I have a dichotomous data and a likert scale and have to analyze variables of both, so in that case we can check the reliability ,but when it comes to validity how to check the validity of these two different construct of questions in which one is dichotomous set of questions and the other one is likert, can we check the validity of both of them as we do in factor analysis?
Factor analysis is not a method for checking validity, which in a survey context is defined as evidence that a variable measures what it is supposed to measure. So, if your variables pass the test for reliability, then you can create the appropriate scales and test those scales for validity. The standard criteria for validity are frequently summarized in terms of construct validity (convergent validity, discriminant validity, etc.)
software is available to check validity and randomness of data compiled .when we compile Likert data from samples we design questions in such way that respondent has to read and cant tick questions in routine lie third choice an all questions .It is very important to design right format so survey is not rejected. Then its randomness and is checked. Regarding validity. Validity means the data obtained reflect the variable relations ship in real and work every time .So normally we first do pilot test or mock run on data to see if the model designed is valid. Then only full survey is done.
Zoiya Naz The amount to which a measure (such as a survey, exam, or questionnaire) genuinely measures what it is designed to assess is referred to as validity. There are numerous methods for determining the validity of dichotomous and Likert scale questions.
One technique to assess the validity of dichotomous questions is to compare the findings to a "gold standard" metric. For example, if you test blood pressure using a dichotomous question (e.g., "Is your blood pressure normal or high?"), you may compare the findings to genuine sphygmomanometer values.
A test-retest reliability study, in which identical questions are delivered to a group of people at two separate times, is another technique to examine the validity of dichotomous questions. A significant correlation between the two sets of findings indicates that the questions have good test-retest reliability and, as a result, validity.
Component analysis is one method for determining the validity of Likert scale questions. A statistical strategy for identifying underlying components or dimensions in a set of data is factor analysis. If the variables discovered by the factor analysis correspond to the constructs that the Likert scale questions were supposed to evaluate, then the questions have strong construct validity. Another method for determining the validity of Likert scale questions is to compare the data from the scale to external criteria, such as a behavioral measure.
It is vital to remember that validity is an issue of degree rather than binary quality (either valid or invalid). A measure may be appropriate for certain purposes but unsuitable for others. Furthermore, it is critical to examine both data reliability and validity, because a measure might be reliable (consistently measuring what it is designed to measure) but not valid (not measuring what it is supposed to measure).
Yes, you need to examine correlations with theoretically specified variables.
Though Cronbach Alpha is used to check for reliability, smart pls helps you to examine the validity and reliability of the instrument. Construct and face validity aside correlation can be used to test the validity of dichotomy variables.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis