Nominal variable in ANOVA report

[#1]

I have 28 music clips. 14 of them are composed in major mode and rest 14 in minor mode. I am looking at emotional response for the music and preference for the music against the mode of the music. 

I have two questions:

1) Should Mode (Major and Minor) be put as nominal variable or ordinal variable?

2) When I put Mode as a nominal variable with value labels - Major = 0, Minor = 0, and then fit the model the Parameter estimate table gives the p value only for Major mode and not the minor one. Why is it so?  I have attached the file. 

[#2]

1) You can create a column with a "Character" type and a "Nominal" modeling type for Minor and Major in your data table.
2) Then, in the Fit Y by X, put your column "Major/Minor" as your X, and your response as Y.

That should work :)

[#3]

Hi Victor!

Thank you so much for your repsonse.

1) The mode is already in 'Character Type' with modelling type as Nominal.

2) It works with Fit Y by X. However, I want to use Fit Model since I have another variable which is the tempo of the music clips (continous variable). I am looking at the main effects and the interaction effects. When I look at the main effect for the mode, the report only shows results for major and not for minor. Why is it so?

[#4 · Accepted solution]

Hi !

I understand better your situation.
Concerning the second point, if you launch the "Fit model" platform with a Standard Least Square personality, you will have a regresssion graph showing for each category (Major/Minor) the regression line (see photo below/attached).

You can also have a look on how significant are main effects and interaction on the "Effect Summary" in order to take into account (or not) the interaction in your analysis. You can also use the Graph Builder in order to see emotional response in function of your inputs, and showing details about regression coefficient in order to help you.

Finally, if you think it may be good to have a look at several parameters in order to see differences in case of interaction, I would give a try to the "Variability / Attribute Gauge chart" in "Analyze", then "Quality and Process". You could be able to see if several parameter seems to influence the response in a particular way. You can do it directly, or create a "class" column (categorical) for your continuous input tempo in order to group similar tempo by categorical groups.

I think from what I see I'm out of other options, but by combining these ideas, you may be able to have a clear view on your topic :)

[#5 · Accepted solution]

I do not think that there is any implied or meaningful order to minor versus major so nominal seems the most reasonable choice for modeling type.

The Effect Tests reports the evidence for the entire model term. The Parameter Estimates to which you refer reports evidence for individual parameters. The coding of the categorical factor levels is such that they must sum to zero for the term. You have two levels for Mode so the parameter for the Minor level is the negative of the estimate for the Major level. You can click the red triangle at the top and select Expanded Estimates to see the estimates for all the levels at once.