SPSSX Discussion

Re: sorting correlations

Classic

List

Threaded

1 message

John F Hall

Re: sorting correlations

Bruce

Perfect. Mille fois merci.

John

GET FILE="H:\F drive backup\Research1a\4 Surveys\Quality of Life\QLGB1971-75\QL3\districtuk.sav".

DATASET NAME raw.

* OMS.

DATASET DECLARE varinfo.

OMS

/SELECT TABLES

/TAG = "varinfo"

/IF COMMANDS=['Sysfile Info'] SUBTYPES=['Variable Information']

/DESTINATION FORMAT=SAV NUMBERED=TableNumber_

OUTFILE='varinfo' VIEWER=YES.

SYSFILE INFO raw.

OMSEND TAG = "varinfo".

DATASET ACTIVATE varinfo.

DELETE VARIABLES TableNumber_ TO Subtype_ Position MeasurementLevel TO MissingValues.

* Change Var1 to A8 and rename it to VARNAME_ to make it match name & format in the next file.

ALTER TYPE Var1 (A8).

RENAME VARIABLES (Var1 Label = VARNAME_ varlabel).

DATASET ACTIVATE raw.

DATASET DECLARE corrmat.

CORRELATIONS var235 TO var250

/MATRIX = OUT("corrmat").

DATASET ACTIVATE corrmat.

SELECT IF ROWTYPE_ EQ "CORR" AND VARNAME_ NE "var250".

EXECUTE.

MATCH FILES FILE = "varinfo" /FILE= "corrmat" /BY VARNAME_.

EXECUTE.

DATASET NAME merged.

DATASET CLOSE varinfo.

DATASET CLOSE corrmat.

DATASET ACTIVATE merged.

SELECT IF ROWTYPE_ EQ "CORR".

COMPUTE absr = ABS(var250).

SORT CASES by absr(D).

VARIABLE LABELS varlabel "Variable".

FORMATS var250 (F8.4).

SUMMARIZE

/TABLES=varlabel var250

/FORMAT=LIST NOCASENUM

/TITLE='Correlations with District'

/MISSING=VARIABLE

/CELLS=NONE.

Correlations with District
	varlabel Variable	var250 C1: Satisfaction with district
1	B1o: PEOPLE	.6778
2	B1j: APPEARANCE	.6699
3	B1n: POLLUTION	.5605
4	B1k: SAFE at NIGHT	.5444
5	B1d: NOISE	.5167
6	B1f: CRIME	.4948
7	B1i: TRAFFIC	.3981
8	B1a: ROADS,PATHS	.3612
9	B1m: NEAR FRIENDS	.3492
10	B1g: SCHOOLS	.3273
11	B1h: PARKS	.3261
12	B1e: ENTERTAINMT	.3108
13	B1l: NEAR FAMILY	.2875
14	B1c: SHOPS	.2854
15	B1b: TRANSPORT	.2257

From: Bruce Weaver <[hidden email]>
Sent: 22 July 2020 17:36
To: John F Hall <[hidden email]>
Subject: Re: sorting correlations

Here's a souped up version of what I posted previously, John. It uses SYSFILE INFO in conjunction with OMS to grab the variable labels for use in the final table of (sorted) correlations. NPR as I like to say. I.e., No Python Required.* I did not take time to comment extensively, but hope things are relatively clear nevertheless!

* Jon P will never agree with me on this, but I really dislike the way IBM has gone to implementing so much new functionality via Python and R programmability rather than fixing and adding native SPSS commands.

Cheers,

Bruce

On Wed, Jul 22, 2020 at 4:31 AM <[hidden email]> wrote:

Jon, Bruce
Here's an extract from the *.sav file for you to test your syntax
John

Bruce Weaver
Research Associate, Centre for Research on Safe Driving, Lakehead University
Assistant Professor of Biostatistics, Human Sciences Division
Northern Ontario School of Medicine, West Campus
955 Oliver Road, MS-2006, Thunder Bay, ON P7B 5E1

Tel: 807-346-7704 Fax: 807-766-7362
Email: [hidden email]
Web: http://sites.google.com/a/lakeheadu.ca/bweaver/

===================== 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