Quality of Survey Questions

Till 1985 the SRF worked mainly to improve the quality of questions as described above. Then we realized that whatever we do, errors will always remain, random errors and systematic errors. These errors can have a strong effects on estimates of the means of variables and on the estimates of the relationships between variables. Therefore, the SRF started to study the errors in questions and, the complement of it, the quality of questions. In doing so we followed Campbell and Fiske (1959) and Andrews (1984) who suggested to use multitrait-multimethod (MTMM) experiments, analyzing the data using Structural equation models. We will introduce briefly the procedures suggested by these authors and after that present the contribution of the SRF to this research tradition.

Estimation of the quality of survey questions:​

•  The MTMM approach  to determine the quality of survey questions
• The True score MTMM model an alternative parameterization of the classic model
• The Split Ballot MTMM approach developed to reduce the burden of the respondents and memory effects
• The EUPD approach developed to be used in case on non-identification of 2-Group SB-MTMM models 

Prediction of the quality of survey questions:

• Prediction by meta-analysis ​
• Prediction by the software SQP