Postcard from the Field
Modes of Data Collection in the 2017 Norwegian National Election Study
Bernt Aardal and Johannes Bergh
In our Postcards from the Field series CSES collaborators provide an update and commentary on election studies recently in the field.
The first Norwegian National Election Study (NNES) was conducted by Stein Rokkan and Henry Valen in connection with the 1957 parliamentary election. There was no election study in the following 1961 election. The next time around in 1965, and in every Norwegian parliamentary election since then, election studies have been carried out. As a result, the NNES is one of a handful of election studies programs globally that has time-series data that spans 60 years or more.
With the exception of the 1965, 1969 and 2001 studies, where the design included a pre- and post-election study, most studies have been based on a post-election design. From 1977 onward the design includes a rolling-panel, where half of the previous sample is re-interviewed at the next election. Since 1997 the NNES has taken an active part in the Comparative Studies of Electoral Systems (CSES) and have included all CSES modules to date.
Face-to-face interviews have been the primary mode of data collection for the NNES. For the first few decades this was the only means of interviewing respondents. In the 1990s, telephone became a possible alternative way to interview people if it was more convenient than doing it face-to-face. Gradually, telephone-interviews took up a larger share, until 2013 when close to half (45%) of the respondents replied over the phone.
It seemed to us that the 2017 NNES was the right time to start thinking about alternative modes of data collection, for a couple of reasons. First, because the secondary mode of data collection (telephone) was about to become the primary mode, and this is probably not the ideal means of replying to such a long survey questionnaire (about 1 hour). Second, the costs of face-to-face interviews have risen dramatically over time. Though funding for the 2017 study is adequate, it was clear that at some point we would have to implement a type of a web-based data collection.
Instead of waiting for the NNES to be forced to abandon face-to-face interviews altogether, we decided to try two different modes of data-collection in the 2017 NNES. The sample was split in two. Interviews in the first half are done “the old way”, with face-to-face as the primary mode, and telephone as the secondary mode. The second half gets a web-questionnaire, and is followed up on the phone as a secondary mode.
The sampling strategy for the 2017 NNES is identical to previous election studies and is the same for both halves of the sample: a random population-based sample. Statistics Norway does the fieldwork, and they have access to an official database of email addresses and cellphone numbers. This enables them to contact respondents electronically, without having to do any adjustments to the sampling strategy.
The main challenge in the planning phase of the 2017 NNES has been that we had to cut the size of the questionnaire by about one third. A web survey that takes one hour is simply too much. We therefore had to run the CSES module as a separate web survey with a population based sample, but with a larger sample size than in previous studies.
So how does it all work? Data collection is still ongoing, but the reports so far are that responses are coming in at a pace comparable to previous Norwegian election studies. We have been concerned about open response categories which we for instance use to tap the “most important issues in the campaign”. In web surveys, respondents have to type in their response to these types of questions. The data that we have so far indicate that the share of nonresponse is almost the same as in previous years. Hopefully, this is an indication that data quality in the 2017 NNES and CSES studies will be on par with previous years. We cannot wait to find out when we get our hands on the final data files in early 2018!
Bernt Aardal is a Professor of Political Science at the University of Oslo and Principal Investigator of the Norwegian National Election Studies since 1985. His research interests include political behavior, public opinion, electoral systems and electronic voting. Johannes Bergh is a Senior Research Fellow at the Institute for Social Research in Oslo and head of the Norwegian National Election Studies. His research interests include political participation, experimental designs, voting, political behavior among minorities, electoral systems and democratic innovations.