Problem with SPSS arithmetic
Locked
 
|
I'm a rank newbie in this group and don't have much experience at all with SPSS. I've just come across this, and it bothers me a lot:
I'm doing a one-way ANOVA and estimating some contrasts. For one of these, I estimate the value of the contrast as ĉ = 2*( 4.6 + 4.4 + 4.8 ) – 3*( 2.1 + 2.1 ) = 27.6 - 12.6 = 15 (by hand). SPSS gets ĉ = 14.4. Now admittedly I'm pretty damn bad at arithmetic for someone with two degrees in math, but I've checked my hand calculations about six times. Have any of you ever seen something like this? Regards, Joe S ===================== 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 |
Re: Problem with SPSS arithmetic
Administrator
|
Two possibilities come to mind: 1. You show the means to one decimal. Are you introducing rounding error? 2. Are the sample sizes all equal? If not, you may want to check the SPSS algorithms to see how it handles that. (I don't know what the answer is, but I'd be checking to see if it changes things.)
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
Problems with the 'SORT' menu
|
|
All, Has anybody here noticed a problem with the 'sort cases' menu? If I have been using PASW for some time, having opened, closed many files and run several procedures, the software seems to get 'tired' of sorting. I might pull up the sort cases menu and try a simple sort. I can pull the variable over and all of the buttons (OK, Paste, Reset, Cancel, Help) appear--none of them greyed out. I click on 'OK' and nothing happens. The image of the button appears to get 'pushed in' like it should but it just stays there. The menu box doesn't close and the sort does not occur. In the same instances, the Paste button doesn't seem to work either. However, the Reset, Cancel, and Help buttons do work. If I write out a sort statement and run it in syntax, the sort runs fine. Once this occurs, the sort menu is useless for any other dataset window until I close eveything and reopen PASW. This strange behavior doesn't seem to occur with any other PASW menu window, just Sort Cases. It occurs with 17.0.2 networked on a big Dell desktop and on an individual license of 18 on my Lenovo laptop. Has anybody else seen this? I don't recall seeing this in the betas of either version. *************************************************************************************************************************************************************** Mark A. Davenport Ph.D. Senior Research Analyst Office of Institutional Research The University of North Carolina at Greensboro 336.256.0395 [hidden email] 'An approximate answer to the right question is worth a good deal more than an exact answer to an approximate question.' --a paraphrase of J. W. Tukey (1962) |
Re: Problems with the 'SORT' menu
|
|
Marc That’s correct Mark. It has been frustrating to deal with the
sort function as it forces you to quit the program and start again. Any resolutions
on that???? Gurvinder Khaneja Gurvinder Khaneja, Ed.D Executive Director, Institutional Research and Planning Passaic County Community College Phone: 973.684.6741 Fax: 973.684.1925 From: SPSSX(r) Discussion
[mailto:[hidden email]] On Behalf Of Mark A Davenport MADAVENP
|
Re: Problems with the 'SORT' menu
|
|
In reply to this post by Mark A Davenport MADAVENP
Mark et al. I noticed this a couple of times when beta testing 17 but it is extremely difficult to replicate; in fact every time I tried to re-create it I failed to provide
a definitive set of commands / steps so I assumed it was something I was doing
J I think it is due to a combination of file size and sorts which hardly change the file – BUT – as I said trying to produce evidence and replicate it reliably
proved to be too hard and life is too short. I just accepted that it happens and just shut down and restart. Best Wishes John S. Lemon Directorate of Information Technology (DIT) -
University of Aberdeen Edward Wright Building: Room G51 Tel: +44 1224 273350 Fax: +44 1224 273372 From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of Mark A Davenport MADAVENP
The University of Aberdeen is a charity registered in Scotland, No SC013683. |
Re: Problems with the 'SORT' menu
|
|
