Accuracy versus Precision
I also added the following example:
Suppose 4 different manufacturing lines are producing cylindrical rods with diameter 3.00 cm. A random sample of 10 rods is taken from each line. The measurements are:
Line 1: 3.05, 3.25, 3.40, 2.55, 2.45, 3.15, 3.00, 3.35, 2.85, 2.80
These are neither precise, not accurate.
Line 2: 2.45, 2.40, 2.50, 2.40, 2.45, 2.45, 2.50, 2.40, 2.45, 2.45
These are precise, but not accurate.
Line 3: 3.15, 3.10, 2.90, 2.85, 2.90, 2.95, 3.00, 3.05, 3.05, 3.10
These are accurate, but not precise.
Line 4: 3.05, 3.00, 2.95, 2.95, 2.95, 3.05, 3.00, 3.05, 3.05, 3.00
These are accurate and precise.
Nice pointer :)
ReplyDelete1. Neither precise nor accurate = Random pattern not clustered & not on the target
2. Precise, Not Accurate = Clustered pattern & not on the target
3. Accurate, Not Precise = Not Clustered pattern but average on the target
4. Precise and Accurate = Tightly Clustered pattern & always on the target
Thanks Athif, for your inputs.
ReplyDeleteexcellent. I could not find this under quality so if you move it under quality it might help another person like me. If you have time will you answer qn 9
ReplyDeletehttp://www.oliverlehmann.com/contents/free-downloads/175_PMP_Sample_Questions.pdf
Hello Anon.,
ReplyDeleteGood catch. Thanks for your feedback. I've added "Quality Management" as a label now.
Regarding Q9 from Oliver Lehmann, I think it's a good question and it stumped me also in the first attempt. But after thinking about it, I think I know the correct answer.
(My) Answer to Question 9 (Oliver Lehmann) is 1 because:
The process is "High Precision" because all the values lie within LSL and USL (not even a single value is outside the limits).
Accuracy is the absolute difference from the mean. As you can see, most values are falling above the mean. Hence "Low Accuracy". If we "adjust" the process such that the values are distributed equally around the mean, the accuracy will improve. That's why the process should be "adjusted". Note the subtle difference between "adjusted" and "improved".
Kudos to Oliver !
Thanks to you for bringing this up here. I think this discussion adds real value to the post.
Thanks. Another doubt As per the PMBOK glossary random points happens during normal cause and non-random during special cause. So I m
ReplyDeleteconfused what is random and non-random and how it looks like in diagram.
The diagram in the Oliver Lehmann question 9 shows data points from common causes (random points). But let's say if you had one point way outside the control limits, it could be due to a special cause (non-random point). Similarly, if you have 7 continuous data points on the same side of the mean, it also indicates a special cause.
ReplyDeleteThanks. This helps
ReplyDeleteAlso remember that if the scenario is accurate but not precise then precision should be adjusted and if the scenario is precised but not accurate then accuracy should be impreoved
ReplyDelete- Sam
I tried to understand this by plotting an area chart using line samples given by Harwinder.
ReplyDeleteMy summary:
1) Neither Precise, Nor Accurate =
Great variations between samples;
Missing the target every time.
"Unpredictable" and "Undesirable"
2) Precise, not accurate =
Very little variations between samples;
Missing the target every time.
"Predictable" and "Undesirable"
3) Accurate, not precise =
Great variations between samples;
Hitting the target sometimes;
other times coming closer.
"Unpredictable" and "Desirable"
4) Precise, and accurate =
Very little variations between samples;
Hitting the target every time.
"Predictable" and "Desirable"
I understand the difference between Accuracy and Precision but I'm not clear on adjust and improve.
ReplyDeleteFrom Oliver Lehmann Q9, The process has high precision but low accuracy. It should be adjusted.
But Sam above replied...
Also remember that if the scenario is accurate but not precise then precision should be adjusted and if the scenario is precised but not accurate then accuracy should be improved
So Low Accuracy - do you adjust or improve?
This comment has been removed by the author.
DeleteUPDATED comment:
DeleteImprove.
Adjust is to move the values close to the desired value.
Improve is to reduce the scatter and bring the values closer.
For better precision you improve (P-I). For better accuracy you adjust (A-A).
Hope that makes sense.
Hi Harwinder,
ReplyDeleteFirst of all, would like to appreciate your efforts!!
As you said above that 'Adjust is to move the values close to the desired value'. This means that you adjust for better accuracy.
Your conclusion does not go well with the statements you mentioned. So, I guess the final conclusion should read as:
"For better precision you improve. For better accuracy you adjust."
Please confirm.
Thanks and Regards,
Aakash Jain
Hello Aakash,
ReplyDeleteMy bad. You are right :)
Thanks for pointing it out.
I've deleted my previous comment and posted the amended one.
Best regards.
Hello Harwinder,
ReplyDeleteI would like to thank you for all your posts which has helped me to clear my PMP certification today!!!
Thanks and Regards,
Aakash
Hello Aakash,
DeleteThat's wonderful. Congratulations. I'm very glad to be of help. Please post your lessons learned on PM Hangout. I'm looking forward to reading your success story.
Best regards.
Really Nice Comments .... it really helped me .. i am planning to appear in PMP on 15th march .... For Better precision we need to Improve Process and for better Acuracy we need to Adjust the process.... or It is true for all process and variable ...... pls confirm ..thanks
ReplyDeleteThanks, that really helps me clear the answer to q9, -> For better precision you improve (P-I). For better accuracy you adjust (A-A).
ReplyDelete