This is archive content that is no longer updated.

Go to the up-to-date statistics page.

Implementation of Continuing Vocational Training Survey 1999

The Continuing Vocational Training Survey (CVTS1) was conducted for the first time in the then 12 Member States of the European Union in 1994. The second survey, concerning vocational training in 1999, was conducted in the year 2000 in all the present 15 Member States of the European Union, as well as in Norway and in the 9 countries having applied for membership in the European Union.

According to Statistics Finland's employment statistics, private sector enterprises employed around 1,305,000 wage and salary earners in 1999. Enterprises with fewer than 10 employees and those operating in agriculture and forestry were excluded from the Continuing Vocational Training Survey. The target population of the survey covered three-quarters of the wage and salary earners in the private sector.

The public sector, that is, central and local government, were excluded from the survey altogether. The total number of wage and salary earners in the reference year was 1,927,000, of whom around 32 per cent were working in the public sector. Therefore, the survey covered approximately one half of the wage and salary earning population in Finland.

Harmonised EU questionnaire

The survey instruments were prepared in joint meetings of the Statistical Office of the European Communities, Eurostat, and the participant countries. The data were collected with a jointly adopted, harmonised inquiry questionnaire. Minor national variations were allowed in individual questions, but the main questions relating to the volume of training were the same in all the countries.

To support the harmonised questionnaire, Eurostat compiled a manual of the used concepts, definitions and classification to assist in adjusting the survey instruments into the national language and the local enterprise culture.

The data collection method

The survey instruments had been designed for mail inquiry-based data collection, which was used by most participant countries. The Commission additionally obliged the participants to conduct face-to-face interviews with a certain proportion of the sample to ensure that the obtained data would be of high quality and to improve the response.

Collection of the data was implemented in multiple stages in Finland. The interviewers first contacted the largest enterprises, which made more than one half of the sample, in order to obtain details of the persons who were responsible for personnel training at them and to whom the mail inquiry could be addressed.

All the enterprises drawn to the sample were sent the inquiry questionnaire with a request to return it. After this, interviewers contacted enterprises that had not responded to the mail inquiry. The interviewers endeavoured to motivate the respondents to reply to the inquiry, conducted telephone interviews or made appointments for personal visits to the enterprises. As a result of this, the response rate of the survey rose considerably. The interviewing stage was important in motivating the respondents, but to succeed it required from the interviewers good knowledge of the training activity of enterprises and its measurement.

During the checking of the returned questionnaires, the statisticians involved contacted the respondents frequently in order to clarify the supplied information or to supplement information lacking from the questionnaires.

The sample

Eurostat had set clear criteria for the drawing of the sample in order to optimise the representativeness and reliability of the obtained data. In Finland, the sample was divided into 60 strata, 3 size classes times 20 industry classes, by the size class and activity of the enterprises.

Because of the small total number and the industrial structure of the enterprises, the stratification had to be done at a somewhat rougher level in Finland than was required by the EU recommendation. In spite of this, almost one half (48%, the largest enterprises) became selected to the sample with probability one.

Response rate and unit non-response

The response rate in the Finnish data was 57 per cent. This can be regarded as a good result because the questionnaire was time consuming and demanding. Nowhere near all the enterprises had the data available at the accuracy required by the questionnaire and the respondents had to work hard to gather together the needed information. The Finnish response rate was also quite good when compared to those of the other EU countries. The east European candidate countries achieved clearly higher rates, but this was generally supported by the fact that responding there was compulsory to the enterprises.

The number of employees in the responded enterprises (365,300) covered 56 per cent of the total number of employees in the sampled enterprises (649,900). This was roughly equal to the proportion of responded enterprises of all sampled enterprises.

Response rate by activity and size class in Finnish CVTS2 data

Field of activity
Size class  
10 - 49    50 - 249 250 +     Total      
% % % % Number
Food industry 68.4 63.8 45.5 62.8 86
Wood and paper industry 61.4 58.4 56.5 59.3 233
Metal industry 59.9 61.7 51.4 59.2 350
Construction 56.3 60.7 77.8 61.3 106
Other industries 71.0 62.6 52.7 64.4 241
Trade 56.4 45.6 47.6 49.5 235
Hotels and restaurants 43.8 50.0 46.2 46.8 52
Transport 54.3 49.7 48.7 51.1 140
Financing and insurance 63.9 56.4 64.7 61.1 88
Other services 53.1 54.2 56.7 54.5 177
Sector Industry 63.6 61.2 54.1 60.9 1016
Services 55.4 50.2 51.7 52.1 692
Total   60.2 56.1 53.0 57.0 1,708
  Number of enterprises 606 849 253 1,708  

The fluctuations in the response rate are relatively minor on the whole, both between activities and size classes. Small enterprises tended to respond more readily than large ones. The reason for this is obvious - small enterprises provide less training so responding to the inquiry was easier for them. Examined by activity, the response rate was generally higher in industrial branches than in service branches. It may be that industrial branches have a longer tradition and more established practices in keeping records about training.

In the reporting of the results, the response data have been raised to the frame population level with weighting coefficients. The weighting coefficients have been calculated for the target units in cells complying with the 60-class stratification used in the drawing of the sample.

Item non-response

It is difficult to describe the quality of survey data unambiguously. The volume of missing data could be used as one measure. Allowing for the difficulty of the questionnaire, there was very little item non-response in respect of the qualitative variables. The rate varied from zero to just short of 10 per cent. As could be expected, the provision of quantitative data proved more difficult. It was easier to give totals concerning training, such as numbers of training days or total training costs, than to give breakdowns of the days by gender or of the total costs by type of cost.

Item non-response in respect of the main quantitative variables varied from 7 to 18 per cent depending on the variable. According to the internationally agreed guidelines, however, deficiencies in quantitative data were not allowed. Individual items of missing data were substituted in accordance with commonly agreed principles. In Finland the substitution was implemented on the basis of the used stratification according to size class and activity. A missing item of quantitative data for an enterprise was added by using the average of the 60 sub-strata divided into 20 activities and three size classes. The method did not affect the distributions in the original data. After this, the parameters concerning quantities could be raised to the frame population level using single weighting coefficients, instead of variable-specific ones.

Last updated 31.5.2004

Referencing instructions:

Official Statistics of Finland (OSF): CVTS, Continuing vocational training [e-publication].
ISSN=1798-0003. 1999, Implementation of Continuing Vocational Training Survey 1999 . Helsinki: Statistics Finland [referred: 7.7.2022].
Access method: