Business Surveys and the Changing Data Environment– How Can We Know Best about Businesses? |
|
| Session Organisers | Mr Xabier Irastorza (European Agency for Safety and Health at Work (EU-OSHA)) Mr Gijs Van Houten (European Foundation for the Improvement of Living and Working Conditions (EUROFOUND)) |
| Time | Friday 19th July, 11:00 - 12:30 |
| Room | D17 |
The interest in business practices and outcomes is only increasing in the context of the changing data environment. There is a wide range of actors that need first-hand information about the status of the economy, even more so in the context of globalisation, which has modified the business and economic structures, with an impact on value adding chains and employment relations. New information technologies have contributed precisely to speeding up the pace of change but also enabled accessing respondents and information providers in a range of new ways. Bearing this in mind, and while there may be a considerable volume of business data on some specific topics and areas, on other policy fields with no available censuses, registers or mandatory surveys in place, much remains unknown.
Some features of businesses pose particular challenges when it comes to surveying them. The heterogeneity of the business population is one of them, with a great number of micro and small businesses on the one hand, and a small number of very large businesses on the other. The structure of the businesses (single vs multi-site, national vs multi-national etc.) is another aspect to take into account, as well as the availability of business registers, the quality of which is often poor, and inconsistent across countries, activity sectors or size classes. The definition and identification of the adequate respondent or respondents and ensuring their participation are clear challenges in business surveys, which, like other types of surveys, have witnessed a downward trend in response rates in recent years.
This session invites contributions that discuss the issue of response rates and response bias in non-compulsory national and international business surveys. To what extent can these issues be alleviated by considering various modes of contacting and questionnaire administration, and at what cost? What avenues are available for reducing respondent burden, and thereby possibly boosting response rates, by making increased use of administrative data? And what possibilities do administrative data offer for identifying and correcting for response bias? Also, what are the experiences in following-up with non-respondents?
The session aims to bring together researchers, users and practitioners in the area of business surveys, to share their experiences – good or bad – of dealing with these issues.
Keywords: business survey, survey mode, non-response, international comparability
Professor Robert Helmrich (BIBB) - Presenting Author
Mr Manuel Schandock (BIBB)
The effects of megatrends, such as demographic change, increasing mobility, globalization, technological development, climate change and scarcity of resources, to name but a few, have an impact on the needs of businesses for their current and future employees.
For the proper orientation of the VET system, information on changes in operational requirements is of particular importance. At the same time, the costs of company surveys are increasing and the returns from such surveys are decreasing dramatically.
As a research institution for vocational training in Germany, we therefore increasingly use data sources that are accessible to the public, which give us a picture of the relevant situation of the companies, give us selection criteria and facilitate access to companies for surveys.
With job advertisements from the Federal Employment Agency and online job exchanges, we receive a multitude of descriptions of requirements, activities and necessary skills and qualifications. We can take these up on a case-by-case basis and supplement them with information from company surveys. In addition, we can add regional and industry-specific features to the operational features in advance of the surveys. In this way, we increase the willingness of companies to participate in our surveys, reduce the effort required for the survey and increase the density of information.
The presentation shows which information job advertisements offer, how the content and information density of the job advertisements can be used and which imputations of structural regional and national data are possible. Selected examples will be used to illustrate how this information and data can be obtained and how this information can be linked to company surveys. The methodological advantages and disadvantages of these enterprise surveys are also described.
Mr Carsten Broich (Sample Solutions BV) - Presenting Author
Mr Nikola Jovanovski (Sample Solutions BV)
Mr Jamie Burnett (Kantar Public)
European establishment surveys remain largely conducted via CATI surveys even though more and more mixed-mode approaches become feasible as well. The basis of all B2B studies which are probability based remains an accurate sampling frame representing the business population within a country and additionally provides contact options (either mail, phone or email). While there are several commercial databases available with high coverage levels, the issue remain non-coverage of specific industries or freelancers, lack of in-depth contact options (eg. phone number missing) and lastly lack of essential parameters to identify sizes or specific industries. This is partly due also to wrong categorization or conversion of local industry codes to NACE Rev.2 codes.
