Identification method research of invalid questionnaire based on partial least squares regression

The researchers often apply the survey questionnaire as a measurement tool to verify hypothetical propositions and structural models in humanities and social science, therefore, the data quality of survey questionnaire can effect the scientificalness of research propositions and models directly. The data quality of survey questionnaire can be divided into 3 important components in research procedure. Firstly, the quality of survey questionnaire is the foundation for theory research; secondly, the reliability of survey data plays an important role in practical application; thirdly, the scientificalness of analyze data is a guarantee for empirical research. These components are inseparable and conformable, the paper make classification to the data which describes data identification in quantitative and qualitative as classification criterion. Thus, identification method of invalid questionnaire will be divided into 2 steps which are qualitative identification and quantitative one. Identification method for invalid questionnaire based on partial least squares(PLS) regression may apply the basis theory of PLS and SIMCA-P software to form ellipses graphs or ellipsoid ones, which will display specific points outside graphs. In order to delete data of invalid questionnaire, deleting specific points in graphs is needed. The paper uses empirical data to verify the feasibility and scientificness of the method in Customer Satisfaction Index Model. The practice shows the popular value of the method in theoretical research and practical application.