Technical details
This page briefly introduces the technical details of the OeNB Euro Survey. It comprises notes concerning sampling, the questionnaire, the design and translation of the questionnaire and information about units of observation, fieldwork and interview techniques, weighting, nonresponse and underreporting, and the plausibility of data.
Sampling
About 1,000 persons per country are interviewed during each survey wave. Interviewees comprise citizens of the respective countries aged 17 years and older. In survey waves prior to 2016 also younger respondents were sampled with some variation in the cut-off age by country and wave. The OeNB Euro Survey is conducted as stand-alone survey or as a part of national omnibus surveys; the sampling procedures differ across countries.
Samples are selected via a multistage stratified random sample procedure, with the exception of Bulgaria and of Bosnia and Herzegovina, where a variant of quota sampling is applied. Samples are representative of the population structure in all countries with respect to age, gender and region. Poland represents an exception, in that from fall 2007 to spring 2012, the population of only the ten largest cities was sampled. Since fall 2012, the surveys have covered Poland as a whole.
For countries that apply a variant of a multistage stratified random sample procedure, the number of strata varies from two to six. In each of these countries, a number of sampling points is drawn with probability proportional to population size and population density. The sampling points represent the whole territory of the countries surveyed according to the Eurostat NUTS II (Nomenclature of Territorial Units for Statistics II, or equivalent) and according to the distribution of the respective nationalities within the resident population in terms of metropolitan, urban and rural areas. In each of the selected starting points, a starting address is drawn at random. Further addresses are selected using standard “random route” procedures from the initial address. Finally, the selection of the interviewee in each household is randomized by standard procedures.
Design and translation of questionnaires
Questionnaires are designed by the OeNB in English and are translated by professional translators in the respective countries in close collaboration with the project team. As a quality check, translated questionnaires are controlled by the Austrian Gallup Institute as well as by Vienna-based native speakers and translators in order to minimize connotation differences.
The standard questions query respondents’ cash holdings in foreign currency, savings at banks, portfolio composition, actual and planned loans as well as economic sentiments, expectations, and trust in currencies and institutions. In some aspects, these standard questions continue earlier surveys commissioned by the OeNB. Over the years, several special topics have supplemented the questionnaires.
Unit of observation
The unit of observation is an individual as opposed to a household. Consequently, the questionnaire addresses personal asset holdings. This approach might constitute a problem wherever it is difficult to distinguish between personal and household holdings (e.g. in the case of a couple with joint holdings). The questionnaire accounts for this issue by asking whether the interviewed person owns assets personally or jointly with a partner.
The OeNB Euro Survey does not constitute a panel at the level of the unit of observation.
Fieldwork and interview technique
The Austrian Gallup Institute conducts the OeNB Euro Survey on behalf of the OeNB. While the Austrian Gallup Institutes coordinates the survey, the actual fieldwork is executed by network partner institutes in the respective countries. From 2007 to 2014, surveys were conducted twice a year, in April/May and in October/November. In 2015, the survey frequency was reduced to once a year (autumn). The first survey wave was in fall 2007.
All interviews are held face-to-face at people’s homes and in the respective national languages. For data capture, CAPI (Computer Assisted Personal Interview) is used in those countries where this technique is available; otherwise, PAPI (Paper and Pencil Interview) is used.
Weighting
For all countries surveyed, a national weighting procedure (using marginal and intercellular weighting) is carried out based on the universe description, accounting at least for gender, age and geographical variables. The universe description is derived from the respective national statistical offices.
Nonresponse and underreporting
The number of refused interviews amounts to some 600 to 700 persons on average per country. Considering the sensitive nature of some questions, the number of interrupted interviews is very low (1% to 2%). The response rate – defined as number of completed interviews over the number of effective contacts – averages out to 60%.
Questions requiring respondents to divulge sensitive information, e.g. questions on the amount of euro cash holdings or net household income, do suffer from underreporting and item nonresponse. The share of “don’t know and “no answer” replies averages 17% for the amount of euro cash holdings and 19% for the net household income, respectively.
Plausibility of the data
In general, the internal plausibility checks show that the vast majority of respondents answer in a consistent manner, such that the opinions expressed by and large correspond with data on people’s actual behavior. Yet it seems that respondents find it rather difficult to answer questions about their expectations.
Comparison with results from other surveys suggests that the OeNB Euro Survey delivers accurate results. In particular, comparisons have been made with data from the Eurobarometer and the Consumer Confidence Survey (both European Commission), the International Household Survey Network (World Bank), the Review of the International Role of the Euro (European Central Bank), and the Labour Force Survey (Organisation for Economic Co-operation and Development).
For further details on the OeNB Euro Survey, see Dvorsky, Scheiber and Stix (2008).