expected, you can declare a reality for millions of people, knowing information from a few hundred.
But to achieve this, the instrument and the respondent's seleccción are critical. The responses should represent a steady and consistent curve in the conditions under which the responses were collected. Factors such as whether survey before and after an election period, a natural disaster, an economic recession after a night sale, after a periodicazo against a product or service ... All these elements can not be controlled directly affect the reliability.
To build confidence, we must revise the wording of the instrument - to obey a neutrality as resistance to exposure to environmental changes is strong - and the timing of sampling - the time when information is collected are constant, equal in status.
To measure reliability, a series of tests are constantly used.
The easiest but less powerful validation is to divide the sample into two timeless surveys, collect information and compare the results by measuring the statistical differences of proportions (cough test, as applicable). Since information is the same and what changes is the time, significant differences can be very common question the reliability of the instrument, as the data produced are not consistent. For this method to be effective, control of sampling in both scenarios is essential to ensure that the conditions and characteristics of respondents are constant at all times.
Another way to validate reliable is the so-called Cronbach alpha coefficient. Is an index of internal consistency, with values \u200b\u200bfrom 0 to 1. In essence, a squared correlation coefficient which measures the homogeneity of the questions, averaging all inter-variable correlation, validating that indeed resemble. As the ratio approaches 1, the more we talk about reliability. From a value .8, talking about a respectable level of reliability and usable.
To calculate this quickly, you can use SPSS or similar program, containing the function. Important to note that for the test is feasible, all variables must be able to combine additively to score total, and all must measure a situation in the same direction.
Validity
In simplified terms, the validity is sure that you are measuring what you want to measure . One element in that less attention is paid and is vital for any instrument piece of research is ensuring that the instrument is appropriate to the research objectives.
I say that often pays little attention because the assertion is obvious: If I learn of pears, apples do not wonder. The reality is more complex than that. When developing an instrument with little information background, it tends to run into errors of wording and content that has been gathering information unwieldy, unresponsive to specific targets. Usually ask for more, or often omit transcendental aspects.
Validity is, therefore, an entire branch in the design of research instruments. There are three basic types of validity:
Content validity: It reflects the ability of the items reflect the construct assessed population. The content, the theme of the variables must be members of the concept being evaluated. Optimally, the specific objectives of research are intertwined to the content of the instrument.
criterion validity: It has to do directly with the scales used in the measurement, which must be faithful to the natural way to discriminate between the concepts measured. That ratings are understandable, that the ranges are typical behaviorally speaking, are examples of this effect.
Construct validity: Right, if the tool So in full measure and reflects the identity of the construct, the topic under investigation. Construct validity is the result of the cohesion of content and approach, designed the general aim of the study.
to construct validity, it is necessary to invest enough time and dedication to the design of instruments that include each and every one of the issues identified in the objectives of the study. It is a process of trial and error. Pilot testing of the instrument should aim to validate the understanding of the respondent, the attitude of responsiveness and coherence of the topics covered. Measure
validity is not easy, and typically responds to the experience and results of pilot tests. Call on experts in the field of instruments and style editorial correction, and validation visits and consistent with the project leaders are the most common ways to validate an instrument.
The relationship between validity and cofiabilidad
synthesize this relationship in a sentence: "A reliable instrument has validity, but not enough" . The reliability ensures that the results of a survey follow the curve of behavior over time, but not sure you measure right.
Validity is defined before the uprising, the reliability is validated by the completion. Reliability is the time and place, and criterion validity is desktop.
Below is an analogy of the relationship between reliability and validity. A reliable instrument Diana hits repeatedly, looking closer to reality. The more reliable the instrument, most about the target. But if the instrument is invalid ... then you are hitting wrong target.
0 comments:
Post a Comment