sampling may have introduced selection bias but no information on response rates was provided to assess this possibility.14 The interview questions assessed implementation in two different ways. In one, respondents were asked for their assessment of implementation. In the other, researchers deduced the degree of implementation from responses to questions about particular IC-related functions.
Nytro et al. (1998) conducted a cross-sectional survey of Norwegian firms to determine the change in prevalence of IC implementation, using data from the previously described study (Saksvik and Nytro, 1996) for comparison. Both studies used a randomized quota sample of private and public enterprises and computer assisted telephone interviewing. The Saksvik and Nytro (1996) study found that 66 per cent of firms had not begun implementation, 25 per cent reported they were in the process of implementation, and 8 per cent reported that IC had been implemented. Of the 66 per cent of firms that had not started implementation, 58 per cent had not heard of the regulation and of those most were small in size. The findings, 4 years after the regulation had come into effect, were that 19 per cent of firms had not yet begun implementation (47 per cent decline), 36 per cent were in the process of implementation (11 per cent increase), and 45 per cent had implemented IC (37 per cent increase) (Nytro et al., 1998). 15 16
The study had a number of design characteristics affecting the internal and external validity of the implementation findings. Randomized quota sampling, based on firm size, industry, and geographic location, was used to obtain the final sample in both cross-sectional surveys (sample sizes were 2092 in 1993 and 1182 in 1996) in order to ensure that the final sample was representative of all Norwegian firms. The sizes of the firms in the 1996 survey were reported as: 43 per cent with less than 10 employees, 39 per cent less than 100, and 18 per cent had over 100. Refusal rates for the computer assisted telephone interviews were not provided and so the possibility of selection bias cannot be ruled out.
As the Nytro et al. (1998) investigation did not have the study of implementation as its primary objective, the published article provided little detail about the items in the interview that collected information on the
14 It should be noted that in a more recent investigation completed by the same research team using similar sampling methods, their response rate was 45 per cent (Saksvik et al., 2003).
15 The study design for the implementation evidence was classified as before after based on their use of data from the previously conducted survey.
16 A more recent study by the authors (Saksvik et al., 2003), which was not received in time for full review, provides a further update of the implementation rates in Norwegian firms. The survey was conducted in 1999 and findings were: 14 per cent of firms had not yet begun implementation (5% decline from 1996), 39 per cent were in the process of implementation (3% increase from 1996), and 47 per cent had implemented IC (2% increase).
Effectiveness of Occupational Health & Safety Management Systems: A Systematic Review