SPSS Data Analysis-Reliability analysis

The most common measure of measurement is the use of surveys. Reliability analysis is mainly to analyze the stability of the questionnaire measurement results, if the results of repeated measurements are very close, it can be considered that the reliability of the measurement is high. The concept corresponding to the reliability is the validity, and the validity refers to the approximate degree of the measured value and the real value. The difference is that the reliability only describes the accuracy of the measurement tool, and the validity of the measurement tool, the effectiveness of high reliability must be high (effective and accurate), and high reliability, validity is not necessarily high (accurate not necessarily effective)

The theory of measurement based on reliability analysis is divided into two kinds, one is true fractional measurement theory, the other is generalization theory true fraction theory believes that reliability can be expressed by the following formula: X=t+e,x is measured fraction, T is true fraction, E is random error. The validity can be expressed by X=v+i+e, V represents the effective fraction, I represents the system error fraction, and the validity of the error is further decomposed into the system error, but the true fraction is also renamed as the effective fraction.

Reliability can be expressed by the reliability coefficient, different analysis purposes have different reliability coefficients, according to the focus of attention is different, can be divided into internal reliability and external reliability, commonly used intrinsic reliability expression method has Kronabahamian coefficient, binary reliability; the commonly used external reliability expression method has the retest reliability, If the reliability coefficient of the score is greater than 0.8 is acceptable, between 0.7-0.8 indicates the need to modify, less than 0.7, then the scale has a large problem, need to redesign.

Most of the reliability analysis coefficients are included in the analysis-measure-reliability analysis of SPSS, but because some reliability analysis can be expressed using correlation coefficients, the correlation analysis process can also be used.

Let's first look at the most common process

ANALYSIS-Measurement-reliability analysis

This is a questionnaire with 10 questions, designed as a 9-component table, to investigate the reliability of this questionnaire

Let's take a look at the reliability of the ratings, we've modeled 5 ratings on 10 questions, assuming the scores are ordered categorical variables

SPSS Data Analysis-Reliability analysis