Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
Choose Language Hide Translation Bar
Highlighted
abmayfield
Level V

Visualizing measurement error

I think this is a JMP/statistics 101 question, so here goes:

 

In a recent project, I am comparing field data ("ABM_DHW" in the data table) vs. what is essentially a standard ("CRW_DHW" in the table). If you regress my field data (y axis) against the standard (x axis) and compare it (blue line) against the identity (i.e., standard vs. standard=red line), you will see that they agree well until a value of ~6 is reached. I don't think a reviewer would complain about this, and I guess this figure shows this error well enough (script is found within table), but I wonder if there is a way to essentially better depict the error. I know you can specify limits but I don't know how that would work unless you specified each data point in the table to equal that of the respective standard. Basically, I would like to visually depict what is essentially the residual, perhaps by shading the area between the two lines or using drop down lines from the measured to the theoretical value. I really feel like I saw this option demonstrated in a webinar. What I have done instead is plot the residual as a second y axis (second script), but I would prefer to use drop-down lines or inter-line shading if that's possible (or any other means of highlighting deviation from a standard). I feel like engineers must have a need for this, so I bet there is a platform in JMP Pro for doing it. 

Anderson B. Mayfield
2 REPLIES 2
Highlighted

Re: Visualizing measurement error

I suggest that you investigate JMP Help about Measurement System Analysis and the Variability Chart. These platforms provide a lot of analyses and options for example this question.

Learn it once, use it forever!
Highlighted
abmayfield
Level V

Re: Visualizing measurement error

Thanks. I did find something that is getting at what I want (albeit not with publishable visuals). Variability/attribute chart. Basically, you put both variables as Y (the field and standard data), with the standard additionally in the "standard" box. In my case, it's not really appropriate to have any sort of grouping, so I just put time. Then, you get a chart showing when the data did and did not agree with the standard. Also, you can select "linearity study" and it will basically show you the residuals vs. the standard, as well as some parameters I've never heard of, but which might be very useful ("bias" and "linearity"). And the "bias test" will show you more data on when your data did not match the respective point in the standard data set. I definitely want to explore this platform more, as it does seem geared towards those comparing field data to a standard. See last two new scripts for what I've done (so far). 

Anderson B. Mayfield
Article Labels