Date:         Fri, 13 Dec 2002 10:14:58 -0500
Reply-To:     John Larsen <johnlar@tfn.net>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         John Larsen <johnlar@tfn.net>
Subject:      Re: Sort Cases (by two variables)
In-Reply-To:  <20021213135214.18011.qmail@web14511.mail.yahoo.com>
Content-Type: text/plain; charset=us-ascii; format=flowed

Yep (actually looked at this file as opposed to my previous email)
they are sorting by thier true values as opposed to the display values,
the true values are
1.88
1.88
1.90
1.90
1.94

Which all round to 1.9 for the display, to fix this you can either
change the format of the field to 6.2 or you could
compute newval=((rnd(fertility*10))/10).
execute.

which creates a new variable rounded to that digit

Andrei Boucher wrote:

>My understanding is that when sorting two variables, precedent is taken by the first variable. Since World 95 is a small database, you may like to check your answer in Excel.
>Andrei
> Michiel van Lieshout <m_vanlieshout@chello.nl> wrote:Hello everyone.
>
>I am a dutch student at the open university. My teacher could not help me with the next problem .
>
>In World95.sav I tried to sort simultaneously fertility (Ascending) and lit_fema (Descending).
>In Syntax Editor:
>SORT CASES BY
>fertility (A) lit_fema (D) .
>
>Fertility is perfectly ascending, but in lit_fema I get orders like the following:
>For cases 29 to 33 fertility is 1.9 .
>Then I expect lit_fema (%females who read) to be descending. But I get the following:
>case29 -> 100
>case30 -> 84
>case31 -> 100
>case32 -> 93
>case33 -> 98
>
>Can anybody tell me what goes wrong?
>
>Michiel van Lieshout
>
>
>---------------------------------
>Post your free ad now! Yahoo! Canada Personals
>
>
>