December 2004 - EpiData-list - University of Manitoba Mailing Lists

New introductory guide to EpIData
by epidata-list＠lists.umanitoba.ca 13 Dec '04

13 Dec '04

Ziad has sent me the pdf file of the notes for distribution. Look on Http://www.epidata.dk/documentation.php for downloading. The e-mail for correspondance to ZIAD is placed on the frontpage of the note. Thanks for the contribution. Jens M. Lauritsen, Consultant,MD. PhD. The EpiData Association,

1 0

Definition of Values, Jumps Reset
by epidata-list＠lists.umanitoba.ca 13 Dec '04

13 Dec '04

Jens, Thank you very much for your help. From reading the help file, I didn't realize that "JUMPS RESET 9" would actually fill in each variable with the full length. That actually works better for me. Is there anyway that I can reset only my numeric values to 9? I would prefer to keep my char variables clear when they're skipped. I tried writing: IF (Q14=55) THEN CLEAR Q15B CLEAR Q16B ENDIF I thought this might overwrite the "JUMPS RESET" for those variables, but it didn't work that way. Is there any other way to do this, aside from writing a bit of code (Q15A=99, CLEAR Q15B, etc...) for each variable between Q14 and ENDNOTE1? Also, regarding the message "Definition of values (comment legal, missingvalue, range ...)": In my program, I have code like in your example: v1 mustenter missingvalue 99 comment legal use label_q01 after entry if (v1 <> 8) and (v1 <> 14) and (v1 <> 21) and (v1 <> 99) then help "Wrong value entered" goto v1 endif end end The reason I did this is because I wanted the users to be able to override the legal values if they felt it was necessary to type in something other than 8, 14, 21, or 99 -- sometimes valid codes are added to our variables. Yet I wanted the users to still get a warning if they didn't use one of the legal values. I also had to add "range -9 99" in order for this to work with "comment legal use". As far as I understand, users cannot override legal values or range values. Please correct me if I am wrong. As a result, I stay away from using legal and range in my EpiData programs, unless I want to have value labels. While the after entry block above is seemingly unnecessary, it at least allows the user to click in the next appropriate field with their mouse and force the program to accept the value. Thank you, Katie Thornton Programmer Institute for Research on Poverty

1 0

EpiData guideline - released for testing
by epidata-list＠lists.umanitoba.ca 13 Dec '04

13 Dec '04

Hello! The file is done now and I am sending it to test it. If someone is interested to have a copy, please write me back with your email address so I can email it to you (it is around 5.5Mb size) in PDF format. It is free for distribution as well! And anyone is welcome to write back with questions and/or suggestions! Note: I have intention to go further in guidelines development in case there is interest and fund (maybe?) to keep the work on? Best regards! /Ziad Karolinska Institute Stockholm - Sweden

1 3

Definition of values (comment legal, missingvalue, range ....)
by epidata-list＠lists.umanitoba.ca 12 Dec '04

12 Dec '04

Often when users of EpIData send in files to epidata.dk for help it turns out that chk files are created more complex than actually needed. One example of this: From a chk file (opened in the editor, not in "Add/revise") example: labelblock LABEL label_q01 8 dog 14 Horse 21 "lazy tiger" end end v1 missingvalue 99 comment legal use label_q01 end In this situation only the values 8 14 21 and 99 plus "blank=missing" can be entered. If the v1 block is changed to: v1 mustenter missingvalue 99 comment legal use label_q01 end Only the values 8 14 21 and 99 can be entered. So there would be no reason to test in a "after entry .... end" block like this: v1 mustenter missingvalue 99 comment legal use label_q01 after entry if (v1 <> 8) and (v1 <> 14) and (v1 <> 21) and (v1 <> 99) then help "Wrong value entered" goto v1 endif end end One other aspect of this is that sometimes people define the same values in a RANGE and LEGAL and COMMENT LEGAL and MISSINGVALUE Any value should only be in one of these definitions. regards Jens Lauritsen

1 0

Definition of values (comment legal, missingvalue, range ....)
by epidata-list＠lists.umanitoba.ca 12 Dec '04

12 Dec '04

Often when users of EpIData send in files that are giving problems it turns out that the chk files are created more complex than actually needed e.g.: LABEL label_q01 8 dog

