2_Editorial_24(3)
10/6/08
5:27 PM
Page 179
editorial Safety culture: snapshot of a developing concept by Ian Glendon
The evidence that safety culture and safety climate are rapidly developing concepts within a number of discipline areas is substantial. How far this development has spread and how fast it has occurred are among the issues addressed in this article, which also considers the relative impact of published articles in this field and contributions by researchers from different countries. Two of the articles in this special issue of the journal contribute to the literature by adding to the relatively small number of empirical qualitative studies published to date. Another contributes to the nascent group of studies that adopt methodological triangulation, while a fourth is a rare example of a longitudinal study of smaller organisations. The final contribution is an extended review of mainly recent literature, which seeks to present the current state of knowledge in the field within a structured framework.
Searching relevant literature Ian Glendon is Associate Professor and a member of the Social and Organisational Psychology Research Unit within the Behavioural Basis of Health Program in the School of Psychology, Griffith University, and also a member of Griffith’s Work, Organisation and Wellbeing Research Centre.
An immediate problem when considering the literature on safety culture and safety climate is that much of it is in the form of reports to sponsoring bodies and other published sources that may not be readily accessed through web-based searches. Therefore, while reports and other documents can be useful when disseminating applied research to a wider audience, when reviewing the literature on these topics, only articles published in refereed journals were sampled. Exclusions included
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
179
2_Editorial_24(3)
10/6/08
5:27 PM
Page 180
Editorial
books, book chapters, dissertations, conference papers, conference proceedings, letters, reports to sponsors, and articles in professional and nonrefereed journals. The justification for excluding at least some of these sources is that a degree of duplication exists, for example, between conference papers and published articles. A further restriction was that only English language articles were considered for this review, although a few articles in other languages were located. All domain-relevant English language published and “in press” articles at 31 January 2008 were included in the review. Extensive web-based searches were conducted, and all articles in the safety culture/climate domain that were referenced within articles retrieved were also identified. This search identified 256 empirical and review articles that might meet the domain criteria. Further inspection resulted in 53 of these being excluded from the final sample — 28 on the grounds that they were published in a professional or non-refereed journal, and 25 because a prime focus was neither safety culture nor safety climate. Related topics included high-reliability organisations, high-performance workplace systems, behaviour modification, attitudes to safety or accidents, risk perception, safety programs, organisational culture, organisational climate, and other aspects of organisational functioning. If an article focused on any of these topics, as well as safety culture or safety climate, then it was retained in the sample. The final sample comprised 203 refereed empirical or review articles where a prime focus was safety culture and/or safety climate. The distribution of these articles over the period of time since the concepts were introduced is shown in Figure 1 (all “in press” articles have been assigned to 2008). The highly skewed distribution reflects the rapid rise in the number of articles on these topics published over a 28-year period. Spikes in the distribution reflect the publication of journal special issues on safety culture/climate (that is Work & Stress in 1998, and Safety Science in 2000 and 20071-3), with the 2008 figure including the 2008 special issues of Safety Science and this issue of the Journal of Occupational Health and Safety — Australia and
180
New Zealand.4,5 The distribution also highlights the time delay between Zohar’s initial 1980 landmark article, Brown and Holmes’ follow-up article in 1986, and the five-year gap preceding the eventual “take-off” in the field.6,7
Discipline areas contributing to the safety culture/climate literature Of interest is the extent to which the related concepts of safety culture and safety climate have penetrated various disciplinary fields. The 54 journals in which the 203 articles appeared were classified into four broad disciplinary fields and a number of subsidiary fields. Table 1 shows the number of articles in each disciplinary field. Over 55% of the articles were published in a safetyrelated journal, and a great majority of these appeared in four journals: Safety Science (50); Journal of Safety Research (20); Accident Analysis and Prevention (10); and Journal of Occupational Health and Safety — Australia and New Zealand (10). Just over 20% of the articles were published in psychology journals, with the greatest numbers being in Work & Stress (14) and the Journal of Applied Psychology (10). In the third most represented field of health, only one journal reached double figures (Quality and Safety in Health Care (12)), while the 23 management articles were spread among a variety of journals. Of the 54 journals in which safety culture/climate articles were found, 31 had only a single article in this domain and, apart from the seven journals cited above, no other journal had published more than five safety culture/climate articles. It remains to be seen whether this high degree of concentration will continue or whether, for example, health and management journals, which tend to have published more recent articles, will publish more safety culture/climate articles in the future. It also remains to be seen whether the number and variety of journals publishing articles on these topics will continue to expand.
