|
Someone may already have pointed out to Lloyd that an easy way to get most
output stats in more than the default decimal places is to increase the "decimals" column in the Variable View. It looks like output is then reported to the number of decimal places specified for the viewer plus 2. However, I have been wrestling with a related problem that may not be so simple. I often copy and paste table output into Word, which doesn't work as well as it did in older versions, but that's another story. However,I have a problem with the "significance" values. That seems to be fixed at 3 decimal places and I would rather have 5 or so. One can edit the cells individually in the table (Cell properties -- Format value)to increase the decimal places, but that is extremely time-consuming when you have dozens of tables. Is there any way to change the default for just those cells? I have fiddled with Table Looks and the like, but so far without success. (BTW, I have had occasion to tell clients that SPSS probably doesn't like to report p values to more places because precision isn't accurate with such tiny numbers, but when we have lots and lots of output all showing p=.000, it is useful to have more info. And you can't be sure that p=.010 is really p<.01 unless you can check to be sure it's not really p=.0104.) Thanks! Dear spssx-l members, I as using spss 16.0 for Windows, release 16.0.1 (Nov 15, 2007). My spss output in procedures such as DESCRIPTIVES; T-TEST; and CORRELATIONS shows the variable means to only TWO decimal places, which renders that output essentially useless for work I am now doing because my means are mostly less than .01. Currently my SPSS output gives means to 2 decimal places, and SDs to 3 decimal places. But I need my means displayed to at least 4 decimal places. Thank you very much for your help. lloyd williams [hidden email] ===================== 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 |
|
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Allan Lundy Sent: Tuesday, June 02, 2009 1:54 PM To: [hidden email] Subject: [SPSSX-L] how to output spss means to >2 decimal places? Someone may already have pointed out to Lloyd that an easy way to get most output stats in more than the default decimal places is to increase the "decimals" column in the Variable View. It looks like output is then reported to the number of decimal places specified for the viewer plus 2. However, I have been wrestling with a related problem that may not be so simple. I often copy and paste table output into Word, which doesn't work as well as it did in older versions, but that's another story. However,I have a problem with the "significance" values. That seems to be fixed at 3 decimal places and I would rather have 5 or so. One can edit the cells individually in the table (Cell properties -- Format value)to increase the decimal places, but that is extremely time-consuming when you have dozens of tables. Is there any way to change the default for just those cells? I have fiddled with Table Looks and the like, but so far without success. (BTW, I have had occasion to tell clients that SPSS probably doesn't like to report p values to more places because precision isn't accurate with such tiny numbers, but when we have lots and lots of output all showing p=.000, it is useful to have more info. And you can't be sure that p=.010 is really p<.01 unless you can check to be sure it's not really p=.0104.) Thanks! [>>>Peck, Jon] Yes, the precision level for many types of cell values is related to the input variable's format, but correlations, being the sort of things they are, don't reflect that. However, with v17 or later, you can use an extension command, SPSSINC MODIFY TABLES, to set the precision you want. Here is an example for a correlation table. CORRELATIONS /VARIABLES=educ jobtime prevexp salary salbegin /PRINT=TWOTAIL SIG /MISSING=PAIRWISE. SPSSINC MODIFY TABLES subtype="'Correlations'" SELECT="Pearson Correlation" "Sig. (2-tailed)" DIMENSION= ROWS LEVEL = -1 PROCESS = PRECEDING /STYLES APPLYTO=DATACELLS CUSTOMFUNCTION="customstylefunctions.SetDecimalPlaces(decimals=4)". This extension command requires the Python plugin and can be downloaded from SPSS Developer Central (www.spss.com/devcentral). The command is not fast, but it's a lot faster than doing this manually. And it can do other tricks, too, such as highlighting the correlations that are significant. HTH, Jon Peck Dear spssx-l members, I as using spss 16.0 for Windows, release 16.0.1 (Nov 15, 2007). My spss output in procedures such as DESCRIPTIVES; T-TEST; and CORRELATIONS shows the variable means to only TWO decimal places, which renders that output essentially useless for work I am now doing because my means are mostly less than .01. Currently my SPSS output gives means to 2 decimal places, and SDs to 3 decimal places. But I need my means displayed to at least 4 decimal places. Thank you very much for your help. lloyd williams [hidden email] ===================== 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 ===================== 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 |
| Free forum by Nabble | Edit this page |
