The Monitoring

logo_pusteblume

Method

To deliver reliable data and find acceptance on the part of the target groups, the Tobacco Monitoring Switzerland has to meet high methodological requirements with regard to representativeness, sampling method, and sample size.
The following sections provide information on the...:


 

Survey Population

The population is all persons residing in Switzerland 14-65 years of age that have a private telephone landline listed in the register of the company AZ Direct AG and that can be interviewed in German, French, or Italian. The age limits in this survey differ from those in the Swiss Health Survey (SGB), which examines persons residing in Switzerland 15 years of age and older (with no upper age limit). The difference in the age limits is viewed as appropriate, in order that the initiation phase of smoking can be captured and also so that the bases for (health) policy measures for working-age people can be delivered. The lower age limit also takes into account ethical standards for interviews and surveys (ESOMAR World Research Codes and Guidelines) (see also Krebs & Demarmels, 1998, p. 43), whereby permission of a parent must be obtained before a child under the age of 14 is approached for an interview. Another methodological difference between this survey and the Swiss Health Survey is that in the Tobacco Monitoring Switzerland, also two persons in the same household may be interviewed.

Periodicity of the survey waves

The survey waves are conducted four times a year; per quarter a new sample of 2,500 persons is surveyed. Very closely related questions have to be asked in the same survey wave or, if data cumulation is required, in successive survey waves. Since survey wave 18, the waves have included a permanent add-on module (like the basic module, four times a year) containing questions on protections for non-smokers and on legal measures, in complement to the add-on modules ‘passive smoking’ and ‘policies / laws.’ The other add-on modules are a part of the survey waves at an interval of approximately two years.

Sampling procedure

In the Tobacco Monitoring Switzerland sampling is done using a two-phase random-random sampling method. In a first step, households having a landline telephone listed in an electronic directory (AZ Direct AG) are randomly selected and contacted by telephone. After recording the structure of the household, a target person among all 14- to 65-year-old household members is randomly chosen and interviewed by telephone, either immediately or on an agreed-upon later date. In order to have for the statistical analyses a sufficient number of persons in some characteristic groups that are small but important for tobacco prevention (young people, pregnant women and mothers of small children), 14-24 year old men and women 14-44 years of age are oversampled. To do this, a second person in a multi-person household is then interviewed, provided that someone in the household fulfills the demographic criteria.
Residents of the French- and Italian-speaking regions of Switzerland are also overrepresented in the sample in order to ensure a sufficiently large sample for each language region in Switzerland.

As to representativeness, certain limitations should be mentioned:
  • Persons can be interviewed only if the interview can be conducted in German, French, or Italian. 
  • Only households having a landline telephone and a telephone number listed in the register of the company AZ Direct AG can be included in the sample. The greater part of the population is still able to be reached by fixed landline connections. However, this percentage is decreasing from year to year, as mobile phones are increasingly replacing landline telephones (-2.8% from 2004 to 2005). At the end of 2005, there were approximately 50 landline telephones per 100 residents (Brambilla, 2007; Bundesamt für Statistik, 2007). According to the feasibility study (Krebs & Demarmels, 1998) in 1997 about 5% of households were reachable only by mobile phone. According to an estimate by the LINK Institute this (number) had already increased to 7% in 2000 (WEMF AG, 2002), and in 2002 the Schweizerische Gesellschaft für praktische Sozialforschung (GfS) (Swiss Association for Applied Social Research) found 9 to 12% of households to be reachable only by mobile phone (Spichiger-Carlsson, 2002). There are no up-to-date estimates available for Switzerland. Assuming that the trend has continued, we can reckon today with 12 to 15% of households accessible only by mobile phone. This poor accessibility is the case mainly for younger persons no longer living at home with their parents. In Germany in the year 2004, approximately 25% of under 25-year-olds were reachable only by mobile phone (Statistisches Bundesamt Wiesbaden, 2004).
  • Since January 1, 1998, it is no longer compulsory for fixed landline telephone customers to list their telephone numbers in the telephone book. For this reason, it is no longer the case that all telephone numbers are listed in the telephone directory. According to the estimate by the LINK Institute in 2000, 3% of households with fixed landline connections no longer listed their numbers (WEMF AG, 2002). No more recent data is available at the present time. But it is reasonable to assume that that percentage has increased.
  • ‚Marginal’ persons can not be reached.