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
2_Editorial_24(3)
10/6/08
5:27 PM
Page 181
Editorial
FIGURE 1 Year of publication of 203 refereed articles on safety culture/climate 35
Number of articles
30
25
20
15
10
5
07 20
03 20
00 20
98 19
91 19
86 19
19
80
0
Year
Sectors studied The industrial sectors studied in the 203 articles are summarised in Table 2 which, as well as revealing that 49 were review articles, shows both a sector spread and a degree of concentration within certain sectors. The average of 1.7 industry sectors per empirical article reflects the fact that, in 18 of the 154 empirical articles, two or more industry sectors were studied. The largest industrial sector was manufacturing, which covered a fair range of subsectors, followed in equal numbers by studies involving health care and transport. Next were the petrochemical, construction and energy sectors. The tail comprised the public/utility, service, education, mining, primary industry, retail, R&D, and military sectors, with three studies not specifying an industry sector. Where the size of an organisation was mentioned, this was generally identified as large. However, while a few smaller organisations were studied, only two articles actually reported that small- and medium-sized enterprises were the prime target. In some cases, small organisations were part of industry-based studies. Respondent organisations were anonymous in all but seven studies.
Further information on multiple industries, sites, occupational groups and other details are summarised in Table 3 which shows that, of the empirical studies, over 43% drew data from more than one organisation or location, nearly 29% secured data from various occupational groups, and just over 9% collected data from multiple workgroups. Twenty-four per cent surveyed different levels at one or more locations (mainly shopfloor workers and managers/supervisory staff). Multi-sector studies (11.7%) were also in evidence, although, at just under 2%, cross-national studies were rare. A few studies analysed data across these multiple groups, so it is not meaningful to sum numbers in the final column of Table 3.
Determining the relative impact of safety culture/climate publications Assessing the impact of published articles might be done in various ways. Some journals publish impact factors, but many do not, and in some cases it is not always easy to find this information. Impact factors could be found for less than half of the 54 journals
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
181
2_Editorial_24(3)
10/6/08
5:27 PM
Page 182
Editorial
TABLE 1 Disciplinary origin of journals publishing 203 articles on safety culture/climate Broad disciplinary field Safety
Sub-disciplinary field Safety — general OHS and environment Ergonomics/human factors Safety/risk management
84 17 8 3 112
(41.4) (8.4) (3.9) (1.5) (55.2)
Psychology — general Organisational/personnel psychology Psychology — other
4 29 8 41
(2.0) (14.3) (3.9) (20.2)
Safety total Psychology
Psychology total Health
Health services Medicine
20 (9.9) 7 (3.4) 27 (13.3)
Management — general Management — other
9 (4.4) 14 (6.9) 23 (11.3)
Health total Management Management total
in this sample. As impact factors vary from year to year, when assessing the impact of a particular article, is it appropriate to use a journal’s current impact factor, the journal’s impact factor for the year in which the article was published (assuming that this can be determined), or some average figure? Impact factors have been criticised, inter alia, on the grounds that they disproportionately reflect citations of a relatively small percentage of articles published in a journal in a given time period, rather than being a representative measure of all articles published in that journal. To avoid the problem of determining and allocating an influence metric that is based on a general measure of journal impact, a search was undertaken to determine the number of citations that each article in the sample had received by a particular date.8 Discounting the “in press” articles (for which no citations would be expected), only 11 article searches returned a nil result. For the articles for which a result was obtained, this exercise revealed a highly skewed distribution in which just over a quarter of the sample had zero citations (n = 52), and around 50% had seven or fewer citations. At the other extreme, five articles had each been cited over 100 times.9-13 As would be expected, number of
182
Number (%) of articles