I may work on a project for several hours and I often have multiple sets open done multiple merges so I use sort cases a lot. A LOT! But, often the sort does not produce large changes in the file. So, you might be on to something. Sadly, as I work on this particular yearly project, it becomes a significant issue. I was getting it last year about the same time of year (after beta testing was over). It does seem to occur only when I have been 'abusing' data for extended periods. As I abuse data with great regularity in the fall, I am having a pretty consistent problem with it now. Thanks for the input. *************************************************************************************************************************************************************** Mark A. Davenport Ph.D. Senior Research Analyst Office of Institutional Research The University of North Carolina at Greensboro 336.256.0395 [hidden email] 'An approximate answer to the right question is worth a good deal more than an exact answer to an approximate question.' --a paraphrase of J. W. Tukey (1962)
Mark et al. I noticed this a couple of times when beta testing 17 but it is extremely difficult to replicate; in fact every time I tried to re-create it I failed to provide a definitive set of commands / steps so I assumed it was something I was doing J I think it is due to a combination of file size and sorts which hardly change the file – BUT – as I said trying to produce evidence and replicate it reliably proved to be too hard and life is too short. I just accepted that it happens and just shut down and restart. Best Wishes John S. Lemon Student Liaison Officer Directorate of Information Technology (DIT) - University of Aberdeen Edward Wright Building: Room G51 Tel: +44 1224 273350 Fax: +44 1224 273372 Diary ( Free / Busy ) From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mark A Davenport MADAVENP Sent: 30 October 2009 14:28 To: [hidden email] Subject: Problems with the 'SORT' menu All, Has anybody here noticed a problem with the 'sort cases' menu? If I have been using PASW for some time, having opened, closed many files and run several procedures, the software seems to get 'tired' of sorting. I might pull up the sort cases menu and try a simple sort. I can pull the variable over and all of the buttons (OK, Paste, Reset, Cancel, Help) appear--none of them greyed out. I click on 'OK' and nothing happens. The image of the button appears to get 'pushed in' like it should but it just stays there. The menu box doesn't close and the sort does not occur. In the same instances, the Paste button doesn't seem to work either. However, the Reset, Cancel, and Help buttons do work. If I write out a sort statement and run it in syntax, the sort runs fine. Once this occurs, the sort menu is useless for any other dataset window until I close eveything and reopen PASW. This strange behavior doesn't seem to occur with any other PASW menu window, just Sort Cases. It occurs with 17.0.2 networked on a big Dell desktop and on an individual license of 18 on my Lenovo laptop. Has anybody else seen this? I don't recall seeing this in the betas of either version. *************************************************************************************************************************************************************** Mark A. Davenport Ph.D. Senior Research Analyst Office of Institutional Research The University of North Carolina at Greensboro 336.256.0395 [hidden email] 'An approximate answer to the right question is worth a good deal more than an exact answer to an approximate question.' --a paraphrase of J. W. Tukey (1962)
|
Re: Problems with the 'SORT' menu
|
|
In reply to this post by Mark A Davenport MADAVENP
What happens if you run the SORT using syntax? From: SPSSX(r) Discussion
[mailto:[hidden email]] On Behalf Of Mark A Davenport MADAVENP
|
FW: Problem with SPSS arithmetic
|
|
In reply to this post by Sharp, Joe P.
More info:
-- the individual data values are counts; -- sample size = 5 for each treatment. Roundoff error is not the problem. ________________________________________ From: Sharp, Joe P. Sent: Friday, October 30, 2009 12:10 AM To: [hidden email] Subject: Problem with SPSS arithmetic I'm a rank newbie in this group and don't have much experience at all with SPSS. I've just come across this, and it bothers me a lot: I'm doing a one-way ANOVA and estimating some contrasts. For one of these, I estimate the value of the contrast as ĉ = 2*( 4.6 + 4.4 + 4.8 ) – 3*( 2.1 + 2.1 ) = 27.6 - 12.6 = 15 (by hand). SPSS gets ĉ = 14.4. Now admittedly I'm pretty damn bad at arithmetic for someone with two degrees in math, but I've checked my hand calculations about six times. Have any of you ever seen something like this? Regards, Joe S ===================== 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 |
Re: Problems with the 'SORT' menu
|
|
In reply to this post by ViAnn Beadle
Forwarding to the list. Sounds to me like there is a bug in the
UI which persists settings from the last invocation of the dialog. At least
users can get around this problem via syntax. From: Mark A Davenport
MADAVENP [mailto:[hidden email]]
|
Re: FW: Problem with SPSS arithmetic
|
|
In reply to this post by Sharp, Joe P.
