2 Data and methods

The data used in the empirical analysis generating the model is a panel of Danish firms, which has been surveyed in five rounds from 1996 up to 2010. Denmark’s Statistics was in charge of the data collection from the start. The first four rounds were part of the DISKO data collection, aimed at collecting representative information on product- and service innovations, organizational change and demands to employee on learning, competence development and training in firms from the private urban sector. The 1996 DISKO survey resulted in information from 1990 firms. The next DISKO survey in 2001 was a matched survey design collecting data from both employers and employee representatives.

Beside the questions on innovation, organizational changes and competence developments, this survey collected information on employer-employee cooperation and employee participation in change decisions. The result of this survey was 2007 employer responses and 473 employee responses. In order to collect information on innovation strategies a third supplementary survey round were launched in 2004.

The fourth round of the DISKO surveys was completed in 2006 on basis of 1552 still economic active firms in the panel. These ‘core’ firms were supplemented by a sample in order to avoid bias and ensure the research sample to be representative. Denmark’s Statistics data collection resulted in 1775 responses from employer representatives. In 2010 there were 1430 of these firms verified as still active and they constituted the GOPA2 panel sample. The data collection resulted in a research panel of 601 firms, which represented a response rate of 39.6%.

This is not a very satisfactory response rate, but the attrition analysis broken down on sector and size indicates no unacceptable bias in the research panel. The research strategy used in the construction of the model is sequential descriptive. The theoretical foundation of the elements in the model is discussed and documented empirically mainly by the 20062010 panel data. However, the empirical documentation of the theoretical dimensions sometimes includes data from the first 1996 DISKO round up to the 2010 GOPA round. A scale of learning organization (LO) which has been verified in prior research (Nielsen 2004, Nielsen & Lundvall 2006) has been used to test the relation between innovation performance and learning organization.