Testing the Web Mode for Collecting Data on Income and Wealth in the Italian Household Budget Survey (303379)

Laura Capparucci, ISTAT 
Massimiliano Degortes, ISTAT 
Romina Gambacorta, Bank of Italy 
Martina Lo Conte, ISTAT 
Loredana Mazza, ISTAT 
*Manuela Murgia, ISTAT 
Andrea Neri, Bank of Italy 
Giuseppina Papadia, Bank of Italy 
Francesca Zanichelli, Bank of Italy 

Keywords: web surveys, income and wealth, mode effect

Many tests and experiments have been carried out on web surveys to study both feasibility and mode effect of this method of data collection. We believe, anyway, that international experiences provides general recommendations that needs to be adapted to the specific surveys needs and cultural characteristics of each country. Besides, to the best of our knowledge, literature lacks studies focusing on household income and wealth.

The survey undergoing our study is the Survey of Italian Household Income and Wealth (SHIW) that has been carried out by Bank of Italy since the Sixties with face to face interviews. The questionnaire has always been interviewer administered, using the PAPI mode until 1998 and, afterwards, through the CAPI technique.

It is well known how the presence of an interviewer can affect the propensity to respond to a survey, especially when dealing with sensitive topics such as income and wealth. For this reason, the Italian Statistics Institute and the Bank of Italy have implemented a web survey on households’ economic situation using a short form of the SHIW questionnaire. A sample of 10,000 addresses was selected from population registers in the same municipalities used for the SHIW CAPI survey.

In the web survey, we also included an experiment, as well as a usability test, for an automatic coding system for the variable Occupation.

In our paper, we will address the following three research questions:

1. Does response behavior relating sensitive topics differ in the two surveys? 2. Have the “don’t know” option and the questions’ wording an effect on the accuracy of the responses? 3. Has the incentive an effect in boosting survey participation and data accuracy?

Data of our study will also be linked to tax records on income in order to have a measure of potential measurement errors in the web survey.