Introduction to Data Prediction
As safety professionals we collect data. It doesn't make a difference if your focus is general safety, occupational hygiene or a combination of the two. We perform safety observations, collect air samples and perform some analysis on the data to make inferences on potential hazards. Wouldn't it be nice if we could use data to predict the future?
Data Collection can predict what?
Collecting the right data points can help in forecasting potential exposures that can then be prioritized and an effective elimination or control mechanism developed. The first thing we have to determine is what data we need to obtain.
There are two types of data categories that we hear discussed in the business world today; leading and lagging indicators (or metrics). Leading indicators tend to be direct or indirect precursors of an incident, such as workplace conditions and behaviors. Collecting this information provides the opportunity to implement preventative actions prior to an unwanted incident or injury. Much like a coach, you manage your team as the game unfolds, trying to score.
Lagging indicators are those that are historic. OSHA recordable, DART rate and experience modifications are like opponent’s points on the scoreboard. At the end of the game, it’s too late to change them.
Let’s say that you are the Safety Director for a large construction firm of 1,000 employees. The safety committee has decided that using worksite observations as a leading indicator will allow us to predict exposures and develop control options. These leading indicators, such as with quality worksite safety observations, diversity of observers, and actions on collected safety data, have proven to predict injuries. For more insight and detail, view this white paper from Predictive Solutions.
Determine how the data will be collected before it is collected. If supervisors or frontline workers are to be collecting information, then a simplified mechanism that allows for data collection and storage with minimal disruption in their work day will aid in obtaining their support. This also provides for the collection of reliable and useable data.
Pre-determine how large of a sample pool that will be effective. Collecting safety observations on small construction crew of less than five people, three times a day, may not be advantageous. Develop a strategy to ensure contractors and crews with the most manpower coupled with higher risk activity are afforded more observations than smaller crews with less at-risk activity.
How should the data be captured and analyzed?
As leading indicators are collected, there must be a plan in place to utilize them beyond the immediate activity. Simply observing and correcting is known as the ‘whack-a-mole’ approach and doesn’t promote safety, because the same things could pop up over and over without effective resolution. For example, a police officer pulls a car over for speeding and gives a warning. Does this stop the at-risk behavior – in this case, speeding? How do you know? Can the observations be tracked– both positive and negative – to establish a tendency or trend? Following an action plan to address a trend, can the observations support a positive shift? A single instance of finding and correcting an at-risk behavior or condition is but the first step. Establishing overall tendencies from the expected outcome over time is the goal. Prioritizing the undesired trends is then the next logical step.
In addition to acting on the data, the findings and the resolutions must be communicated. Providing feedback to observers on action items resulting from the observations is crucial. This way they understand their efforts are being heard and acted upon. Coaching observers on good quality observations is also vital so that management is confident enough to act on the data obtained.
Unlike lagging indicators that measure a process purely on failures, leading safety metrics and indicatorscan measure a process on accomplishments. Developing a sound strategy on what is being done to achieve safety is much more effective than hoping and praying that no injuries occur.