Julian,     Thanks for your response. I am all too familiar with copying/pasting into Adobe Illustrator, in which I can move around anything I want in seconds. I just wanted to make sure I wasn't missing a GraphBuilder option (and I am anti-Adobe, despite using their products daily)! Funny enough, I had to stop using my JMP16Beta (in which, not that you mention, I DO recall being able to drag things around more easily) because I am still running Mojave (and I think the last JMP16 beta release migrated to Catalina and above), so maybe it's time to just upgrade my OS.   BTW, after my post, I played around with the x and y axes (using the "hand" icon), and GraphBuilder will eventually "figure out" the best orientation for the labels to where they don't get cut off; it just takes tinkering (and I am notoriously impatient). As I told Xann and others, my goal is to become so proficient with GraphBuilder that I can do 100% of my figure making in JMP (no Adobe or PPT), but maybe this is unrealistic..... ... View more

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).  ... View more

The newest Venn diagram add-in, which I last used in the past few days) is great and has way more features than before. I would strongly consider formally incorporating it as a full feature under "Graph," not just as an add-in. I know many molecular biologists like myself use this add-in weekly, if not more often! ... View more

I think I actually made this same request at some point! ... View more

I am beta testing JMP 16, and the box-cox transformation is now basically in all platforms that I can see (Distribution, Fit Y by X, Fit Model). AND, even better, a little graphlet pops out to show you what your distribution would look like upon transforming at various lambda levels.  ... View more

I think I kind of got what I wanted by playing around with the formulas. Basically, I suppressed the calculation of the DHW under 28C (since nothing interesting is happening then) and then did a trailing mean DHW over the preceding 90 days since eventually the DHW will fall to 0 once the temperature hits 28. This then means the trailing mean DHW eventually tapers off. Then, the plot effectively resets from year to year. It's kind of cheating, but in the end, it gave me what I wanted! Still, I am open to suggestions if people can think of a better, more objective way to do this that doesn't require so much arbitrary binning and tweaking.    ... View more

Thanks! That is a handy formula, and I did want that information (DHW/year).  But that I really want to do is basically suppress the DHW calculation at time in which no heat is accumulating (i.e., the plateau phase).  Basically, I told JMP to only calculate the DHW when temperatures were above 29 because I want the DHW to fall back to 0 once the temperature has dropped. Is it possible to define "year" in the moving average formula? In that case, I could plot DHW over month for each of the year. If I don't do that, the DHW will keep accumulating from one year to the next!   ... View more

I think a figure may better explain what I want to do:  so the red line represents the degree-heating weeks. Instead of plateauing, it should basically drop back to zero as temperatures fall (blue). But for the life of me, I can't think of how to script this! ... View more

There IS a JMP add-in to convert ESRI files to shape files ( ESRI to JMP Map Converter ), but it is not currently functional in JMP15 or 16(beta). Although it wouldn't have the nice imagery, this could be one way of getting higher-resolution shape files into JMP. ... View more

Bummer! But it's what I expected. Funny enough, both Google Earth and ArcGIS claim to be WMS, but I think they mean you can pull images from OTHER WMS servers into THEIR software (not vice versa).  ... View more

Thanks for the advice and suggestions. I should have explained the experiment better in the initial post. I am guessing they opted for this strange design because the aquaria are so large and so hard to maintain, so they opted to have a few large ones with split plots, rather than 10-20 individual ones. In fact, the strange nature of their design was the main impetus of my post to begin with!   Even when pooling the data across fragments within each tank x light x food (n=18 samples) in the summary file, you can still see 1-2 multivariate outliers. One of them I had already flagged from the PCA because it basically sits off by itself. It seems to be the growth data that are driving it, so I will need to ask the student about it. Maybe some fragments were dying.     They measured light at the three levels (100, 150, and 250 umol/m2/s photosynthetically active radiation), so I am not concerned about that except if one were to argue that, because the depths were inherently different (30, 20, and 10 cm, respectively) this could bias findings. This is a legitimate concern if something else varied across depths (like flow). If nothing else, it's been a learning experience for me because I have never seen an experiment set up this way! Anyway, thanks for all your help and advice.    ... View more

