Bar Graphs in Graph Builder - new in JMP19 ...

[#1]

In Graph Builder, users do not need to drag a value onto the y-axis in order to create a bar graph.
Without specifying a Y value, JMP will count -- or calculate an aggregate value.
Which one? For some of the settings, this depends on the version of JMP being used.

comparison JMP18 vs. JMP19:

Names Default to Here(1);
dt = Open( "$SAMPLE_DATA/Big Class Families.jmp" );
Graph Builder(
	Variables( X( :sex ), Overlay( :age ) ),
	Elements( Bar( X, Legend( 3 ), Summary Statistic( "Median" ) ) )
);

Median is great to get rid of outliers, so maybe use it as a default setting in the preferences?
Think twice - and decide whether you would prefer meaningful results for 'values' or for 'counts'.

[#2]

Lots of those plots look the same, what's the summary of your findings here?  Is this only an issue if you override the default Summary Statistic in the GB settings?

[#3]

Yes, somehow.
If a user has the 'Mean' setting selected in their preferences, it is very likely that they won't notice any difference.
They start with 'Mean' and will get 'Mean' values, or counts if no value is dragged onto the Y-axis. Great!
Perhaps they will use a different setting and benefit from the new options (besides median, minimum and mode, other calculations have changed as well. I suppose there is a benefit of te new mode - maybe for Cumulative Sum?]

On the other hand:

A user who selected 'Median' in the preferences needs to be aware of the change - in order to switch to 'Mean'(N or Sum) for counts.

They also have to change all the created scripts because the preferences are automatically added to them. Even if the setting did not have an effect for counts in the past, it does now in JMP19 ...

Unfortunately for me: I was brave enough to define 

Summary Statistic( "Median" )

as standard in the default script which new JMP users get when they start with JMP.
So, let's see how many of my colleagues will notice, wonder, find out ... or find the post.

[#4]

Besides mitigating the impact for users with manually changed preferences - why the change? 
Where is the benefit? 

my 1st guess: to better teach/understand the differences between weight and freq - Freq and Weight: better documentation 

(e.g. the impact on median).

let's try ...

Names Default to Here(1);
dt = Open( "$SAMPLE_DATA/Big Class Families.jmp" );

new column ("freq", formula(random integer(10)));
new column ("weight_", formula(random uniform (2)));
Graph Builder(
	Variables( X( :sex ), Overlay( :age ), Frequency( :freq ) ),
	Elements( Bar( X, Summary Statistic( "Max" ) ) )
);

Doesn't seem to be the case -  even "max" is 1 for freq = 1 ... 10:

Some of the aggregations [min, max, median ...] seem to calculate the result on row level ,

not on on "bar" level like mean, cumulative %, percent of .....

Plots with freq=1 and random freq. I added colors to the background to highlight the differences.