1 0

Re: Jumps reset
by epidata-list＠lists.umanitoba.ca 12 Dec '04

12 Dec '04

The command "jumps reset value", e.g. Q14 TOPOFSCREEN JUMPS RESET 9 55 ENDNOTE1 END will do: after entry in field q14 EpiData will check "Is q14 = 55" if so then jump to field endnote1, but before "landing in endnote1" : RESET all numerical variables from Q14 to ENDNOTE1 to missing ("." or ""). and furthermore : RESET 9 : fill all variables from q14 to ENDNOTE1 with as many 9's as the width of each field. That is if you have a field of width 2 the value will be 99, a field of length 10 will be filled with all 9's. So you cannot do: jumps reset 99. Only one digit is allowed, but the variables passed with the jump will get 99 if these fields are of width 2. Regards Jens Lauritsen Initiator and Coordinator of EpiData

1 0

MISSINGVALUE / JUMPS RESET commands
by epidata-list＠lists.umanitoba.ca 09 Dec '04

09 Dec '04

Send the chk + qes file to info(a)epidata.dk and we will check if it is a limit problem or what could be the cause. Also check that you are using most recent version, which is 3.02, but v3.1 is on the test page (www.epidata.dk/testing.php) - we are lacking final. In general it is a good suggestion to partition a study such that each type of data is in a particular file connected with relate. Also if you press F2 twice a small "debug" window will be shown with current file, variable and memory status. This can give further information. E.g. if you are working on the correct folder. Regards Jens Lauritsen Initiator and Coordinator of EpiData

1 0

MISSINGVALUE / JUMPS RESET commands
by epidata-list＠lists.umanitoba.ca 08 Dec '04

08 Dec '04

Hello, I've really enjoyed using EpiData, and I have to say that the application I am creating will be a HUGE improvement over what our data collectors had been using. They're excited to start using EpiData. I am having problems using some check file commands, however. When I use the command "Jumps Reset 99", this code mysteriously disappears from my .chk file without warning after I save and exit. If I use "Jumps Reset 9", it stays. I need to use 99, though, because 9 is a valid code for some of my variables. I am also having trouble using the Missingvalue command. When I type in "MISSINGVALUE 98" in a field, this code will also mysteriously disappear from my .chk file, though not always immediately, like "Jumps Reset 99" does. I have tried using the missingvalue command in the before file block, but I've had no luck with this. I am wondering if the size of my .chk file has anything to do with these problems. I saw in the help file that there is a limit of 64Kb if you work in the editor, but I use the add / revise checks feature so I'm not sure if this affects me. My .chk file is 90Kb, and over 4200 lines long. If I can't use Jumps Reset 99, this will increase the size of my .chk file even further. Besides the above problems, everything is working OK. Any advice that you can give would be greatly appreciated. Thank you, Katie Thornton Programmer Institute for Research on Poverty

1 0

Labels in stata after export from epidata
by epidata-list＠lists.umanitoba.ca 07 Dec '04

07 Dec '04

Correct. Thanks! However, in order for the value label to work properly in Stata, it seems that the value label must be written in ascending order, i.e. 0 No_Adjuv 1 Adjuv not the other way around. Halvor Sommerfelt At 00:28 07.12.2004 +0100, you wrote: >Most likely the problem is a Stata matter of how labels are shown in the >data browser > >Try in stata: >use xxxxxxxx (your file) >d >* check the formats >* set format for the variable with problems : >format %9.0g x (assume name is x) >browse > >Works here for me. If the display format is too narrow for a given >variable, then the label is not showing correctly in browse, but does show >correctly at : >label list >or >tab >or >freq >of that variable. > >If you still have problems send the rec+chk file as an attachment to >info(a)epidata.dk or indicate adress for download. > >regards > >-- >Jens Lauritsen >Coordinator and initiator of EpiData Project >http://www.epidata.dk > > >_______________________________________________ >EpiData-list mailing list >EpiData-list(a)lists.umanitoba.ca >http://lists.umanitoba.ca/mailman/listinfo/epidata-list

1 0

Re: labels in stata after export from epidata
by epidata-list＠lists.umanitoba.ca 06 Dec '04

06 Dec '04

Most likely the problem is a Stata matter of how labels are shown in the data browser Try in stata: use xxxxxxxx (your file) d * check the formats * set format for the variable with problems : format %9.0g x (assume name is x) browse Works here for me. If the display format is too narrow for a given variable, then the label is not showing correctly in browse, but does show correctly at : label list or tab or freq of that variable. If you still have problems send the rec+chk file as an attachment to info(a)epidata.dk or indicate adress for download. regards -- Jens Lauritsen Coordinator and initiator of EpiData Project http://www.epidata.dk

1 0