Surveys: methods, procedure, publications | GFS Bern

Surveys and opinion research: Frequently asked questions

gfs.bern specialises in representative surveys and data analyses. We have summarised the most frequently asked questions about methods, procedure and publications for surveys.

gfs.bern ist Spezialist für repräsentative Umfragen in der Schweiz.

Since founded as an independent institute in 2004, gfs.bern has conducted more than 1,000 surveys. We have reached more than a million people throughout Switzerland.

Who are the clients commissioning representative surveys?

Our clients include companies, political parties, associations, cantonal governments and non-governmental organisations. We have been working with many of them for years, providing analyses, help with their campaigns and other services to support their strategic communication. They value our proven expertise in the fields of public opinion and political processes, our far-reaching experience in the use of media, and our state-of-the-art tools.


Once or repeatedly?

In addition to one-off studies on specific topics, we also offer our clients periodical surveys. This grants them valuable insights into changes that happen over time. Our regular surveys include the Credit Suisse ‘Worry Barometer’, the TCS Travel Barometer and the Bank Monitor for the Swiss Bankers’ Association.

Face to face, over the phone or by online panel?

We use our own, dedicated survey service. More than 200 employees are available for face-to-face interviews (FtF/CAPI) and computer-assisted telephone interviews (CATI) in all linguistic regions of Switzerland. They specialise in all forms of field research, particularly quantitative and qualitative surveys.

We also operate our own online panel with several thousand registered participants. It enables us to carry out online surveys at very short notice. Even with large samples, we can ensure rapid response times. During a free initial consultation, we will advise you on your best option for your task: online, offline or both.


How long does a representative survey take?

It usually takes around eight weeks from the initial consultation to the completed evaluation. In a first step, our project team determines the client’s specific requirements. It decides on the most efficient representative survey methods for the case at hand. Working closely with the client, we then develop a questionnaire and test its feasibility using a trial run. If everything is in order, we launch the online and/or offline survey(s). As soon as the required number of participants have responded, our data analysts start evaluating the results. Finally, our project managers turn the facts and figures into written and visual content.

What makes a survey representative?

A survey is representative when a random sample resembles the population in its composition and the structure of its relevant characteristics as closely as possible. We use computer-assisted models to achieve representativeness. To obtain representative statements about German-speaking Switzerland, French-speaking Switzerland and Ticino, a sufficiently large number of people need to be interviewed in each area. The following numbers have proven successful: around 700 people for German-speaking Switzerland, around 250 people for French-speaking Switzerland and 50–100 people for Ticino. Surveys that are intended to represent all of Switzerland, then, ideally require at least 1,000 respondents.

How do we put together our samples?

Our survey experts use a range of methods to determine samples, depending on the case at hand. Random digit dialling (RDD) is one of them. This method was developed in response to the deteriorating quality of telephone directories. Rather than being selected from a directory, numbers are picked and dialled by a systematic algorithm. We also use a wide range of other approaches, such as Bayesian methods based on non-representative random samples and quota samples in cases where, for instance, only people over the age of 65 are to be interviewed.

What results and information do you publish?

Survey-based research focuses on the present. Its purpose is to represent the current situation reliably. Such a snapshot is often sufficient as a means of orientation and a basis for decisions. In itself, a survey provides no information about the future. To produce a forecast, we need to calculate special statistical models, much like meteorologists do. gfs.bern only offers this service in within the scope of longitudinal studies. This means that the same survey must be repeated at least once during a specific period.




Are surveys predictions, too?

In general, the decision to publish or withhold the results of a representative sample is the onus of the client. If they decide to publish, all results must be included. The industry association VSMS, which unites Switzerland’s main survey institutes, also demands that its members publish the technical details about their surveys. Opinion researchers are obliged to disclose who their clients are and who is responsible for each survey. They must also publish details about the sample, state when the survey was carried out and provide information about the scientific limtits and interpretation of the survey. The gfs.bern research institute provides those technical details about all its published surveys in the form of short or final reports in the ‘Publications’ section of its own website.


Do surveys influence referendums or elections?

Researchers are intensely exploring the potential influence of surveys on elections and referendums. The following influences have been proposed:

  • Mobilisation effects: surveys could boost voter turnouts, especially if the outcome of the referendum or election in question is uncertain.
  • Defeatism effect: surveys might lower participation among voters whose side is expected to lose.
  • Lethargy effect: surveys might lower participation among voters whose side is anticipated to win.
  • Laziness effect: surveys might lower participation among undecided voters.
  • Bandwagon effect: surveys might sway public opinion in favour of the predicted winner.
  • Underdog effect: surveys might sway public opinion in favour of the predicted loser.

Many studies have been conducted on those theoretical effects worldwide. So far there is no evidence of any of these proposed effects.

This proves that surveys do NOT influence the outcome of referendums or elections. But they may influence the decisions of active stakeholders, for example their campaign measures or their communications strategy. In our experience, they also influence public expectations about the outcomes.

Participate now in our survey panel

We have made it our purpose is to conduct relevant social research to answer questions that matter to society. You can help us by participating on our online panel. Especially in Swiss politics, the people’s voices are of central importance. Your answers will help our policy makers understand public sentiment on new laws, for instance. If you join our online panel, we will regularly invite you to take part in surveys about politics, the economy and communication.