- JMP User Community
- :
- Discussions
- :
- Order of addition experiments, simultaneous addition, and delayed addition

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

Created:
Mar 16, 2023 07:43 AM
| Last Modified: Jun 8, 2023 9:28 AM
(1432 views)

Hello - I am working on an process that has an obvious order of addition effect. I have a plugin from JMP that uses a pairwise ordering scheme that can be plugged in as a covariate factor. But, one of the components is a powder, so there is the choice to predissolve it in the solvent being used, or to add it directly as a powder. Also, there could be some benefit to having some components simultaneously, while others in order. Finally, there could be some benefit of adding some components simultaneously, but delayed, such as, for example, the addition of component A and B is 2 mins each, but we start adding B after 30 seconds of starting to add A, so there is an overlap of a minute and a half, and then B continues on for another 30 seconds.

The brief literature search I did seemed to only give either complete simultaneous addition, or completely separated addition, with no mixture of the two, nor somewhat overlapping addition between components. Is there a complete way to design this sort of thing, or is this more of an active area of research kind-of thing. If a methodology is out there, is there a straightforward way to implement it in JMP? I'm an engineer so I don't know the literature well at all, so I apologize if this is a rather trivial process.

Thanks!

Edward Hamer Chandler, Jr.

5 REPLIES 5

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

Created:
Mar 16, 2023 09:05 AM
| Last Modified: Mar 16, 2023 6:10 AM
(1421 views)
| Posted in reply to message from ehchandlerjr 03-16-2023

Order of addition is a little tricky.

There is a really good Discovery Summit talk on the subject (there used to be a video too, not sure where that went)

At that link there is also a helper script that is super useful for helping to set up the experimental plan.

JMP Systems Engineer, Health and Life Sciences (Pharma)

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

Hey @Byron_JMP Thanks so much for the suggestion!

Yea so I've already found that talk, and downloaded the addon and everything. However, it only does the standard pairwise method for OoA experiments, where its entirely sequential. I was asking more about, say, where sometimes you might need to add two things together simultaneously. For example, you add components A-C sequentially, then D may react with C in a way that is not preferable unless E is also present. So the question isn't just whether D should be added before or after E, but also should it be added simultaneously. This is an example, and not exactly what I'm working on, but it demonstrates the point that there are situations where one needs the flexibility of the option for concurrent/simultaneous addition for some steps (or concurrent addition with a delay of one of the compounds) rather than the simple binary of purely sequential addition for the entire process.

Does that make more sense?

Thanks!

Edward Hamer Chandler, Jr.

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

It sounds like A-C would become a factor level of the first component with a constraint to disallow level C with A-C?

JMP Systems Engineer, Health and Life Sciences (Pharma)

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

Hmmmm. Maybe I'm misunderstanding, but I think you are saying to constrain it de facto. My example was simply to show that that might be necessary to constrain it. I'm asking about a more general situation where you wouldn't know this ahead of time, and so you want the option in the design so you aren't constrained to a purely sequential addition, and possibly miss steps where concurrent addition is preferable. I could see this being a three level experiment (-1,0,1), but I get the feeling that the scale isn't continuous as its not a time scale, but just a before/after scale, so I'm not sure the midpoint actually makes sense.

Did I clear it up at all or make it more confusing?

Edward Hamer Chandler, Jr.

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content

- © 2024 JMP Statistical Discovery LLC. All Rights Reserved.
- Terms of Use
- Privacy Statement
- About JMP
- JMP Software
- JMP User Community
- Contact