LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (November 2005)Back to main SPSSX-L pageJoin or leave SPSSX-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:   Fri, 25 Nov 2005 15:13:49 -0600
Reply-To:   "Peck, Jon" <peck@spss.com>
Sender:   "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:   "Peck, Jon" <peck@spss.com>
Subject:   Re: RECODE glitch?
Comments:   To: Marta García-Granero <biostatistics@terra.es>
Content-Type:   text/plain; charset="iso-8859-1"

Recode is designed to process the specifications in the order they are given. There is no requirement that they not overlap. In fact overlapping intervals can be quite convenient (lo - 25) (lo - 50)

etc.

This is useful in dealing with open/closed intervals.

And, of course, the gui generates syntax according to the order in which the intervals are specified.

Regards, Jon Peck

-----Original Message----- From: SPSSX(r) Discussion on behalf of Marta García-Granero Sent: Fri 11/25/2005 12:46 PM To: SPSSX-L@LISTSERV.UGA.EDU Subject: [SPSSX-L] RECODE glitch? Hi everybody

I've been faced recently with the following "SPSS surprise":

My students had recently an exam about SSPSS usage. One of the exercises consisted in recoding a continuous variable into 3 categories and finding the percentages of every category:

* Part of the dataset they used *. DATA LIST FREE/height weight (2 F8.0). BEGIN DATA 167 81 181 82 164 78 179 80 159 60 163 70 167 57 159 54 170 59 154 61 162 49 162 77 173 95 177 63 164 85 175 67 184 116 155 51 156 85 160 86 161 73 171 86 159 76 188 90 185 108 167 62 167 77 176 63 165 69 173 67 159 48 194 86 180 81 170 79 167 59 182 106 187 116 173 63 178 97 170 48 165 85 167 65 164 72 168 73 152 75 150 70 173 63 169 77 158 39 165 75 END DATA. COMPUTE bmi=weight/((height/100)**2).

* They to recode BMI (Body Mass Index) according to the following cut-points: =<25; 25-30; >30 *.

* This was the correct way: *. RECODE bmi (Lowest thru 25=0) (25 thru 30=1) (30 thru Highest=2) INTO obese . VALUE LABEL obese 0'No' 1'Overweight' 2'Obese'. FREQUENCIES VARIABLES=obese.

* One of my students (they use the GUI), made a mistake in the first category; she erased it and defined it again. When she hit OK, the code she actually run was: (see the difference?)*.

RECODE bmi (25 thru 30=1) (30 thru Highest=2) (Lowest thru 25=0) INTO obese . VALUE LABEL obese 0'No' 1'Overweight' 2'Obese'. FREQUENCIES VARIABLES=obese.

The 2 methods gave different results (check by yourself), at least on SPSS 11 (& 12, I have just checked). What do you think of that? I'm careful enought to select the correct order for the categories, but novice users might have problems (I needed a bit of time to spot the problem using the GUI), both RECODings look the same...

Awfully tired (what a week I've had!) Happy week end!

Marta


Back to: Top of message | Previous page | Main SPSSX-L page