During the last decades , the internet has become the largest source for B2B information however in a unstructured format. This vast resource of unstructured data (Big Data) B2B data can be found from a companies website, open business directories or social media profiles together with information such as employee size, growth rates, relevant industries and contact details. This paper analyzes current ways that have been implemented to overcome issues in European establishment studies with regards to coverage, accuracy but also different contact modes and further suggests to a roadmap for an EU wide business establishment sampling frame and also attempts to evaluate the limitations of Big Data with regard to coverage-error and bias.
Mr Andrew Cleary (Ipsos) - Presenting Author
Ms Ahu Alanya (Ipsos)
Ms Femke Dekeulenaer (Ipsos)
Mr Gijs van Houten (Eurofound)
Mr Giovanni Russo (Cedefop)
The European Company Survey 2019 is a pan-European survey of establishments. The survey uses a push to web survey design, with telephone initial contact and short CATI screener questionnaire, and web interviewing for the full substantive sample. The survey is undertaken by Eurofound, with this 4th edition taking place in partnership with Cedefop and fieldwork delivery commissioned to Ipsos. The survey informant is the manager responsible for human resources in a randomly sampled establishment and when possible an employee representative. Main survey fieldwork will take place in the first half of 2019, with over 20,000 interviews across 28 European countries, following a full pilot late 2018.
Cross-national surveys of establishments face several challenges when it comes to sampling. First and foremost, registers of establishments – the local units of companies, rather than companies as a whole – are not available in all countries. This implies that a company level register needs to be used, followed by a within company screener to select one or more eligible establishments. Secondly, there is considerable variability in the quality of the available registers – be they on the establishment or company level.
In our presentation we will discuss the efforts that have been made in ECS 2019 to overcome these challenges, which include the use of frame data (e.g. number of establishments) to improve sampling in countries with a company level register and applying phone number matching based on lists acquired from different sample providers. We will also discuss the approach that was used to screen for establishments within companies, assessing the impact on the distribution of single-establishment companies, and headquarters and subsidiaries of multi-site companies.
Miss Catherine Grant (-) - Presenting Author
With increasing numbers of surveys moving online there appears be less of a role for paper self-completion surveys, particularly for surveys of businesses. Common concerns about levels of online access tend not to apply to the same extent to business populations so why employ a paper survey? This paper looks at the impact using a paper questionnaire as part of a reminder strategy can have in improving response rates for business surveys. Most business surveys aim to collect the organisational views or information about an organisation rather than a personal opinion. Respondents may be required to collate information from multiple sources or to discuss their response with other members of the business. A paper questionnaire can offer two main advantages, it can be easier for the respondent to share and collate information and it can serve as a physical reminder to complete the survey.
Working with the Financial Conduct Authority we conducted an experiment on the annual survey of regulated firms to test the impact of issuing a paper questionnaire as part of the reminder strategy to improve response. This experiment built on previous experiments on this survey which have shown that hard copy letter invites have a positive impact on response.
Mr Xabier Irastorza (European Agency for Safety and Health at Work (EU-OSHA))
Mr Arnold Riedmann (Kantar Public) - Presenting Author
Fieldwork for the third wave of the European Survey of Enterprises on New and Emerging Risks (ESENER) is to be carried out in spring 2019. Building on the experience of the first two waves, in 2009 and 2014, ESENER focuses on the management of occupational safety and health (OSH) risks in practice. In the last wave almost 50,000 establishments across all business size classes and activity sectors were surveyed in 36 European countries, the target respondent being ‘the person who knows best about health and safety in the establishment’.
As is the case for other voluntary CATI establishment surveys, ESENER suffers from low response rates (particularly in some countries) and a possible selection bias, which may put into question some of the findings. In order to tackle the non-response bias, a more rigorous and improved approach to sampling has been envisaged for ESENER-3, starting from the sample design, the business registers used and the weighting to be adopted, as well as the contact strategy to engage respondents.
Among the increased efforts to minimise non-response bias, ESENER-3 is trying to gather information about the non-respondents in order to see whether they differ from respondents in a systematic way. Two short questions are planned for non-respondents: (1) the establishment size (due to quality issues with some business registers) and (2) their approach to OSH management by providing three statements on preventive culture. The same questions are to be asked to respondents in order to contribute to an analysis of the non-response bias.
These questions will be tested at the pilot in February 2019 (30-40 interviews in 33 countries) and it is the findings from the pilot that will be presented in the conference, as well as the final decision taken to implement them (and how) in the main survey.