Need Help Creating an MSA
Six Sigma – iSixSigma › Forums › General Forums › Tools & Templates › Need Help Creating an MSA
- This topic has 12 replies, 7 voices, and was last updated 2 years, 6 months ago by
Strayer.
-
AuthorPosts
-
November 14, 2019 at 9:09 am #243487
SebvlarsenParticipant@SebvlarsenInclude @Sebvlarsen in your post and this person will
be notified via email.Can anyone help explaining the concepts of MSA and how to create one?
Im doing a project on improving delivery time for orders sent – but in this relation i cannot figure out how to create an MSA for this…
0November 14, 2019 at 12:40 pm #243490
Katie BarryParticipant@KatieBarryInclude @KatieBarry in your post and this person will
be notified via email.@Sebvlarsen Did you search our site before asking your question? We’ve got a bunch of content on Gage R&R/MSA in our Tools & Templates section: https://www.isixsigma.com/tools-templates/measurement-systems-analysis-msa-gage-rr/
0November 14, 2019 at 6:55 pm #243501
SebvlarsenParticipant@SebvlarsenInclude @Sebvlarsen in your post and this person will
be notified via email.Hi,
I did not see these, thank you!
i will take a look.
0November 14, 2019 at 7:10 pm #243502
SebvlarsenParticipant@SebvlarsenInclude @Sebvlarsen in your post and this person will
be notified via email.Hi again,
I have looked through all the topics now, but it still deosnt make sense with the project im working on.
My main Six Sigma goal os to improve on delivery time of orders being sent.
I have collected data from 1000 orders, whereas 512 of them were delivered too late (later than 5 days).
i also have the top 5 reasons for these.
But how do i do an MSA for this kind of data?
0November 14, 2019 at 10:52 pm #243508
Katie BarryParticipant@KatieBarryInclude @KatieBarry in your post and this person will
be notified via email.@sebvlarsen What do YOU think you should do? Why or why not? Be more specific about what problem you’re having and you’re more likely to get a response. People want to see that you’re putting in a good-faith effort to figure out the solution on your own.
0November 15, 2019 at 12:16 am #243509
StrayerParticipant@StraydogInclude @Straydog in your post and this person will
be notified via email.You are using attribute data (yes/no) rather than proportional data where there is a range of values for the variable. Look up MSA for attribute data and assessing reliability of qualitative measurement. In your case I’d be more concerned about sample size and sample selection, unless there’s a question about accurately measuring days to delivery.
0November 17, 2019 at 10:54 am #243525
FabioCacciottiParticipant@FabioCacciottiInclude @FabioCacciotti in your post and this person will
be notified via email.MSA stands for Measurement System Analysis therefore it is related to the “measurement system” (combination of gauges, instruments, methods, people, ecc..) used in the process to collect data. In this respect, from what is shown in the conversation, it seems that the process data are those related to collection of delivery times which supposingly are derived from “clocks” (the measurement instrument). Purpose of the MSA is therefore to check whether the “clocks” are correctly recording the process data (delivery times) in terms of: resolution, accuracy, stability, linearity, repeatability and reproducibility. That said, at a first sight and from my personal experience, usually there is not too much need and value in conducting MSA’s about “clocks” especially if the granularity of the measurements (in the case of delivery times) it’s hours or even days. In conclusion I would suggest that you can even skip the MSA step in your project, but now you know why!!!
0November 17, 2019 at 9:53 pm #243533
StrayerParticipant@StraydogInclude @Straydog in your post and this person will
be notified via email.@FabioCacciotti MSA should always be part of the process. We need to know about confidence in our measurements and evaluate how much of the variation we’re seeing may be due to our measurement system. Reputable polls, for instance, state a margin of error based on sample size relative to population and, if they’re sophisticated enough, considering randomness of the sample. The problem many beginners have with MSA is that the tools they’ve been taught may not be applicable to the measurement system they’re using. For instance, you wouldn’t use GR&R unless your measurements come from a calibratable instrument such as a clock.
0November 18, 2019 at 12:25 am #243534
FabioCacciottiParticipant@FabioCacciottiInclude @FabioCacciotti in your post and this person will
be notified via email.@Straydog, I know very well what is (and it is much more valuable of a “simple” calibration) and the importance of MSA and I do push all my trainees and coachees to use it, but on the other hand I also convey the message to not “overprocess” projects where there is not a real need to use a specific tool and from what I could realize about Sebvlarsen project I could not see the real need to carry out an MSA about the clock used to record delivery times especially because the granularity required is about hours or even days. But if you could see something I could not, maybe you could share your advice to Sebvlarsen about the specific MSA he should carry out which is what he’s looking for.
0November 18, 2019 at 8:56 am #243542
[email protected]Participant@[email protected]Include @[email protected] in your post and this person will
be notified via email.@sebvlarsen The purpose of MSA is to make sure that before moving on to process analysis, we analyze how the process data is being collected and the people or instruments collecting data have an accurate and consistent procedure to do it. As this is the primary metric for the project, if I were to perform MSA on delivery time reduction project, I would go and observe the delivery time data collection process – how delivery times are captured, who captures it and where it is captured and reported. I would observe this process at the start of the process – when the package is despatched and when and where the package is received if this is all done manually, I would look for operator to operator variation in the time recording procedure. If there is varation in the way data is collected, I would work on addressing to ensure ‘after’ data is recorded correctly. I would document the process as part of MSA, discuss it with the team to see if they see any defect opportunities and if satisfied, conclude my MSA with gained confidence that the metric is being measured reliably. Hope that helps😊
0November 18, 2019 at 12:19 pm #243541
Alexander SilantyevParticipant@SilantyevInclude @Silantyev in your post and this person will
be notified via email.Hi,
you can check out this article:
To understand how to do an MSA, you need to look into the information flow. The lead time for each order is characterised by the start time and the end time. What is the variation (actual or potential) of these measurement points. How large is the potential deviation in comparison to the total lead time.
Even though this analysis may not be as technical as Gage R&R you will understand, for example, that the measurement variation is in hours and the process itself lasts a few days – then it’s okay. Or there could be a short actual lead time, then a few hours difference in the data may distort it significantly.
Do you see what I mean?
0November 19, 2019 at 2:08 am #243564
huonghxParticipant@huonghxInclude @huonghx in your post and this person will
be notified via email.Hi,
I think You have to identify what are You measuring data by – which equipment that you use and how do you measure? If you only count the days I think you can skip MSA step. But you have to review the method or procedure for collecting above data.
0November 19, 2019 at 8:46 pm #243577
StrayerParticipant@StraydogInclude @Straydog in your post and this person will
be notified via email.You shouldn’t skip MSA. Per my previous comments, you need to use the appropriate tool/method rather than force-fit a tool such as GR&R. Also beware that there may be hidden (uncaptured) data when you’re dealing with something like delivery time. I recall a case where the data showed that delivery times were nearly always met and measurement of days to delivery was accurate, according to MSA. But customers complained. It turned out that customer service did not enter orders until they could confirm with production since the system made it a real pain to change production schedules and/or orders and keep them in synch. So this, and negotiating with the customer, was all done off-line with no record!
0 -
AuthorPosts
You must be logged in to reply to this topic.