Performing one-way (or multiway) ANOVA on many variables without displaying graphs using JSL

[#1]

Hello,

I would like to run a one-way ANOVA (and eventually multiway) on a table for a very large list of endpoints followup by a post-hoc test such as Dunnett's.

For a large list of endpoints, displaying each graph is both slow and unpractical if one only needs the aggregated p-Values.

Is there a way in JSL to only retrieve those values by bypassing the graph creation?

[#2 · Accepted solution]

One can easily run the platform in an invisible mode, and then just output the results.  See below fhe script and the re

Names Default To Here( 1 );
// Open Data Table: big class.jmp
// → Data Table( "big class" )
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

colNames = dt << get column names( continuous );

ow = Oneway( invisible, Y( Eval( colNames ) ), X( :site ), Means( 1 ), Mean Diamonds( 1 ) );

Report( ow[1] )["Analysis of Variance"][Table Box( 1 )] << make combined data table;

[#3]

Hi Jim,

I did not think about that way. It did the trick.

Thanks a lot!

Best,

Sebastien

[#4]

 I also thing you should look into using the Response Screening platform.  It provides you the PValues plus additional stats.  It also has an interactive table script that will display the scatterplot for any selected variables.

Names Default To Here( 1 );

// Open Data Table: semiconductor capability.jmp
// → Data Table( "semiconductor capability" )
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

colNames = dt << get column names( continuous );

Response Screening( 
	Y( Eval( colNames ) ), 
	X( :SITE ), 
	PValues Table on Launch( 1 ) 
	);