At about 10.30am Mary went for her rest break and slipped over on some spilt milk (tripped over a box) and hurt her arm. An ambulance was called to take her to hospital. She has broken her right arm, which has been plastered. She will be off work for at least 3 weeks [strained her wrist and will be off work for a day].
Page | 83 Figure 2.
Example of additional information provided in the maximum information and major injury scenario - Slip version
Mary does not usually work on Thursdays but was covering for a friend who was on holiday.
Mary closed her checkout at the usual time for her mid-morning break and waited for a friend on the checkout next to hers to serve her last customer and they both went to their break together as usual.
They were walking together past the checkout when Mary slipped over on some spilt milk and fell awkwardly on her right arm.
The First Aider attended and an ambulance was called to take Mary to hospital. At hospital she was found to have a broken right arm. She will be off work for at least three weeks with her arm in plaster.
A customer had seen the milk and reported it to Bill the Supervisor.
Bill confirmed that the spillage had been reported by a customer and the Cleaner had been asked to clear it up five minutes before the accident but had not got round to dealing with it.
No warning signs had been put out.
It is not known how long the milk had been on the floor before it was reported by the customer.
Spillages around the checkouts are very common.
According to the Accident Book four other people had been injured in slipping accidents in the past six months.
The scenario-based exceptional event was repeated and reinforced in the Maximum detail version as the fact that Mary was covering for her friend and did not usually work on Thursday was given twice, once in the general introduction and again in the additional information. More respondents selected the scenario exceptional event (Mary working on Thursday) under the minimum detail condition (22) than under the maximum detail condition (9) and this difference was significant (χ2 (1) = 5.45 p = 0.20). Notwithstanding this, Mary’s decision to work on Thursday was not selected significantly frequently by any of three respondent groups. See Tables 15, 16 and 17 for details of frequency of the selection of the scenario exceptional event for the counterfactual, prevention and causal sentences.
Page | 84
Manipulation of injury severity
The severity of the slip or trip injury was manipulated by having two levels of injury. In the major injury version Mary suffered a broken arm and a three week absence from work, whereas in the minor injury version she suffered a strained wrist and a day’s absence from work. More extreme outcomes (death and no injury) were considered but they did not allow respondents the possibility of both upward and downward counterfactual alternatives as both were anchored at the extremes. It was important to maintain the ecological validity of the scenarios as the vast
majority of reported slip and trip accidents actually result in strains or fractures. Referring to Figure 3, if outcome severity is considered on a linear scale with 0 being no accident and 10 being a fatality, it is suggested that the minor injury suffered by Mary would be at about point 2 on the scale and the major injury (broken arm) would be around point 8 on the scale.
Figure. 3 Design of the scenario allowing for better and worse outcomes. Scenario design allowing for a better or worse outcome
Norm = No Accident Minor Injury Major Injury Fatality 0. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Increasing severity of injury / Exceptionality
Page | 85 Piloting of the scenarios
The research scenario and questionnaire were piloted with 10 local businesses and 10 Safety Professionals. As a result of the responses obtained the questionnaires were amended by improved formatting and layout, some questions were omitted and others refined. Piloting of the questionnaire indicated that the scenario actor was not often identified spontaneously in the completed sentences so a further question was added which asked the respondent to indicate from a list of scenario actors which of them their sentence referred to. The person identified in this question was later used to judge the degree of control they had over the specific subject referred to in the completed sentence.
After reading the scenario, respondents were asked to complete a questionnaire which included writing three sentences focusing on how the outcome of Mary’s accident could have been different (counterfactual thoughts), how Mary’s accident could have been prevented and what the cause of the accident was, and to complete a number of other questions and scales.
In order to enhance ecological validity respondents were asked to complete the questionnaire as they would respond to such an accident as a Safety Professional, a Manager or, in the case of an Accident Subject, to imagine themselves as being Mary.
The counterfactual sentence was prompted by the following statement – ‘After Mary’s accident you found yourself thinking ‘If only…’. How would you continue this thought?’ Respondents were requested to complete the counterfactual sentence – ‘If only… things could have been different’. Respondents were then asked if they believed that Mary's accident could have been prevented. 95.8% of respondents considered the accident could have been prevented and they were then asked to complete the prevention sentence – ‘Mary’s accident could have been
prevented…’. Finally all respondents were asked to say what the cause of Mary’s accident was by completing the causal sentence – ‘The cause of Mary's accident was...’. Respondents who received the longer questionnaire version were then
Page | 86 asked to complete a modified Consideration of Future Consequences (CFC) Scale based on Strathman et al. (1994).
The CFC Scale was developed to measure the consideration of future consequences in a broad way. This research focused on a specific aspect of respondents’
consideration of future consequences, namely that of future safety, and the questions making up the scale were modified to reflect that aspect. The tone and structure of the questions was maintained to ensure the integrity of the scale. Strathman et al.’s (1994) original questions and the modified versions are shown in Table 4. The questions relating to the modified CFC Scale were coded and scored according to Strathman et al. (1994).
The modified 12 item CFC Scale gave a Cronbach’s Alpha result of .651 but this was increased to .735 if item 8 was removed. Subsequent analysis of respondents’ CFC scores was based on an 11 item scale.