citations was highly correlated with years since publication (ρ = .86, p < .001).14 The correlation remained significant when the sample was reduced to 140 by removing “outliers” — in this case, all 2007, 2008 and “in press” articles, as well as Zohar’s 1980 article (ρ = .70, p < .001). An analysis of covariance (ANCOVA) was undertaken on this reduced sample to determine whether disciplinary field significantly influenced the number of citations of articles in the safety culture/climate domain, controlling for year of publication.15 Mean numbers of citations were: for psychology journals 42.26; for safety journals 26.04; for health journals 25.68; and for management journals 10.83. There was a significant effect (F(3) = 4.30, p = .006, partial η2 = .087, power = .86), with post hoc tests revealing that citation means for psychology, safety, and health journals did not differ significantly, but that all three were significantly higher than the citation mean for management journals (p < .05). While impact factors may be a general guide to a journal’s influence, their variation over time and the wide differential between disciplinary areas (for example, medicine and many other fields) means that they may be a problematic indicator of a
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
2_Editorial_24(3)
10/6/08
5:27 PM
Page 183
Editorial
TABLE 2 Sectors studied in 203 articles on safety culture/climate Sector
Sub-sector
Number of articles
Totals (%)
Review articles
General Health care Other specified sectors Meta-analyses
32 7 8 2
49 (18.1%)
Manufacturing
Not specified Metals Wood products Automotive products Others (specified)
18 8 4 3 16
49 (18.1%)
Health care
Not specified/general Hospital Nursing home Others
14 11 4 4
33 (12.2%)
Aviation Marine Road Rail Not specified
14 7 4 4 4
33 (12.2%)
Petrochemical
Offshore/oil Chemical (not specified) Petrol/oil refining Others/not specified
10 7 6 5
28 (10.4%)
Construction
All types
20
20 (7.4%)
Energy
Nuclear power Others
7 8
15 (5.6%)
Public sector
All categories
10
10 (3.7%)
Service
All categories
8
8 (3.0%)
Education
All categories
6
6 (2.2%)
All types
4
4 (1.5%)
Agriculture/grain/forestry
4
4 (1.5%)
Retail
All categories
4
4 (1.5%)
R&D
All types
2
2 (0.7%)
Military
All types
2
2 (0.7%)
One referred to as “high-risk”
3
3 (1.1%)
Transport
Mining Primary industry
Not specified Total
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
270
183
2_Editorial_24(3)
10/6/08
5:27 PM
Page 184
Editorial
TABLE 3 Multiple group studies Number of studies (% of empirical articles)
Multiple group descriptor > > > > > >
1 1 1 1 1 1
organisation/location/site within an industry sector occupational group within or across organisations/sites level within an organisation (eg worker/supervisor/manager) industry sector team/workgroup country
journal’s influence within a particular domain. However, to reflect the fact that in all disciplines a recognised “pecking order” of journals exists, some way of representing this structure would be desirable. Impact factors are often quoted to three decimal places, suggesting a spurious degree of accuracy for a field that is characterised by some uncertainty. Many universities use a tiered structure to represent the relative impact of journals within broad disciplinary domains. While not without its limitations, this approach recognises that the rank of any particular journal (for example, as represented by its impact factor) might vary over time within a fairly broad band. For the sample of journals in this study, an arbitrary three-tier ranking structure was adopted. Most of the “mainstream” middle-ranking journals (for example, Safety Science, Journal of Safety Research, Accident Analysis and Prevention) were assigned rank 2, the Journal of Applied Psychology and Academy of Management Journal were assigned rank 1, and journals for which no impact factor could be identified were assigned rank 3. In the analyses, these numbers were reversed so as to produce more interpretable statistics. The same sample of 140 articles was subjected to another ANCOVA, again with years since publication as the covariate and number of citations as the dependent variable. The grouped variable was journal tier, with the mean number of citations for high-ranked journals being 46.00, for mediumranked journals 31.12, and for low-ranked journals 17.47. The overall effect was significant (F(2) = 6.66, p = .002, partial η2 = .089, power = .91), with post hoc tests revealing that the citation mean for
184
67 44 37 18 14 3
(43.5%) (28.6%) (24.0%) (11.7%) (9.1%) (1.9%)
high-ranked journals was significantly higher than those for medium- and low-ranked journals, but that the means for medium- and low-ranked journals were not significantly different.