Hi Joe
Sharp, Joe P. WROTE: > More info: > > -- the individual data values are counts; > Do you mean they are integers? > -- sample size = 5 for each treatment. > If the individual values are truly integers, then there is something wrong with some of the means you report, like 2.1, since there is no way you can get a mean with value 2.1 when you divide and integer value by 5. HTH, Marta GG > Roundoff error is not the problem. > ________________________________________ > From: Sharp, Joe P. > Sent: Friday, October 30, 2009 12:10 AM > To: [hidden email] > Subject: Problem with SPSS arithmetic > > I'm a rank newbie in this group and don't have much experience at all with SPSS. I've just come across this, and it bothers me a lot: > > I'm doing a one-way ANOVA and estimating some contrasts. For one of these, I estimate the value of the contrast as ĉ = 2*( 4.6 + 4.4 + 4.8 ) – 3*( 2.1 + 2.1 ) = 27.6 - 12.6 = 15 (by hand). > > SPSS gets ĉ = 14.4. > > Now admittedly I'm pretty damn bad at arithmetic for someone with two degrees in math, but I've checked my hand calculations about six times. > > Have any of you ever seen something like this? > > Regards, > > Joe S > > ===================== > 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 > > -- For miscellaneous SPSS related statistical stuff, visit: http://gjyp.nl/marta/ ===================== 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 |
Ver. 17.0.1 Date calculation error
|
|
In reply to this post by Mark A Davenport MADAVENP
Good Morning
list, I would like to
know if anyone else has experienced this issue or whether I am doing something
wrong which is resulting in the following: I am trying to calculate three earlier dates
based on the difference between some given future dates and a value indicating number
of days. However, within each
column of dates I get the same values, and this happens for all of the four
dates regardless of the number of days (numdays) I subtract from the given
dates (date1, date2,...) which are all different(?) My code is below, and if anyone can point
out if there is anything incorrect with it, I would appreciate it very much. This is what the resulting data look like:
Here is the code: date1 through date4 vary and
so does numdays(?)! compute
dt1=(date1-(numdays * 86400)). compute
dt2=(date2-(numdays * 86400)). compute
dt3=(date3-(numdays * 86400)). compute
dt4=(date4-(numdays * 86400)). formats
dt1 dt2 dt3 dt4 (adate10). Any help, advice or suggestions would be very much
appreciated. TIA Mike |
Re: Ver. 17.0.1 Date calculation error
|
At 12:25 PM 10/30/2009, Roberts, Michael wrote:
>I am trying to calculate three earlier dates based on the difference >between some given future dates and a value indicating number of >days. However, within each column of dates I get the same values, >and this happens for all of the four dates. My code is below. Could you post the input data that gives the results you get? We can't do much without it. I don't see anything obviously wrong with your code, on simple inspection. ===================== 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 |
Re: Ver. 17.0.1 Date calculation error
|
|
Below is the input data:
numdays date1 date2 date3 dt1 dt2 dt3 dt4 34339 1/1/2099 1/2/2099 1/3/2099 12/29/2004 12/28/2004 12/27/2004 12/26/2004 34355 1/17/2099 1/18/2099 1/19/2099 12/29/2004 12/28/2004 12/27/2004 12/26/2004 34356 1/18/2099 1/19/2099 1/20/2099 12/29/2004 12/28/2004 12/27/2004 12/26/2004 34358 1/20/2099 1/21/2099 1/22/2099 12/29/2004 12/28/2004 12/27/2004 12/26/2004 -----Original Message----- From: Richard Ristow [mailto:[hidden email]] Sent: Friday, October 30, 2009 1:09 PM To: Roberts, Michael; [hidden email] Subject: Re: Ver. 17.0.1 Date calculation error At 12:25 PM 10/30/2009, Roberts, Michael wrote: >I am trying to calculate three earlier dates based on the difference >between some given future dates and a value indicating number of >days. However, within each column of dates I get the same values, >and this happens for all of the four dates. My code is below. Could you post the input data that gives the results you get? We can't do much without it. I don't see anything obviously wrong with your code, on simple inspection. ===================== 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 |
