Getting the measure of safety culture

Bill Dixon explains how an analytical approach can improve safety campaigns. 

“We need to improve the safety culture on this site!” is commonly heard on sites around the country. Easy words, but what do they really mean? What is safety culture, and how do we know if we have improved it? When we introduce a new safety campaign – perhaps a poster competition or commissioning a safety video for toolbox talks – do we spend our money and time to best benefit? It’s a topic I’ve been researching for the past four years as part of my professional doctorate research at Salford University.

Improvements in construction site safety are always being sought, with three distinct tranches of historical development combining to improve safety on site: technology, systems and culture.

We are all familiar with recent technological progress, such as the design and introduction of podium steps, developed after the introduction of the Work at Height Regulations 2005. Company systems have evolved to a point where their use is seen as a commonplace item in risk assessments, and a culture shift has seen workers embrace podiums as a normal site tool.

The most recent tranche, culture, hit the headlines after the investigation into the Chernobyl accident in 1986 (Cooper 2000). There is general agreement that the seven core facets of safety culture are:

• Management attitudes & commitment
• Safety training
• Communications
• Safety practice
• Risk perception
• Employee involvement
• Work pressure.

Many companies spend time and money trying to influence the safety culture in the belief that this will improve worker safety. Balfour Beatty’s Zero Harm campaign (above) is a high-profile example of safety culture intervention. But do these activities actually have any academic rigour, or are staff just doing what they feel is right? Do culture change interventions, such as promoting safety to the first item on every meeting agenda or inviting staff to “skip level” meetings, actually work?

Can we target our time and money better? Which of our intervention strategies yields the best or worst results? The only way is to find an accurate and cost-effective way of measuring culture before the intervention, then remeasuring after the intervention has been embedded.

Typically, the construction industry is numbers driven and embraces quantitative reports. Statistics, percentages, dashboards and graphs typically populate most project reports. In contrast, however, safety culture is a qualitative entity. The conventional way to assess a qualitative issue is via structured or informal interviews with a highly trained, impartial person to spend large amounts of time with hand-picked workers.

This raises issues of cost, integrity, impartiality and confidentiality. In addition, qualitative analysis tends to give rise to verbose reports that do not appeal to busy construction executives who are normally used to numerate reports. There has to be a better way.

Optical mark recognition software

We have seen that safety culture is felt to consist of seven key facets. By analysing the questionnaires used by six other researchers it has been possible to distil their questionnaires into a new one. The six sets of questions were selected from differing countries and different industries to present a cross section of sample questions. A total of 382 questions from the six original research works were examined. From this, a total of 46 questions were found that summarised all of the 382, and a focus group was used to hone the proposed questionnaire. The 46 questions aim to elicit feelings with respondents indicating their feelings on a Likert scale (strong negative to strong positive).

But on a large site of say, 500 workers, a 20% uptake would lead to 100 x 46 question responses that need collating. As the main focus of the research was to produce a simple procedure for implementation on site, then manually inputting each piece of data was felt to be inappropriate. Optical mark recognition (OMR) software was used to analyse the data. The type of OMR that most of us are familiar with, from our school days, relies on a question paper and a response paper, which used to look like an A4 version of the lottery number forms. But new software that could be easily programmed to read the responses from any bespoke questionnaire was purchased and tested.

Testing of the questionnaire’s initial design has now been carried out on two busy sites in England, and the results shared with the site teams. On site, workers completed the questionnaires in an average of 14 minutes and the responses were scanned using the site’s multi-sheet scanner and saved as a PDF file in less than a minute.

The OMR software had graph reporting capability, which produces several default graphs, with the most useful showing the aggregated scores of each of the seven main facets of safety discussed above.

Capturing survey responses online was considered, but in view of the practical issues around gathering perhaps 100 responses in a short space of time, paper and pen was considered simpler.

Implementing this system would allow companies to take a snapshot of safety culture on each of their sites, allowing them to isolate the facets that need more attention and concentrate their investment in those areas.

It is hoped that this simple, accurate and cheap system will allow companies, over time, to better assess the safety culture intervention strategies they choose to implement.

Bill Dixon FCIOB is a health and safety manager, currently studying for a professional doctorate at the University of Salford while working on a major site in Shanghai