Article authorship One alleged consequence of the growing pressure (for example, from threatened or imposed research assessment exercises) on researchers to increase their published output is the increasing number of multiauthored articles. Table 4 shows author number frequencies for the articles in the present sample. While the modal authorship for the entire sample was one, 71% of the articles in the sample were multi-authored. The mean number of authors per article was 2.61 (SD = 1.56). To test the hypothesis that multi-authorship has increased over time, number of authors was correlated with years since publication, which resulted in a (ρ) value of –.154 (p = .014, 1-tailed) — indicating a significant trend towards an increasing number of authors per article over time. TABLE 4 Numbers of authors per article Number of authors 1 2 3 4 5 6 7 8 9 Total
Frequency 59 50 44 30 11 3 3 2 1 203
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
2_Editorial_24(3)
10/6/08
5:27 PM
Page 185
Editorial
To assess the possibility that articles in the health/medical domain could have particularly affected this result, an analysis of variance (ANOVA) first tested for differences between mean number of authors for the different journal types (F(3) = 6.66, p < .001, partial η2 = .176). While safety (mean n = 2.43), psychology (mean n = 2.24) and management (mean n = 2.22) journals did not differ significantly, all three differed significantly from health/medical journals (mean n = 4.26). However, re-running the correlation with the 176 cases that were not from the health/medical domain still resulted in a significant correlation with years since publication (ρ = –.157, p = .019, 1-tailed), confirming a general trend towards an increasing number of authors per article. An interesting question concerns the global distribution of article authorship on safety culture/climate. A designated country of origin was provided for all authors of the sample of 203 articles. Altogether, 25 countries were represented by the authorship criterion. That both China and India were represented in the list of countries of authorship helped the aggregate country of author origin to represent just over 51% of the world population. This is a crude measure of the extent to which the safety culture/climate concept has the potential to reach a global audience of interested researchers and practitioners. However, due to the language selection criterion for this sample of studies, the inevitable bias against countries in which English is not widely spoken is likely to be reflected in the country of authorship. While most articles were written by researchers from a single country, 24 (11.8%) were written by authors from different countries. For single-authored articles, and those where all authors were from the same country, an initial score of one was assigned to the article. In the case of articles where the authors’ designations were given as being from different countries, a score of 0.5 was assigned to the first author and 0.5 was split equally among all other authors.16 To reflect the journal tier in which the article appeared, each article’s points score was multiplied by the appropriate journal ranking. From this points allocation, an overall score for each
country of origin of the articles’ authors was derived, from which each country’s aggregate score could be calculated. Table 5 shows the points score for the 25 countries from which authors had contributed to the safety culture/climate literature, as defined in the present article. The points column in Table 5 is dominated by countries in which English is the prime language or in which English is widely used, which, given the bias in sampling for this article and the general distribution of scientific articles over a range of disciplinary areas, is unremarkable. The final column shows that, for this weighted sample, almost 50% of available points for this domain were assigned to researchers based in the United States and the United Kingdom.17 TABLE 5 Countries represented by authors with at least one refereed article on safety culture/climate (1980–2008) Country
Points
% of points
Australia Belgium Canada China Denmark Finland France Germany India Ireland Israel Japan Kuwait The Netherlands New Zealand Norway Poland Portugal Spain (South) Korea Sweden Switzerland Taiwan United Kingdom United States
33.25 0.84 14.00 8.75 4.00 5.00 2.00 1.00 1.00 4.00 25.00 0.34 0.50 14.16 3.07 13.50 1.00 1.00 10.34 0.67 8.50 6.00 11.68 84.24 91.20
9.64 0.24 4.06 2.54 1.16 1.45 0.58 0.29 0.29 1.16 7.25 0.10 0.15 4.10 0.89 3.91 0.29 0.29 3.00 0.19 2.46 1.74 3.39 24.42 25.43
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
185
2_Editorial_24(3)
10/6/08
5:27 PM
Page 186
Editorial
Another way of considering these data would be to weight them according to a criterion that reflected some relative national position. While numerous options might be available, many of these are subject to different reporting, recording and data analysis systems between countries.18 For this exercise, it seemed simplest to take into account the relative populations of the countries represented in this analysis. Estimates of the populations of all 25 countries were readily available from a web-based search and, in all but two instances, the July 2007 estimate was taken for each country and for the world population figure, from which percentages of the world population could be readily estimated for each country.19 From these calculations, an index obtained by dividing each country’s percentage of the world population by its percentage contribution to the safety culture/climate literature could be obtained. Table 6 shows the rank order of those countries for which this index exceeded unity. TABLE 6 Relative contributions to literature on safety culture/climate (1980–2008) by authors’ country of origin Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Country
Index*
Israel Norway Australia United Kingdom Ireland Finland Sweden The Netherlands Switzerland New Zealand Denmark Taiwan Canada United States Spain Kuwait Portugal Belgium
65.91 55.86 31.10 26.54 19.33 18.13 17.57 16.40 15.80 14.83 14.50 9.69 7.96 5.58 4.92 3.75 1.81 1.50
* Index derived by dividing a country’s points score by its percentage of the world population.