Re: Ver. 17.0.1 Date calculation error
|
|
At 02:09 PM 10/30/2009, Roberts, Michael wrote:
Below is the input data:numdays date1 date2 date3 34339 1/1/2099 1/2/2099 1/3/2099 34355 1/17/2099 1/18/2099 1/19/2099 34356 1/18/2099 1/19/2099 1/20/2099 34358 1/20/2099 1/21/2099 1/22/2099 You didn't give a DATE4, but that's OK; I'll do it for 1-3 only. You had complained that within each column of dates I get the same values, and this happens for all of the four dates regardless of the number of days I subtractas in this example you posted: dt1 dt2 dt3 dt4 12/29/2004 12/28/2004 12/27/2004 12/26/2004 12/29/2004 12/28/2004 12/27/2004 12/26/2004 12/29/2004 12/28/2004 12/27/2004 12/26/2004 12/29/2004 12/28/2004 12/27/2004 12/26/2004 Using the test data you just sent, I get compute dt1=(date1-(numdays * 86400)). compute dt2=(date2-(numdays * 86400)). compute dt3=(date3-(numdays * 86400)). formats dt1 dt2 dt3 (adate10). LIST. List |-----------------------------|---------------------------| |Output Created |30-OCT-2009 15:45:04 | |-----------------------------|---------------------------| numdays date1 date2 date3 dt1 dt2 dt3 34339 01/01/2099 01/02/2099 01/03/2099 12/26/2004 12/27/2004 12/28/2004 34355 01/17/2099 01/18/2099 01/19/2099 12/26/2004 12/27/2004 12/28/2004 34356 01/18/2099 01/19/2099 01/20/2099 12/26/2004 12/27/2004 12/28/2004 34358 01/20/2099 01/21/2099 01/22/2099 12/26/2004 12/27/2004 12/28/2004 Number of cases read: 4 Number of cases listed: 4 So within each column, all the calculated dates are the same. It looks like that's the correct answer: that the 'numdays' values differ between rows exactly as much as the input dates differ. You'll have to think about what that means, given what the data is about. -Best of luck, Richard ============================= APPENDIX: Test data, and code ============================= DATA LIST LIST/ numdays date1 date2 date3 (F5, 3ADATE10). BEGIN DATA 34339 1/1/2099 1/2/2099 1/3/2099 34355 1/17/2099 1/18/2099 1/19/2099 34356 1/18/2099 1/19/2099 1/20/2099 34358 1/20/2099 1/21/2099 1/22/2099 END DATA. compute dt1=(date1-(numdays * 86400)). compute dt2=(date2-(numdays * 86400)). compute dt3=(date3-(numdays * 86400)). formats dt1 dt2 dt3 (adate10). LIST. ===================== 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 |
Re: Ver. 17.0.1 Date calculation error
|
|
Richard, Thank you for
checking the code, and for your response.
After posting and further examination, I noticed the incremental changes
in the dates and number of days was the cause of the date being the same! Looks like I have to do some additional
digging into my data J Mike From: Richard Ristow [mailto:[hidden email]] At 02:09 PM 10/30/2009, Roberts, Michael wrote: Below is the input data:
numdays date1 date2 date3 within each column of dates I get the same values, and this
happens for all of the four dates regardless of the number of days I subtract as in this example you posted: |
Re: Problems with the 'SORT' menu
|
|
In reply to this post by ViAnn Beadle
Hi
Mark,
I spent some
time today trying to replicate this on Release 18 and was not able to do so, but
I've opened a support case and will contact you directly. If anyone is able to
provide a consistent replication scenario, we would be most
appreciative.
Best
regards,
Dave
David Nichols, Ph.D.
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of ViAnn Beadle Sent: Friday, October 30, 2009 11:05 AM To: [hidden email] Subject: Re: [SPSSX-L] Problems with the 'SORT' menu Forwarding
to the list. Sounds to me like there is a bug in the UI which persists settings
from the last invocation of the dialog. At least users can get around this
problem via syntax. From: Mark A Davenport
MADAVENP [mailto:[hidden email]]
|
| Free forum by Nabble | Edit this page |