Smoother line between to examine a relationship between two categorical variables?Hello Jmp-folks:
I love the feature of smoothing line on graph builder. I see that this smoother line is available in categorical variables
which I found particularly interesting.
When you hover over the icon for smoother line it says it is best for continuous variables
Does it make any sense to use it to examine relationship of three categorical variables?
If yes how do you interpret this graph?
Are there any examples available?
The variables are three:
academic upreparedness: likely or not likely
gpa performance
and outcome attrition (dropout or not)
Thanks to the amazing "Save Script option"
here is the script:
Graph Builder(
Variables( X( :GPA ), Y( :Academic_Unprepared ), Overlay( :Attrition ) ),
Elements( Smoother( X, Y, Legend( 7 ) ) )
);
Re: Smoother line between to examine a relationship between two categorical variables?Makes so sense! Thank you for this great explanation
Re: Smoother line between to examine a relationship between two categorical variables?Hi @triunk,
Sometimes this can be useful, especially when your X variable is ordinal (has a natural ordering). What you're seeing is a reflection of the proportion of observations in each level of Y for each level of X. For example, here's such a plot using the Consumer Preferences sample dataset:
We can easily see that the proportion reporting school-aged children changes across the lifespan. But, without a Y-axis this only tells us generally about the relative proportionality. Seen as a mosaic plot we get much more information:
So, when would the smoother be useful? One case might be when you're overlaying groups and your main point is to simply demonstrate a difference in the proportions in each level of Y for each level of X for those overlain groups. For example, again using the Consumer Preferences dataset, what if we look at school age children by age, but separately for those people who do and do not come from a large family: 
I've turned off the points so we can just focus on the smoother. In this can we can see that people who come from a large family seem to have school-age children more often, and do so earlier in their life. There are certainly other ways to show this, but you can see the smoother does tell the story (as long as you know what you're seeing).
@julian