186
Taking into account a country’s population in order to derive a weighted “contribution index” reveals a different pattern. Countries with relatively small populations tend to score well using this criterion, while more populous countries tend to perform less well than when raw publication scores are used as the metric — the US is the prime case.20 The results of this analysis are also skewed by a small number of highly productive researchers in countries such as Israel and Norway, and by the journal tier in which some key articles have appeared.21 It has been claimed that a lack of funding in Australia for OHS research over the past 12 years has seriously hindered this country’s capacity to deliver in this field.22-25 While this is likely to be true across a broad spectrum of OHS research, on the evidence presented in Tables 5 and 6, the capacity of Australian researchers to contribute to the safety culture/climate field appears to remain strong. From the author’s personal knowledge, this capacity is represented by some level of research activity in around one-third of all Australian universities.26 Australian-based researchers have made contributions to the safety culture/climate literature in top-ranked journals; they have contributed to literature that is anchored in both the safety climate and safety culture fields, and have authored empirical quantitative and qualitative, as well as review, articles.
This special issue An objective for this special issue of the journal was to include international contributions alongside home-based articles to reveal something of the current state of research in the safety culture/climate field. Space limitations meant that only a small number of articles could be accepted and those published include articles from researchers who are based in the Netherlands and the UK, as well as from Australia and New Zealand. An extended review article complements the four empirical contributions, each characterised by different methodological and conceptual approaches.
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
2_Editorial_24(3)
10/6/08
5:27 PM
Page 187
Editorial
The special issue begins with two qualitative studies. These articles go some way towards redressing the gap in the literature, in which the majority of empirical studies to date have been quantitative. Walker’s study used an unobtrusive observation technique to record safety-related comments made by participants attending safety training sessions in two organisations. Bentley and Tappin’s small-scale study of a utility company involved interviews and focus groups as the main methodology. Coincidentally, these two qualitative studies resulted in 10 final safety culture categories deriving from initial lists of 18 categories. While Walker’s categories were exclusively grounded in the data collected, Bentley and Tappin’s methodology was based on the increasingly popular developmental model as a framework for considering safety culture. Despite these contrasting approaches, there is a fair amount of overlap between the final categories developed from the two studies. Walker’s “Managing safety” category is like a combination of Bentley and Tappin’s “Top management commitment” and “Health and safety management systems” aspects of safety culture. These two authors’ “Employee safety commitments, attitudes and behaviour” aspect of safety culture is similar to a combination of Walker’s “Employee behaviour” and “Individual responsibility” categories. Both studies include similar categories referring to health and safety training, health and safety communication, and reporting policies and procedures. These studies offer a useful framework for classifying organisations on the basis of safety culture categories emerging from their respective analyses. Guldenmund’s article also used the 18-category safety maturity framework (in this case, as one component of a methodologically triangulated approach to safety culture in a service organisation). His case study approach allows appreciation of a situation where different methodologies (questionnaires, structured focus group sessions, and interviews) might offer alternative diagnoses and suggest varied interventions. As the author observes, contrasting sentiments are accessed by the
different instruments, which offer distinct perspectives on the multi-faceted notion of safety. The article by Clarke and Flitcroft is a rare example in the safety climate literature of a longitudinal intervention study. It examines the vitally important issue of the impact of management on employee perceptions of safety. This article is unusual in that it studied small organisations in high injury-rate sectors. The authors examine the impact of an intervention designed to influence managers’ transformational leadership skills on employees’ safety perceptions. Their findings support the limited amount of other published research in this field, indicating that transformational leadership style has a positive impact on employees’ perceptions of safety climate. In the final review article of this special issue, Glendon summarises material from selected articles published since the start of 2006 (augmented with relevant material from earlier studies) to develop a model of safety culture based primarily on contemporaneous sources. The author’s model identifies safety culture as an explanatory variable that assumes a position of mutual influence with organisational culture. In this model, safety climate (perceptions of safety) is one of a number of mediating variables, which also include other cognitions, feelings, behaviours, and social interactions. Outcome variables, in addition to traditional OHS measures such as injury rates, include major disasters as examples of system risk outcomes and environmental hazards.
