Mind the mode: lessons from a web survey on households finances Romina Gambacorta 1 , Martina Lo Conte 2 , Manuela Murgia 2 Andrea Neri 1 , Roberta Rizzi 2, Francesca Zanichelli 1 1 Banca d’Italia, Via Nazionale, 91 - 00184 Rome 2 Istat, Via Cesare Balbo, 16 - 00184 Rome, Italy Abstract Surveys on household income and wealth are generally carried out through personal interviews. In recent years, Internet has been increasingly used as a data collection method, both for economic reasons and for the timeliness with which the collected data are available. To the best of our knowledge, however, there are no many studies on the use of web for collecting information on household income and wealth. This work contributes to fill this gap by illustrating the results of an experimental survey conducted in 2016 by the Banca d’Italia in collaboration with Istat (Italian National Institute of Statistical). Data quality is assessed through the use of fiscal administrative data and the comparison with a parallel survey conducted through personal interviews. Our results suggest that the web can represent a valid alternative for qualitative questions and to gather information on less sensitive sources of income (such as those for employees and retirees). To reduce the bias associated with the use of this instrument, it is essential to have auxiliary information on the whole selected sample. Key Words: web survey, mode effects, non-sampling errors 1. Introduction Face-to-face surveys are potentially those able to ensure the greatest data accuracy, especially when the CAPI (Computer Assisted Personal Interviewing) methodology is used. This data collection mode also allows to reach any population of interest as each selected unit can be contacted, reached and interviewed. An alternative method of data collection is based on self-compiled electronic questionnaires in a web environment (Computer Assisted Web Interviewing, CAWI). This method has obvious advantages in terms of costs and timeliness with which the collected information is available. Moreover, there are at least three other aspects that make this survey mode appealing. First, the Internet allows the transmission of audio and video material, as well as text. For example, the use of Voice over Internet Protocol (VoIP), such as Skype, would allow "in-person interviews" to be conducted via the Internet. Second, social networks could be useful for making snowball-type sampling designs (Salganik and Heckathorn, 2004, Heckathorn 2002, Poynter 2010). A third interesting element is that the absence of the interviewer could lead to a more truthful reporting behavior in the case of sensitive questions (Tourangeau and Yan, 2007; Kreuter et al., 2008). For these reasons, Banca d'Italia and Istat have decided to collaborate in carrying out an experimental web survey aimed at collecting economic information on Italian households. The experimentation has two main purposes. The first is to investigate the impact of the use of a different survey mode on the quality of answers to complex 168 JSM 2018 - Survey Research Methods Section