Thanks so much for spending all this time looking at my data. Allow me to answer some of your questions.    The experimental design is basically: 5 coral colonies broken up into 25 2-g fragments, a subset of 18 of which were used in experiments (n=90 coral fragments). I actually took some of Jian Cao's suggestions to look at the nesting random coefficients feature to ensure that the corals essentially all started off similarly (this could very well NOT be the case since different fragments could display differentially recovery from being torn off their parent colony).    30 corals fragments were placed in each of three tanks. Within each tank, 10 fragments were placed at each of three depths (which correspond to different light levels). Half of the 10 colonies in each light x tank group are fed. Feed is actually done by transporting them somewhere else, so it's not as if there is a physically stratified light x food "bin" in each tank. So "bin" in my table is actually just "tank x light x food" = 3 x 3 x 2 = 18 bins, each with 5 fragments. In other words, there are six fixed-factor treatments (food x light groups) with three tanks. So I think you are right: it is tank as the whole plot and light x food as the within-tank subplot. Fragment is the repeated subject and time is the repeated measure. Does that mean that, under random effects in the mixed model platform, I should include tank AND light x food(tank)? I think it is just the latter since light x food(tank) is literally the whole design (subplot nested within whole plot). I have attached a new data table with this script. The last one perhaps didn't save properly.   Your figures look similar to mine. What jumps out is that food seems to have a stronger effect than light. That is the main point of the study, in fact. I had never seen these control and variability plots before, but this could be a good way of showing relative influence of time and tank. I usually use predictor screening or fit-model+random effects for all parameters to get at this in a different way (i.e., which factors drove the most variation in your responses). Actually, if you look at BW alone, light has a stronger effect than food (which is not significant), which is why the "specific growth rate" found in the second table I'm attaching might be the way to go (i.e., relative rate of size increase rather than raw size data).   Two factors correlate so much as to be redundant. I think I didn't include one of them (TLE) in the file I sent. Basically, we looked at size 2-3 different ways since analyzing size of corals is surprisingly disputed. But in the end, they respond similarly to treatment. Fv/Fm is a proxy for photosynthesis, and, while statistically significant, we are talking about a range of 0.7=mean low to 0.74=mean high. I literally wrote a sentence in the manuscript saying that "Although highly significantly influenced by the experimental factors, it was unclear whether such differences are biologically relevant." The photosynthesis people will butcher me because it has become the norm to build huge stories on minuscule changes in this response variable (which I obviously disagree with). In contrast, the weight increased from 2 g to 9 g over 140 days (over 4-fold), which will likely to be the selling point of the whole paper because it's unprecedentedly high.    I have been box-cox transforming the decidedly non-normally distributed data. The color scores are themselves essentially ranks, which I guess makes the color analysis non-parametric.    The second file "linear extension split by time" is really the "meat" because this reflects all the data as a relative or percent change (i.e., the temporal element has been removed). The exception is Fv/Fm, where I just show the file data. One could argue that the actual time by time temporal changes are less important (and will probably end up as an online supplemental figure) vs. the overall effects shown in this second table so I am going to try out the aforementioned split plot with this table (as well as with your pooled summary file). ... View more

Hello,      Thank you for looking at that. I'm attaching the data table with a script since that might explain it better. BW=buoyant weight (coral size), and Fv/Fm and color are two additional response variables. I think I have the repeated measures part set up right (the coral fragment is the subject that is repeatedly assessed over time). I guess I was thinking that the tank was literally split into three plots so I would need to accommodate that somehow in the model since one could argue that the within-tank light levels aren't independent of one another since they are within the same replicate block (i.e., the tank). The fact of the matter is that the feeding effect is so strong that no matter what the final model is, the results hardly change, but I still want to make sure I present it correctly! Maybe what will be important to show is the interaction of tank and light(tank) to ensure that light effects were similar across tanks. I've never set up an experiment like this, so I'm still trying to wrap my head around it! ... View more

Karen,    Thank you for your response. I am going to go through the link you sent. It seems like it could be used for repeated measures on the same subjects over time, so now I just need to figure out how to input it. Also, that's weird that I can't save the script under the red triangle (the first place I looked). Maybe it's because I have been using the beta version of JMP Pro 16. I will try it again on JMP15! ... View more

If I understand correctly, "each pair" is referring to each pair of methods. The mean of method 1 does does differ significantly from that of method 2. In this case, since there are only two methods tested, there is only one pair to test, so I can see why the wording might be confusing since it sounds as if you are doing multiple tests. It is worded that way because, if you had over two methods, you would need to do multiple tests (preferably Tukey's in that case), in which case it would show you the results of "comparisons for each pair using student's t-test." On that note, I like turning on the "connecting letters report" option when you have many comparisons (again, preferably using a more conservative approach like Tukey's) because those letters are, in my field, what usually make it into the publication since they show individual mean differences so clearly and obviously in most figure types. ... View more