Rank: Forum user 
|
Hello fellow safety professionals
A while back I posted a topic on this site relating to Incident trends, to which i gained quite a few replies, well i want to post it again:
I am currently working on a large oil and gas project in Kazakhstan, total cost is in excess of 40 billion dollars, Its a multi national project with Russian and English as the two "project languages"
We have identified a trend with incidents, we have long periods of no incidents but when one does occur it is generally followed by another 2 - 3 all within a short period of time, I have been here a number of years and have witnessed this happen on a number of occasions. It has happened often enough for the trend to be highlighted
I have some ideas why this happens, but would like to ask (again) my fellow safety professionals for their opinions
Spasibo Bolshoe
|
|
|
|
|
|
Rank: Super forum user
|
Staggsy,
I have noticed the same trend in our system but cannot find any statistical or causal link.
I have examined our stats on these occasions and compared age, time of day, season, holidays, shift patterns, type of injury, causal factors, sex, trade, location, sporting events, etc,etc.
The problem is that with a good safety record you are always looking at small data sets and it's easy to spot a trend that's not there. It's noise in your data.
Sorry I can't help you further than saying good luck with keeping your accidents to a minimum.
|
|
|
|
|
|
Rank: Forum user
|
It is probably the real randomness of random events against the human inability to randomise. (which is why computers have to randomise for us - and we reject their non-random selections such as 1,1,1,1 or 5,4,3,2,1).
Having eliminated an obvious reason, accept that there is no reason.
|
|
|
|
|
|
Rank: Super forum user 
|
Any truly random sequence contains clusters like this - it's an effect of randomness. If the events were at regular intervals (say 1 every 7 days), that's when you'd have a non-random pattern. What you have here is a mix of short intervals and long ones and that's exactly what a statistician would expect if they happen at random intervals.
It's our human perception of what a random sequence ought to look like that's at fault. This is classically demonstrated by the experiment of asking someone to draw some random dots on a piece of paper, and comparing with a picture of random dots made using a proper randomising process. The human being will draw the dots quite evenly spaced; the randomising process will include clusters.
|
|
|
|
|
|
Rank: Guest
|
I have had similar trends in a nursery. I looked at the reason why and found all the accidents happening on a specific day occcured after lunch when the children had had a full on morning followed by lunch and were subsequently very tired.
The solution was to even out the day including changing the lunch hour and encouraging short naps after lunch.
Reminded me of when I was in infant school many years ago (1950s) when napping after lunch was part of the day.
Rich
|
|
|
|
|
|
Rank: Super forum user
|
The weird statistically thing we found was that for several years the number of incidents reported each quarter was 18+ 2 but the types of incidents each quarter was always different; slips/trips/falls in winter, bad backs in summer etc . I could never work out why it always came to the same average number.
|
|
|
|
|
|
Rank: Forum user
|
Thanks for the replys people, But im still no further forward lol
|
|
|
|
|
|
Rank: Super forum user 
|
staggsy
I'll put my penny's worth in, many years ago when I was working for a train service provider it was identified that many railway incidents took place after the individual had a break ie day after resuming from a rest day, annual leave and even after a lunch break.
The problem here is not identifying the causal link but implementing a control. You cannot stop staff from taking breaks! My theory was that where people who work in offices slowly get back into a routine ie making a cuppa, checking emails, etc, this is obviously not possible for some operational staff like train drivers. So they went from a relaxed mode to a fully operational one in an instant. That was the crux of the issue - it may be of some use or on the other hand...
Ray
|
|
|
|
|
|
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.