multiway frequency analysis, Canonical correlation

#1
I have a set of survey report of my college students who decided to change to other schools. There were 3,000 students surveyed over the 10 year period.

There are 12 schools in our university. Each school is coded with a number, other is coded as 13.

1. pharmacy
2. engineering
3. management...
.
.
.
.
12. nursing
13. other

The survey questionnaire is yes or no.

Like:
Do you find the courses too difficulty

Yes
No

The information this survey collected about the students are their gender and their major in our school and future school.

I would like to find what this set of data tells me. I did frequency analysis, and also separated male and female students frequency on each question.

Here are my thoughts which I would like to use two steps to conduct analysis.

1. Multiway Frequency Analysis

Would multiway frequency analysis be good to check exactly which major of students we are more likely to say, for example, YES, to the question if the courses are too difficult?

Would I be able to make a 13X13 table and see the traffic of specific major group of students "traveling" to, say, "liberal arts" major???

2. Canonical Correlation

There are 40 "yes" and "no" question, measuring from if courses are difficulty, if students get help, if students attend classes... and such...

Could I divided these questions into two groups and use Canonical correlation. If so, how could I go by doing it since it is coded as dichotomous way.

How could find a dichotomous correlation syntax from SPSS?

Thank you in advance for your help
 
#2
I am taking a multivariate statistic course. I am hoping to use this survey for my course final project. If this group helps its own university student, then I am sorry.