Workplace observations are the key to success in any Behavioral Safety process as the data provide the means to give feedback and rectify any problematic safety issues. Yet, many processes struggle with this fundamental element. What are the issues and the solutions?

There are a number of related features surrounding observations that include checklist types, actual observations and observers, each of which can enhance or detract from results. To obtain accurate observation data that will do the most good requires consideration of the following issues.

Observation Checklists
The observation checklists are not targeting the accident causing behaviors. This usually occurs because the accident records have not been analyzed, and people's perceptions about their 'pet hates' have been allowed to dominate. Examining your existing accident / near miss (Hit?) incident database using data-mining techniques helps to identfiy behaviors of concern.

Often company's simply borrow a checklist from another facility that attempts to cover every conceivable behavior (and psychological states in some cases!) without attempting to identify the small number of behaviors involved in the lions share of incidents. The company is then surprised when the incident rate does not diminish as expected.  An example 'catch-all' checklist is on the left.

Pinpointing 'common' unsafe behaviors can be easily achieved by breaking down your incident records by geographical location and type of injury. A frequency count of the same types of incident will then direct you to the 20 percent of unsafe behaviors triggering 80 percent of your problems. If your accident records are less than adequate (as they often are) and do not specify the behaviors involved, Applied Behavioral Analyses techniques should be used to ascertain the actual unsafe behaviors involved in your incidents.

The unsafe behaviors have not been defined with sufficient precision. This is a very common problem with many Behavioral Safety systems. Some for example, use a category of behavior, like 'Line of Fire' as an individual checklist behavior. Some of the most well known Behavioral Safety processes tend to operationally define a workflow process or task (See the example on the right which is only half of those defined) on a separate sheet. For each task, such operational definitions tend to include between 6-8 behaviors within each definition. The observers then record any observed 'Unsafe' or 'At-risk' behaviors against these. However, if 4 of these relate to the non-use of Personal Protective Equipment (PPE), it becomes unclear as to which one of them was not being used. The sheer number of behaviors operationally defined can be in excess of 100 or more. This means, that in practice, observers rarely refer to them.

Developing very specific checklists forces the identification of the 20 percent of behaviors triggering 80 percent of the incidents! The golden rule is 'the more specific the checklist behaviors, the bigger the impact'. In this way, each behavior is already 'operationally defined' making it very clear which behaviors are being targeted for change (as the example below demonstrates).

The observation checklists focus on unsafe condition's not unsafe behaviors. The observation checklists are not workplace audits of unsafe conditions, they are behavior sampling checklists. As such, the vast majority of items on the checklists should relate to actual behaviors. It is correct to monitor temporary unsafe conditions created by an unsafe behavior (e.g. hoses left lying across walkways), as they are within peoples control. Pure unsafe conditions (e.g. potholes in floor) should be noted, recorded and passed on to the appropriate people to get the problem resolved, but should not form a part of the observational checklist.

Observation Issues
Observations take place at the same time everyday. The observation sampling should be undertaken at random times throughout a week. In other words, the timing of each day's observation must be as unpredictable as possible, else all that might be achieved is that people behave safely only for that particular time period.

Observations are quota driven. Many process require only require people to observe one or twice a  quarter. This means in the last week or so of the quarter numerous observations are turned in, most of which are focused on PPE. Research has shown the impact of Behavioral Safety is much much bigger if daily observations are completed. 

People's names, where they were working, what was said to them, and their responses are recorded. For people to willingly engage in a behavioral safety system, it must be psychologically safe for them to do so. The only way that this can be achieved is for the observations to be anonymous so individuals cannot be identified. Although some argue against this, on the grounds that such information is seen only by a steering committee, I would strongly argue that if it is not collected in the first place, it couldn't be used against people at all.

Observer Issues
Observations not being undertaken: Observers sometimes do not complete their 'scheduled' number of observations. This can happen for a variety of reasons (e.g. a lack of time). The only remedy, is for steering committees or Behavioral Safety project teams to monitor the number of checklist returns to ensure it is not a  problem with any particular observer. If problems are detected, it is wise to discuss any issues with the person concerned.

The percentage safe scores do not reflect reality on the 'shop floor' because observers are trying to convey an optimistic picture of safety in their work area. This self-defeating strategy tends to come about from line-management pressure, that emanates from a fear that their area of control will look bad to their superiors. If people cannot be honest about the safety issues in their work areas, then no Behavioral Safety process is right for them, as they are not yet mature enough to even contemplate its use.

Observers do not provide verbal feedback at the point of observation. In many processes, observers are required to provide verbal feedback - both positive and negative, but it does not always happen. There are numerous reasons why, from people being shy to being intimidated by those being observed. Verbal feedback is simply a means to fix problems associated with people's work that may range from a lack of equipment to standing under a suspended load. As such, all observers should be encouraged to give verbal feedback during every observation.