Multivariate analysis

[#1]

There are two files attached;  'Data on female catches EP' and '06-22 all reps Catches_per_vessel_per_day v5'.

In both files you will find a saved script for Analyze>fit Model. For the first file the analysis produces tables and graphs of the variables whilst in the second file the analysis produces tables but no graphs. Why is this?

[#2]

With just a quick look, your data tables are not the same and the fit model script from the one table can't be run the same on the 2nd table as the response variables do not exist in the second table. The following is the script for the fit model in the first table:

Y( :"Catch/pot total"n, :"Catch/pot small"n ),

Effects(

:"East-West"n, :"Inshore-Offshore"n, :"East-West"n * :"Inshore-Offshore"n

),

Catch/pot total is not in the second table

[#3]

'Catch/pot total' and 'Catch/pot' are measuring the same thing but in the second table the heading is derived from the IBM output. If I change the variable 'Catch/pot' in the first table to 'Catch/pot total' in the first table it makes no difference to the output. I presume that the lack of graphs deriving from the analysis of the second data set has something to do with the relationship between the variables?

[#4 · Accepted solution]

I did not run the scripts, but just looked at them. For the Data on females catches EP file, you are using Standard Least Squares with an Emphasis of "Effect Leverage". For the other file you are using Standard Least Squares, but the Emphasis is Minimal Report. Change that to "Effect Leverage" and the plots will appear by default again. Note that even with the Minimal Report all of the graphs are available from the red popup menu.

[#5]

Thanks for the help - much appreciated. I'm not sure how I got to choose 'Effect leverage' for the first data set but by applying it to the second the graphs appeared as you said they would.

[#6]

The default choice for Emphasis in the Fit Model launch dialog window is not constant. There is a complex flow of comparisons to determine the initial choice. I usually over-ride it and pick Minimal Report. I am not suggesting that this choice is the best one for you. I only want to point out that you should learn what the choices mean and pay attention to the choice before you click Run so that you start the regression the way that you want.