We specialize in surveys of all kinds. Since the turn of the millennium we have conducted more than 1000 surveys and questioned more than a million people throughout Switzerland.
Surveys help our clients to get a glimpse of public opinion and the needs of target groups and stakeholders. They identify respondents’ awareness, attitudes, moods or desires. There are many methods for gathering empirical data. The aim of all methods is to make and quantify observations on real world data. Together with our clients, we select the method best suited to the question as well as the size of the population sample.
Surveys or market and opinion research are still on trend despite big data and digitization. They form the basis of our work and enable us to produce verifiable analyses based on the systematic polling of opinions. They also question “perceived truths” or subjective assessments of individuals.
Many of our clients have enjoyed our support for years. To answer concrete questions, we develop bespoke projects for them, in which the survey forms the core element of the project. The results of our work help our clients in their strategic communications and in the development of a genuine dialogue.
"Surveys are a part of almost all the projects we work on. I never cease to be surprised that the prior estimates our clients make are very far removed from the results that we obtain using our methods.”
Who do we conduct surveys for?
Our clients include companies, political parties and associations, cantons and NGOs. We come up empirical and evidence-based answers to concrete societal, political or economic questions for them. The questions range over a wide variety of topics, such as the EU, concerns about everyday life, the future of the Roman Catholic church, services and products, campaigns, perception of democratic principles or e-mobility.
"Our clients appreciate our proven expertise in public opinion, societal trends and political processes – and also our huge experience in dealing with the media, as well as our interactive tools, which we use not only in our collaboration, but also in reporting."
gfs.bern expertise in the organization, conduct and analysis of surveys
For the development and design of questionnaires, we regularly fall back on our tried and tested gfs.bern methods: focus groups, qualitative in-depth interviews, planning and accompanying studies for campaigning, data modelling, regional analyses and media analyses. For the optimum development of quantitative questionnaires and representative surveys, we draw on the existing groundwork and expert knowledge of our clients.
The starting point for any quantitative survey is the data collection based on a rigorous questionnaire. We gather data using face-to-face interviews in the field or on the phone – or using the online panel polittrends.ch. More than 200 associates conduct face-to-face (FtF/F2F) interviews, computer-aided personal interviews (CAPIs) and computer-aided telephone interviews (CATIs) in all language regions of Switzerland.
We also have our own online panel with several tens of thousands of registered participants. This enables us to conduct online surveys within a very short time. Using the online panel, we can also conduct panel interviews in which the same study participants are questioned several times over a prolonged period (panel studies). For example, we carried out a panel study of this kind on the subject of Brexit for the University of Zurich.
“The study served as a valuable basis for our change of name from CVP to The Centre.”
Problematic aspects and points of criticism
Market and opinion surveys often attract a lot of media attention. This also has to do with the topical nature of the questions surveyed. Despite the sophisticated quality of the methods, the high level of media attention brings with it some problematic aspects and points of criticism. We take these points seriously and include them in the design of the survey:
- With all surveys, questions of how representative they are, especially with open online surveys (opt-in survey) and online panels
- Respondents are negatively disposed to telephone surveys and the willingness to participate falls with these surveys
- Some surveys are hounded by the reputation of being surveys designed to serve political-ideologies
- Survey results depend too heavily on the quality and objectivity of the method and weighting processes. These are said to be not transparent enough.
- Opinions expressed by respondents differ from actual actions: if respondents hold opinions/behaviours perceived to be socially undesirable, some will tend not to admit to them or give specious answers
- The interviewer may provoke answers using suggestive questions and thereby confirm preconceived opinions.
- Sample populations selected tend to be to small
"With every referendum, our trend studies are measured against the real world. This compels us to keep developing our methods and making sure they are representative. The trend for good-quality methods in survey research is moving in the direction of combining methods and improving weighting models."
Frequently asked questions about surveys:
One-off or repeated surveys
We conduct one-off surveys for our clients or and also regularly repeated surveys on specific questions. Surveys that are conducted periodically enable us to draw conclusions about changes over time that amplify what are otherwise only snapshots of opinion. Examples of recurrent surveys are the Credit Suisse Worry Barometer, the TCS Travel Barometer, the Health Monitor for Interpharma and the Banking Barometer of the Swiss Bankers Association..
When is a survey representative?
Representativeness is achieved when a sample population is as similar as possible to the general population in its composition and structure of relevant characteristics. Data on the structure of Switzerland and its society are constantly fed into our models for this purpose. To reach representative conclusions on German, French and Italian-speaking Switzerland, enough people must be surveyed in these language regions. Population sizes that have proved reliable are a minimum of 700 people for German-speaking Switzerland, about 250 for Suisse Romandie and 50 to 100 for Svizzera Italiana. For representative nationwide surveys we recommend at least 1000 respondents.
How do we form sample populations?
Our survey experts use different methods for determining sample populations depending on the questions – for example, random digit dialling (RDD) in telephone surveys. This procedure was developed to counteract the declining quality of telephone directories. With this approach, no numbers are selected form a directory, but instead a systematic algorithm is used to select the numbers. But other procedures are also used, such as Bayesian weighting for dealing with non-representative population samples or pure quota samples, if the aim is to survey only people aged over 65, for example.
What results and information are published?
In principle, it is at the discretion of the client who commissions a representative survey whether the results are published or not. If they opt for publication, the basic principle is that all results are disclosed. The industry association SWISS INSIGHTS, which represents the relevant research institutes in Switzerland, also demands of its members that they publish the technical details on the survey. Opinion researchers are required to state who the client is and who is responsible for the survey, what the sample population was, when the survey took place and what the statistical significance levels are. The research institute gfs.bern publishes these technical details in all the published surveys in on its own website in short or key reports.
Are surveys also forecasts?
Survey research is presence research. The aim is to provide a reliable picture of the present situation. This snapshot is often sufficient as a guide and a basis for decisions. But a survey as it stands does not say anything about the future. For conclusions about the future, i.e. forecasts, we would have to apply special data modelling procedures, as they do in meteorology for weather forecasts. This is only done by gfs.bern in the context of recurrent surveys, i.e. involving at least one repetition of the same survey.
Do surveys influence referendums or elections?
esearch is working intensively on possible influences of surveys on elections and referendums. The following effects have been described:
- Mobilization effects: surveys can lead to greater participation in the decision, especially when the outcome is uncertain.
- Defeatism effect: surveys can reduce the mobilization of likely losers.
- Lethargy effect: surveys can reduce the participation of presumed winners.
- Complacency effect: surveys can reduce the participation of undecided voters.
- Bandwagon effect: surveys can lead to a change of opinion in favour of the likely winner.
- Underdog effect: surveys can lead to a change of opinion in favour of the likely loser.
Although numerous studies have been conducted on these theoretical effects, a marked and statistically significant effect has never been demonstrated in practice. Recent findings suggest that an effect is possible when surveys indicate a close result. This usually mobilizes additional voters, and the side that benefits in this case tends to be the one that has the edge in surveys. We assume that surveys have hardly any effect on the outcome of referendums or elections. However, they can impact the decisions of active players. For example, how they do what in the course of a campaign. And how they organize their strategic communications. In our experience, they also influence the assumptions as to the outcome that can be expected.
On behalf of the Swiss Broadcasting Corporation (SRG / SRR), for example, we conduct the trend surveys on federal and cantonal referendums and also follow-up analyses.