The sample size for the basic module was set at 2,500 telephone interviews per survey wave. In some of the add-on modules, in-depth questions are asked of certain groups of persons. The sample is stratified by language region. Of the 2,500 interviews conducted per quarter, on average 1,426 interviews are conducted in the German-speaking, 711 in the French-speaking, and 363 in the Italian-speaking part of Switzerland. Motivation to participate in the survey is optimized by informing about 75% of the selected households about the planned interview via a written announcement. About 25% of the households are listed in the telephone directory marked with an asterisk, which means that they do not wish to receive advertising material; because also scientific studies fall into the category of advertising material, no announcements may be sent to these households. 

Data collection

The data are collected using computer-assisted telephone interviewing (CATI), performed by the telephone laboratories of the LINK Institute in Zurich und Lausanne. The interviews are conducted by trained interviewers who are monitored by supervisors. The interviews are conducted in German, French, and Italian.

Weighting and analysis methods

The data collected in the CATI laboratory are first checked by the LINK Institute for plausibility; after follows the weighting of the data. The weighting corrects distortions in the sample, so that it correctly represents the population.
The weighting of the data is performed in four steps (Raemy & Grau, 2002):

Oversampling weighting
Persons having the oversampling criteria mentioned in section 3.3 above have a greater chance of entering into the sample. For this reason, in a first step, the distortions caused by the oversampling are corrected.

Household transformation weighting
The household transformation weighting serves to correct the distortion caused by the selection procedure. Here the weighting factors are calculated based on the composition of the households interviewed. 

Regional weighting
To be able to draw meaningful comparisons between the German-, French-, and Italian-speaking parts of the country, the sample in any one part must not be too small.  For this reason the sample was drawn disproportionately – that is, French-speaking and Italian-speaking Switzerland are disproportionately (over) represented. Afterwards, the disproportionate set is corrected such that the distribution of the respondents from the German-, French-, and Italian-speaking parts of Switzerland accords with the distribution in the population (72% German-speaking Switzerland, 23.5% French-speaking Switzerland, 4.5% Italian-speaking Switzerland).

Sociodemographic weighting
The sociodemographic weighting is a cell weighting by age and sex. In this step the sample is adjusted to the distribution of age and sex in the population. The weighting factors are based on the updating of the Annual Population Statistics (ESPOP) provided by the Swiss Federal Statistical Office. 

After that, LINK Institute provides a basic evaluation and a technical report on the main data collection. It also delivers a data set that has been made anonymous and prepared, as an SPSS file. After an additional plausibility check by the Social and Health Psychology unit of the Department of Psychology at the University of Zurich, the reports on the results are drawn up.


1 Krebs, H. & Demarmels, B. (1998). Indikatoren für den Tabakkonsum in der Schweiz. Machbarkeitsstudie im Auftrag des Bundesamtes für Gesundheit.

2Brambilla, M. (2007). Amtliche Fernmeldestatistik 2005. Datenerfassung bei den Fernmeldedienstanbietern. Biel: Bundesamt für Kommunikation.

3 Bundesamt für Statistik (2007). Indikatoren Telefoninfrastruktur [On-line]. Available: http://www.bfs.admin.ch/bfs/portal/de/index/themen/systemes_d_indicateurs/
indicateurs_de_la/introduction.html

4 WEMF AG (2002). Report: Blick hinter die Forschungskulissen [Online]. Available: http://www.wemf.ch/d/rep_august2002/report.shtml.

5 Spichiger-Carlsson, P. (2002). Handies: Repräsentativität von Umfragen gewährleistet. Newsletter Schweizerische Gesellschaft für praktische Sozialforschung (GfS), 1, 2.

6 Statistisches Bundesamt Wiesbaden (2004). Fakten und Trends. Deutschland aktuell – Ausgabe 2004. Wiesbaden: Statistisches Bundesamt.

7 Raemy, N. & Grau, P. (Oktober 2001). Studienbeschrieb Tabakmonitoring (1. und 2. Welle). Luzern: LINK Institut.

 

 

Tobacco Monitoring Switzerland. Development and implementation on behalf of the Swiss Federal Office of Public Health. Funded by the Tobacco Prevention Fund.
Copyright © 2009 Tobacco Monitoring Switzerland. All rights reserved.