You can understand it as reliability, consistency, and stability. For example, when the same object is measured, the results of multiple measurements are very close, and everyone will think that the result is credible and real, that is, the reliability is high. If there is a large difference in the results of each measurement (for example, the same test paper is used to test the same group of students, and the average score of each test varies greatly), then the reliability is low. Similarly, in questionnaire research, reliability analysis is also used to measure whether the sample answer results are true and reliable. The higher the test reliability, the more credible the result.
Reliability can be divided into two types: intrinsic reliability and external reliability. Intrinsic reliability: refers to whether multiple items in the research questionnaire measure the same concept or content, that is, whether there is internal consistency between the items. External reliability: usually refers to whether the measurement results are consistent when measured at different times. For example, the example of test scores mentioned in the example just now.
Validity analysis, in simple terms, is the validity and accuracy of the questionnaire design. When we design a questionnaire for a research topic, we all hope that what the question actually measures is what we want to measure, so that the research data can accurately explain the problem. For example, if we want to know the comprehensive grades of students in a class, if we only test one subject of mathematics in the formal research, and then draw a conclusion, the validity of such a research may be very low, because there is a gap between the actual measurement direction and the research direction. Big deviation.
Validity can be divided into content validity, construct validity, convergent (convergent) validity, discriminant validity and so on.