Coding of responses
As part of the questionnaire respondents were asked read a slip or trip scenario and then to complete sentences describing how the outcome might have been different (counterfactual), how the slip or trip might have been prevented and finally to describe the cause of the slip or trip. The individual responses were collated into a single Word document along with the Accident Subject’s counterfactual thoughts relating to their own accident. I examined the sentences and identified the relevant aspects of their structure relevant to the research and these were coded in to SPSS statistical software. The coding scheme used for the respondents’ sentences is contained in Appendix 1, and the respondents’ completed sentences are contained in Appendix 5 .
Each completed counterfactual sentence was coded against 14 structural dimensions, whilst the prevention and causal sentences were coded against 13 structural dimensions. The structural dimension of direction was only applicable to the counterfactual sentence. All the coding options for the structural dimensions
Page | 87 produced categorical data, for example a sentence may relate to either an action or inaction. (Table 5 sets out the structural dimensions that this research considered).
Content analysis
After reading the stimulus scenario, respondents completed three sentences to record their counterfactual, prevention and causal thoughts and these sentences were subject to conceptual content analysis in which the coders examined the wording to identify the presence of specific words, their meaning and the concepts contained within the sentence (Krippendorff, 1980; Weber, 1990).
Using a content analysis approach I examined each respondent’s counterfactual, prevention and causal sentences against each of the structural descriptions set out in Table 5. Each respondent’s completed sentence was copied from the hand written questionnaire and typed in to a Word document containing a table in which the respondent’s number, job group and questionnaire version were recorded along with the text of their three sentences. For each sentence the scenario actor identified by the respondent as being associated with a particular sentence was also noted. An example of a single respondent’s completed sentences is given in Figure 4. The full responses are provided in Appendix 5.
Page | 88
Table 4.
Original and modified Consideration of Future Consequences Scale questions
Original questions Modified version used for consideration of safety in the future
1. I consider how things might be in the future, and try to influence those things with my day to day behavior.
I think about safety in the future and try to influence things by my day to day behaviour.
2. Often I engage in a particular behavior in order to achieve outcomes that may not result for many years.
I think about safety in the future and do things now to achieve safety in the years ahead.
3. I only act to satisfy immediate concerns, figuring the future will take care of itself.
Thinking about safety I only do things to deal with the immediate situation, not worrying about the future.
4. My behavior is only influenced by the immediate (i.e. a matter of days or weeks) outcomes of my actions.
What I do about safety is only influenced by how things work out in the short term. 5. My convenience is a big factor in the decisions
I make or the actions I take.
My convenience is a big factor in how I make decisions or take actions about safety.
6. I am willing to sacrifice my immediate happiness or well-being in order to achieve future outcomes.
I am willing to put in extra time, effort and money now to ensure that the job is safe in the future.
7. I think it is important to take warnings about negative outcomes seriously even if the negative outcome will not occur for many years.
I think it is important to take warnings about safety seriously, even if it is unlikely that an accident will happen for many years.
8. I think it is more important to perform a behavior with important distant consequences than a behavior with less-important immediate consequences.
I think it is more important to do something about serious accidents in the future than minor accidents now. 9. I generally ignore warnings about possible
future problems because I think the problems will be resolved before they reach crisis level.
I generally ignore warnings about possible risks in the future, because they generally get sorted out before that happens.
10. I think that sacrificing now is usually unnecessary since future outcomes can be dealt with at a later time.
I think it is unnecessary to change things now to prevent a possible future accident as problems can be dealt with nearer the time.
11. I only act to satisfy immediate concerns, figuring that I will take care of future problems that may occur at a later date.
I only act when there is an immediate risk, I prefer to take care of future problems that may occur at a later date. 12. Since my day to day work has specific
outcomes, it is more important to me than behavior that has distant outcomes.
I believe that safety today is more important than safety at some time in the future.
Page | 89 Figure 4. Example of a respondent’s completed counterfactual prevention and causal sentences Counterfactual sentence If only Bill the Shop Floor Supervisor had taken
immediate action when the spill was first reported things could have been different.
Prevention sentence Mary’s accident could have been prevented if Bill had taken immediate action… closed the checkout, placed a cone near the spill and stood by the spill until the Cleaner arrived.
Causal sentence The cause of Mary’s accident was the failure to have a
procedure in place to deal with spillages.
Each sentence was examined by asking a series of questions about its contents to determine the relevant structural dimensions and entering the appropriate coding in to IBM SPSS Statistics 20 for analysis. The complete set of coding instructions is given in Appendix 1. All the structural dimensions that were identified and coded related to different categories of the various dimensions being tested, therefore this study relied on the use of appropriate non-parametric statistical tests.
Data cleaning and coding checks
The data set was checked for general coding errors and corrected by reference to the original responses contained in the questionnaire where necessary.
