Statistics: An Introduction To An New Way Of EHS Analysis and Reporting
Fessler, Mark
(Tokyo Electron America (TEA), Chandler, AZ)
You must log in to view the full proceedings.
LoginEHS in our industry radically improved through the late 1990\\\’s but especially during the last 7 years due to top-down safety cultures established by many leading U.S. Chipmakers who in turn, pushed it further down to the major capitol equipment suppliers. Initial EHS improvements focused on \\\”low hanging fruit\\\” (Training, Equipment Design, PPE) but was less focused on the way to prove the effectiveness of their programs. This was because the initial improvements could be readily seen in lagging indicators like OSHA\\\’s Recordable Incident Rate (RIR). This has been acceptable until now…but not moving forward, primarily because we have unfortunately stabilized above \\\”zero-incidents\\\”. The rapid EHS improvements which reduced the RIR are no longer occurring. If we mathematically model our improvement over the last decade, and if we continue at the current improvement rate of our industry, it will be years, even decades to reach the goal of zero incidents. So, as EHS Professionals, we must ask ourselves, these important questions: 1.”Is there more information in the data we collect than we are currently aware of?”; 2. “Should we collect EHS data in a different ways to better capture this potential hidden information?”; 3. “Is there a way to prove the effectiveness of our existing programs… or planned future programs?” … The answer to these questions is…YES, with the aid of statistics! “Statistics are numbers, but the practice of statistics is the craft of measuring imperfect knowledge”- John Sall, JMP. Proven statistical methods already exist, and new software developments can allow easy application of even complex statistical methods. It is time to introduce statistical methodologies to our EHS metrics for success. This introductory discussion will include some basic statistical data collection procedures and analyses but due to time limits, it cannot provide adequate attention to ensure full statistical training. Thus, to help bring clarity to this topic, some example analyses will be shown, and example proposals for future investigation will be presented. The goal of the discussion is to generate questions, and promote interest in interest in the collaboration of professionals in other industries (automotive, financial, etc.) which have already initiated such analyses. With the improvements made in statistical software development over last decade, you no longer have to be a PhD in Statistics to understand the complexities of raw data variations. Semiconductor EHS professionals can now use such methods as a discovery tool to unveil information not anticipated by our current straight forward analysis.