|
|
1. I would much welcome advice on how to overcome the problems I have
encountered in trying to conduct canonical correlation analysis of survey
data in SPSS v14, using the built-in macro, on my PC (Windows XP Home
Edition).
2. The data consist of 10 IVs (Set1) and 4 DVs (Set2). More specifically,
the IVs comprise mainly binary (dummy) predictor variables but also
include some numerical (continuous) ones, while the DVs are all
continuous. 6 of the IVs are composite variables and the remainder are
singular; all DVs are singular.
3. Drawing on specimen syntax, including SPSS's own, the procedure and
sequence I have used were as follows:
Open data file/Open new syntax window/ Type macro into new window/
Run /All. The macro together with all relevant
variables took this form:
INCLUDE `Canonical correlation.sps'.
CANCORR SET1=MDAANDTFT MGTCOM COMPLAN EXSPRED VISITS NOQTSC CommAgt
OrgExp4 Both14J05 PrcCPMB/
SET2=TOStrTnd Sharto StrengthTrend27Jun05 ExPfyOall/.
I have also tried running this macro with the full path specified:
C:\Program Files\SPSS\Canonical...and likewise without the final forward
slash. Neither has had any effect.
4. Run with either the full or the curtailed path, the macro generates
only these 2 warning and 2 error messages:
Warning # 206 in column 9.Text: An invalid character has been found on a
command.
Warning # 208 in column 35.Text: A text string is not correctly enclosed
in quotation marks on the command line. Literals may not be continued
across command lines without the use of the continuation symbol `+'.
Error # 32 in column 9.Text: Invalid syntax of file reference. Enclose the
file name in quotes. This command not executed.
Error # 1. Command name: CANCORR. The first word in the line is not
recognised as an SPSS command. This command not executed.
5. Assuming the references to columns 9 and 35 refer to columns in the
data file, these columns (both containing purely numerical data) are
respectively the two ( SET2) DVs TOStrTnd and StrengthTrend15Jun05.
Enclosing one or both of them in single or double quotes has no effect on
the outcome. Nor does inserting a `+' symbol half way along the list of
variables in SET2 validate any wrapping into the next command line that
may be taking place (Warning 208).
I shall be much obliged for all advice on how to overcome these problems.
While I would have liked to make the query much shorter, I did not feel I
could do so without leaving out detail that might prove vital to the
solution.
With advance thanks =
Owen Murphy
|
Locked