Further developments Evidence presented in this article points to the buoyant state of research in the safety culture/climate field, acknowledging that, at publication, the data will already be out-of-date. The analyses reveal a highly skewed distribution for citations of articles in this field, accounted for by a combination of time since publication, disciplinary field, and journal ranking. While mainstream safety journals continue to provide a home for a majority of articles in this field, a relatively small proportion
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
187
2_Editorial_24(3)
10/6/08
5:27 PM
Page 188
Editorial
of safety culture/climate articles have, to date, been published in high-ranking journals, indicating that safety culture/climate researchers could intensify their publishing endeavours by seeking to influence a wider audience. The analyses could be readily updated by adding new references as they appear, and could be extended by including other categories of publications (such as books, book chapters, and reports). Other refinements might include greater discrimination between journal rankings, for example, based on impact factors. However, decisions about further additions and refinements would need to be justified by an increased use of the outcomes of analyses.
Acknowledgments The author would like to thank Alistair Cheyne, Sharon Clarke, Frank Guldenmund, Kathryn Mearns, Andrew Neal, Arlene Walker and Dov Zohar for their speedy and conscientious reviewing work for this special issue. References 1. Cox, S and Flin, R. Editorial: safety culture. Work & Stress 1998, 12(3): 187-188. 2. Hale, A. Editorial: culture’s confusions. Safety Science 2000, 34(1-3): 1-14. 3. Baram, M and Schoebel, M. Editorial: safety culture and behavioral change at the workplace. Safety Science 2007, 45(6): 631-636. 4. Swuste, P. Editorial: WOS2006, regulatory issues, safety climate, culture, and management. Safety Science 2008, 46(3): 345-348. 5. Glendon, I. Safety culture: snapshot of a developing concept. J Occup Health Safety — Aust NZ 2008, 24(3): 179-189. 6. Zohar, D. Safety climate in industrial organisations: theoretical and applied implications. J Appl Psychol 1980, 65(1): 96-102. 7. Brown, R and Holmes, H. The use of factor-analytic procedure for assessing the validity of an employee safety climate model. Accid Anal Prev 1986, 18(6): 455-470. 8. All citation indices were determined through Google Scholar on 31 January 2008. While other sources of citation counts are available (for example, Web of Science), Google Scholar was selected as being particularly comprehensive and likely to provide the greatest variability in citation data for analysis. For more information on citations and related performance indicators, see www.heeact.edu.tw/ranking/EngIndicator.htm.
