Web Survey Bibliography
AGS covers many topics, including student satisfaction for his/her university experience. Using the complete AGS data set, we calculate two satisfaction factors from a set of satisfaction questions. This is done using the factor analysis (via principal components) method. First factor may be interpreted as a “general satisfaction” index, the second is a contrast between software (teaching, exams, graduation organization) and hardware (classrooms, libraries, cafeterias) evaluation. Factors are not directly observable, but we treat them as they are, for simplicity’s sake. They are continous, approximately normal variables.
Is the non-response process MAR (“Missing at Random”)? To say this we regress the response indicators on all the administrative variables (including sex, age, number of years needed for graduating, graduation mark, high school final mark, faculty, type of course) and the two satisfaction factors. There is a weak evidence that the respondents and non-respondents have different distribution of factor scores.
If we have to take any inference on the non-respondents we must assume that they are related to respondents in some way. The way is often the assumption that they are related through the auxiliary information, that is through variables known for both respondent and non-respondents.
We bootstrap the population to evaluate the ability of the calibration correction to improve the estimators of non-response. We try first to use the quasi-randomization approach to estimate propensity, then use these weights as a basis for calibration. Several combination for calibration variables are used. Faculty is always included as they are main subdivisions of the University and estimates by faculty are routinely required.
The paper analyzes reweighting adjustments for non-response in surveys carrying out a bootstrap evaluation of non-response adjusted estimators. In our study we consider a population made of students from the Bergamo University graduating in a specific period of time.
This population has been surveyed twice (web mode in both cases), before and after graduation. The ante-graduation survey (from now on, AGS) is a compulsory survey, the post-graduation survey (PGS) is not compulsory, therefore there was 56% non-response rate. Administrative (archive) data available for all the students. We apply the non-response process of the PGS in the analysis of AGS data. In this way, we have a controlled situation in which all survey variables, for both respondents and non-respondents are known. We avoid artificial assumptions on the non-response process.
Web survey bibliography - Biffignandi, S. (17)
- Targeted letters: Effects on sample composition and item non-response; 2017; Bianchi, A.; Biffignandi, S.
- Web-respondent-driven sampling; 2014; Bianchi, A., Biffignandi, S., Artaz, R.
- Improving web survey quality; 2014; Steinmetz, S., Bianchi, S. M., Tijdens, K. G., Biffignandi, S.
- Web Panel Representativeness; 2013; Bianchi, A., Biffignandi, S.
- Responsive design for mixed-mode panel data; 2013; Bianchi, A., Biffignandi, S.
- Responsive Design for Web Panel Data Collection; 2013; Bianchi, A., Biffignandi, S.
- Innovation in Data Collection: the Responsive Design Approach; 2013; Bianchi, A., Biffignandi, S.
- Online Data Collection in the Agro-Food Sector; 2012; Biffignandi, S., Artaz, R.
- Panel retention rate and data quality: experimental results drawing on Reciprocity design; 2012; Biffignandi, S., Artaz, R.
- Challenges and pitfalls of measuring wages via web surveys - some explorations; 2012; Steinmetz, S., Bianchi, A., Tijdens, K., Biffignandi, S.
- Web Surveys: Methodological Problems and Research Perspectives; 2012; Biffignandi, S., Bethlehem, J.
- Using survey data collection as a tool for improving the survey process; 2011; Biffignandi, S., Perani, G., Laureti, A.
- Modeling non-sampling errors and participation in Web surveys; 2010; Biffignandi, S.
- Imperfect frames and new data collection techniques ; 2009; Biffignandi, S.
- An experiment on the effects of non-response reweighting on estimators' precision in a web survey; 2009; Fabrizi, E., Biffignandi, S., Toninelli, D.
- The Electronic Questionnaire Experience in Business Surveys: mode effects on quality and on response...; 2009; Biffignandi, S., Siesto, G., Zeli, A.
- Calibration and Propensity Score Weighting in Web Surveys; 2007; Fabrizi, E., Biffignandi, S.