Specific cross-referencing was undertaken to check the respondents’ group ( Safety Professional, Manager or Accident Subject) against the questionnaire version. The main purpose of this was to ensure that the data relating to Managers was not influenced by Safety Professionals who were also managers. Where there was any indication that the respondent was or had an element of being a Safety Professional they were recorded as such. For example, if a questionnaire indicated that the respondent was a Manager but the questionnaire was one sent to a Safety
Page | 90 questionnaires were sent to local authorities or colleges it is quite possible that a Safety Professional in a management position completed one. They might describe themselves as being a Manager but they are professional safety officers who are managing a team. It would have introduced a bias if these responses were coded alongside other Managers who had no specialist safety experience.
In 59 cases there was some uncertainty as to the appropriate coding for the respondent’s group. Forty three questionnaires designed for Safety Professionals were returned by respondents who classified themselves as either Managers or Supervisors (40) or Accident Subjects (3). For each of these questionnaires the respondent group coding was cross-checked against the details provided regarding their employment sector and number of people they managed to ensure they were coded appropriately. Sixteen questionnaires designed for Managers were completed by respondents who indicated they were not Managers (15 Safety Professionals and 1 Accident Subject). It is likely that when a Manager’s questionnaire was sent to a business who had a manager / supervisor with specific health and safety
responsibility that it was directed internally to them as being the most suitable person to complete it. All such questionnaires were coded as Safety Professionals and the single questionnaire from an Accident Subject was coded as such.
Managers do have accidents and the occupation of Accident Subjects was not requested in their questionnaire. There were no coding anomalies from the Accident Subjects’ questionnaires, all the respondents indicated they were from that group.
In total the respondent’s group could not be satisfactorily determined in a total of 19 cases and these were excluded from analysis.
Inter-rater reliability
A 10% sample of randomly selected responses were coded by a work colleague who had been trained to identify the different structural dimensions from the completed sentences but who was blind to the research aims. Differences in coding were discussed and agreed. An inter-rater reliability analysis using the Kappa
Page | 91 statistic (Landis & Koch, 1977) was performed to determine consistency between the two raters. Details of the inter-rater reliability for the structural dimensions of each of the three sentences is given in full in Tables 86 and 87 in Appendix 2. The mean Kappa score for all parameters was .753 p = < . 001 with a range from 0.634 p = < .001 to 0.903 p = < .001.
Research design
Eight versions of the scenario were produced in a 2 (slip / trip) x 2 (minimum / maximum detail) x 2 (minor injury / major injury) study, the versions of the scenario are summarized in Table 6. The questionnaires for each population were colour coded for ease of identification - green for Safety Professionals, yellow for Managers and pink for Accident Subjects. Examples of the scenario versions and full and short questionnaires can be found in Appendix 4.
Long and short questionnaire versions
When the questionnaire was used for actual data collection the response rate from the Manager and Accident Subject populations was lower than anticipated. One possible reason for this was the length of the questionnaire, so a shortened and simplified questionnaire was produced for these two groups. Table 91 in Appendix 4 sets out the questions used in the full version completed by Safety Professionals and the shorter version completed by Managers and Accident Subjects.
Page | 92 Table 5. Structural dimensions of the counterfactual, prevention and causal
sentences Number Counterfactual sentence Prevention sentence Causal sentence 1. Direction (better or worse outcome) Yes Not applicable Not applicable
2. Action or inaction Yes Yes Yes
3. Addition or
subtraction
Yes Yes Yes
4. Exceptional or
routine antecedent
Yes Yes Yes
5. Antecedent
timescale
Yes Yes Yes
6. Locus of control Yes Yes Yes
7. Dynamic or
passive antecedent
Yes Yes Yes
8. Case specific or
general
Yes Yes Yes
9. Known or inferred
antecedents
Yes Yes Yes
10. Personal or
situational
Yes Yes Yes
11. Spontaneous
identification of the scenario actor
Yes Yes Yes
12. To whom did the
sentence refer?
Yes Yes Yes
13. The specific
subject of the sentence
Yes Yes Yes
14. The ‘domain’ of
the specific subject
Page | 93 Table 6.
Summary of questionnaire versions
Accident subject Manager Safety Professional
Accident type Outcome severity Background information Accident type Outcome severity Background information Accident type Outcome severity Background information Slip Minor injury Minimum information Slip Minor injury Minimum information Slip Minor injury Minimum information Slip Minor injury Maximum information Slip Minor injury Maximum information Slip Minor injury Maximum information Slip Serious injury Minimum information Slip Serious injury Minimum information Slip Serious injury Minimum information Slip Serious injury Maximum information Slip Serious injury Maximum information Slip Serious injury Maximum information Trip Minor injury Minimum information Trip Minor injury Minimum information Trip Minor injury Minimum information Trip Minor injury Maximum information Trip Minor injury Maximum information Trip Minor injury Maximum information Trip Serious injury Minimum information Trip Serious injury Minimum information Trip Serious injury Minimum information Trip Serious injury Maximum information Trip Serious injury Maximum information Trip Serious injury Maximum information
Page | 94
Results
Previous research identified a number of different structural dimensions to the way that respondents record their counterfactual thoughts following an unwanted outcome. The unwanted outcome has usually been presented in the form of a vignette or scenario and respondents are asked to record their counterfactual thoughts by completing an ‘if only...’ type sentence. Counterfactual thoughts bring