Open Access Repository

Community Innovation Survey. Improving the quality of CIS data for innovation expenditures and turnover shares


Downloads per month over past year

Arundel, A and van Cruysen, A 2010 , 'Community Innovation Survey. Improving the quality of CIS data for innovation expenditures and turnover shares', paper presented at the Working Group Meeting on Statistics on Science, Technology and Innovation, 22-23 November 2010, Luxemburg.

[img] PDF
Improving_the_q...pdf | Request a copy
Full text restricted
Available under University of Tasmania Standard License.


Since the first CIS survey, there have been concerns over the quality of the CIS results
for the amount that enterprises spend on different innovative activities (Question 5.2
in the standard CIS 2010 questionnaire) and in the share of sales from new to market
and new to firm product innovations (question 2.3). For the first three CIS surveys
these concerns were partly driven by low item response rates for these questions,
particularly for innovation expenditures, and the effect of these low response rates on
the representativeness of the results. These concerns have decreased over time with
notable improvements in item response rates, from approximately 50% for CIS-1 to
80% for CIS-4.
However, concerns over the accuracy and comparability (across sectors within a
country or across countries) of the answers to these questions remain both among
National Statistical Offices (NSOs) that implement the CIS surveys and data users.
Several NSOs have reported that the total R&D expenditures reported in the CIS
survey are greater or less than reported R&D expenditures in the national R&D
survey. Users have noted that the data for the share of product sales from new-tomarket
innovations are not comparable across countries, with very high shares in
previous CIS surveys in Italy and in some of the new member states and low shares in
highly innovative countries such as Denmark. Although the cause is partly known –
what is new to the market for a transitional economy such as Romania is very
different from what is new to the market in an advanced and sophisticated market
such as Denmark, the goal of policy users is to obtain results that are comparable
across countries.
In response to these concerns, Eurostat asked MERIT to examine possible changes to
the design of CIS questions 5.2 and 2.3 with the goal of improving the quality of the
responses. This report provides an interim report for this task, based on the results of a
survey of 32 European NSOs that conduct the CIS. The purpose of this interim report
is to elicit suggestions on how to improve these questions. An example is provided in
the German proposal (see the presentation by Christian Rammer). All suggestions will
be evaluated and a short-list of concrete proposals for a redesign of these two
questions will be included in the final report, with the goal of cognitively testing the
redesigned questions in several member states. Of note, it may not be possible to
solve all problems through a redesign of each question. Alternative for improving data
quality include checking results with respondents, comparing CIS data with other data
sources where possible, or adding additional questions which can either provide
missing information or a reference for respondents.

Item Type: Conference or Workshop Item (Paper)
Authors/Creators:Arundel, A and van Cruysen, A
Item Statistics: View statistics for this item

Actions (login required)

Item Control Page Item Control Page