At 05:36 PM 10/30/2008, Jennifer Plath wrote (under subject line "Re:
SPSSX-L Digest - 28 Oct 2008 to 29 Oct 2008 (#2008-299)", which I'm
changing here):
>I am using SPSS for some time series correlation analysis. I'm using
>the cross-correlation analysis, but I'm having a hard time
>interpreting the results. The data I am correlating is not based on
>the same measure, one is a sales figure, the other is an industrial
>index. Do I need to convert this data first?
>
>The example below should really not have a lag of any sort, and
>should be highly correlated (and not negatively). That's not what
>this is telling me. I don't understand...
This is one of the cases (why do they happen so often?) when test
data that should have been in multiple columns comes through with
space between columns replaced by two line breaks, so all data is in
one column and almost unreadable.
For the benefit of any who may wish to address this question, here is
the posted table, reformatted into tabular form:
Cross Correlations
Series Pair:PneumaticBookings with NFPAOrders
Lag Cross Std. Error(a)
Correlation
-7 -.284 .080
-6 -.294 .079
-5 -.255 .079
-4 -.339 .079
-3 -.282 .079
-2 -.288 .078
-1 -.305 .078
0 -.119 .078
1 -.272 .078
2 -.278 .078
3 -.245 .079
4 -.332 .079
5 -.283 .079
6 -.303 .079
7 -.295 .080
>a. Based on the assumption that the series are not cross correlated
>and that one of the series is white noise.
>
>Jennifer Plath
>Marketing Research Analyst
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
