nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How can I know what test I just did after using JMP's "Fit X and Y" tool ?

slamer2000

I'm trying to do relationship analysis between age and recycling frequency for a data set. After selecting and filtering some of the ages (since a lot of my respondents didn't give their age for some reason), I used the Fit X and Y tool tool and selected age for the X factor and recycling frequency for the Y factor. Then I clicked Ok and it gave me the below results.

jmp forum 3.jpg

However, JMP doesn't indicate what specific type of test it did for your data. Although it does show on the bottom left a useful table showing Bivariate, Oneway, Logistic and Contingency, it doesn't show the specific test type in the results window. But if I want to do future analysis, how can I know what specific type of test it has done, especially if its a test I haven't heard of? JMP is very different from SPSS in that you didn't really choose your test type. Any suggestions.

Accepted answers

txnelson

The outline box at the top of the displayed output indicates the analysis that has been performed:

Logistic Fit of 17 On average, how often do you recycle on campus: By Age

You can also go to the red triangle and select additional options and analyses that are appropriate for the type of data you provided to the Fit Y by X platform. JMP does not give you a static display environment once you make your initial selections. This is where JMP and SPSS differ. JMP allows you to see your initial analysis, and then continue directly from that point, to do sub analyses and charting and subsetting, without having to go back to the initial dialog box and make changes, and then rerun the analsis all over again.

And finally, if you go to the red triangl, and select "Save Script", you can save the script that will rerun the analysis. One of the places you can save the script, it do the data table you ran the analsis on. This way, you can always have a way of going back and recreating the analysis directly from the data table, whenevery you need to.

All comments

txnelson

The outline box at the top of the displayed output indicates the analysis that has been performed:

Logistic Fit of 17 On average, how often do you recycle on campus: By Age

dale_lehman

TXNelson is correct but I want to add an (unasked for) comment. With 9 data points and such a narrow age range, this logistic regression is almost meaningnless. Actually, I'd say it is meaningless. You would be better served to just make age nominal and do Fit Y by X (you'll get a contingency table and an meaningless chi-squared test). It does make me wonder about your data. If age is missing for many observations, you should not omit those points but code them as "missing" and use them in the analysis (with age as nominal). If the ages of 21-24 are meaningful differences in ages, then the contingency table should suffice for seeing what the relationship looks like, if there is any. Logistic regression, in this case, seems to me to hide the real limited nature of the data you have.

slamer2000

Thanks a lot for your insightful comments. The data here was just from a pilot test for my survey. So if I find that I'm missing a lot of age data, I should include them as missing yes? May I ask how it would benefit to still include this missing data in the analysis since they don't represent anything?

dale_lehman

Missing data represents something and is often informative - indeed, JMP has a number of platforms that use the option "informative missing" for exactly that purpose. Data is usually missing for a reason - if it is because someone forgot to record it, it probably won't help much. But, variables like age (or income) are often missing because the data collection is voluntary (e.g., surveys) and there is a selection going on in terms of who responds and who does not. So, in your case, perhaps a sub-group of people are interested in recycling but do not wish to provide their age (perhaps because they are older than their peers). Their recycling behavior may be significantly different than those that do provide their age. So, omitting them from the analysis potentially misses an important finding. If the ages of those that don't provide age are truly random, then the "age missing" category will not be informative - but otherwise it may be.

slamer2000

I see... it is a bit on the speculative side but it can still be used in research to discuss results. Thanks!