188
9. Zohar, D. Safety climate in industrial organisations: theoretical and applied implications. J Appl Psychol 1980, 65(1): 96-102 (290 citations). 10. Zohar, D. A group-level model of safety climate: testing the effect of group climate on microaccidents in manufacturing jobs. J Appl Psychol 2000, 85(4): 587-596 (156 citations). 11. Guldenmund, F. The nature of safety culture: a review of theory and research. Safety Science 2000, 34(1-3): 215-257 (147 citations). 12. Flin, R, Mearns, K, O’Connor, P and Bryden, R. Measuring safety climate: identifying the common features. Safety Science 2000, 34(1-3): 177-192 (130 citations). 13. Dedobbeleer, N and Béland, F. A safety climate measure for construction sites. J Safety Res 1991, 22(2): 97-103 (114 citations). 14. Spearman’s rho (ρ) is used in this and other instances in this article because of non-normal distributions. 15. Strictly speaking, the highly skewed distribution would preclude parametric analysis in this and the other instances in this article. However, as ANCOVA is relatively robust to breaches of normality, the analysis was undertaken as indicative rather than definitive. 16. This arbitrary scoring system was selected on the grounds that, typically, an article’s first-named author contributes the greatest amount, for example, writing the first draft. Less than 12% of all articles were authored by researchers from more than one country, and this points allocation was consistently applied to the 24 articles in this sub-sample. 17. For more information on the top 20 countries by all scientific articles and citations, see www.in-cites.com/ countries/2002allfields.html. 18. Indices might include number of universities, percentage of population with degrees, or levels of active research. Other options could include some measure of national-level occupational injury data. However, these are known to be subject to differences in reporting, recording and analysis — even definitions of occupational fatalities differ — see, for example, Glendon, A and Clarke, R, How occupational accidents and diseases are reported in the European community, Luxembourg: Office for Official Publications, 1988 (also available in Spanish, Danish, German, Greek, French, Italian, Dutch and Portuguese). 19. The exceptions were France, for which the 1 January 2008 figure was the one available, and Israel, for which the CIA World Factbook estimate for 2007 was the one available. Given the error variance in all of these estimates, it is highly unlikely that the different sources for these two estimates would have made any difference to the calculations. 20. It is interesting to compare the analysis of scientific articles in this particular OHS field with a comment on the production of occupational health scientific articles, which cited evidence to indicate that, while the US, the UK, and Scandinavian countries had the highest production rates, when adjusted for population size, Scandinavian countries led the field. See Smith, D, The occupational health and safety editorial, J Occup Health Safety — Aust NZ 2008, 24(1): 3-6.
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
2_Editorial_24(3)
10/6/08
5:27 PM
Page 189
Editorial 21. This analysis is at variance with a recent statement made by an Australian OHS researcher (in response to a questionnaire compiled by the professional journal National Safety) who stated that, “In more general areas of OHS such as safety climate, accident causation and human error, the leading research is being done in the UK and Holland. Australia is not even on the radar on these issues” (OHS researchers speak out, National Safety 2007, 78(10): 46-48). The analyses described in the current article suggest that, while it is correct to state that the UK and the Netherlands are among the high performers in this field, greater scrutiny of relevant publications strongly indicates that, in the safety culture/climate domain at least, Australia is holding its own in a group headed by Israel and also including Ireland, Switzerland, New Zealand and the Scandinavian countries. 22. Quinlan, M. Forget evidence: the demise of research involvement by NOHSC since 1996. J Occup Health Safety — Aust NZ 2000, 16(3): 213-227. 23. Capra, M. The future of OHS education: a time to bring all parties together. J Occup Health Safety — Aust NZ 2006, 22(4): 283-286.
24. Mayhew, C. Funding for OHS research in Australia. J Occup Health Safety — Aust NZ 2006, 22(6): 499-504. 25. The spread of safety culture/climate research in Australian universities indicates strong potential for local expansion of research in this field in the event that further dedicated research funding becomes available. For a broader analysis of the mechanisms by which social scientific research in Australia and elsewhere is being increasingly channelled towards serving corporatist ends, see, for example, Thornton, M, The retreat from the critical: social science research in the corporatised university, Australian Universities Review 2008, 50(1): 5-10. 26. Of greater concern might be results from a small-scale annual survey of chief executive officers of large Australian companies who responded to three “wish list” questions on recent achievements, current desires and a 2008 “to do” list (Durie, J, Corporate captains learn to price in risks, The Weekend Australian, 22-23 December 2007). Of 25 CEO responses, only two mentioned safety — one was the CEO of an oil company and the other was from a large construction company. While this might be an improvement on responses to similar surveys in previous years, a reasonable target for OHS researchers could be to seek to exert greater influence at this level of industry.
J Occup Health Safety — Aust NZ 2008, 24(3): 179–189
189
