- EpiData-list - University of Manitoba Mailing Lists

Re: EpiData Analysis - Definition of histogram
by epidata-list＠lists.umanitoba.ca 14 May '07

14 May '07

Good point by Jamie Hockin. We should incorporate into histogram (and epicurve) the ability to define length of interval. Until this is done users can work like this (in the test version) * generate the grouped variable: (age into five year groups) gen i ag = age div 5 labelvalue ag /3="<20" /4="20-24" /5="25-29" /etc ....... (3 because in this example 16 was the minimum age) histogram ag /xa // which will work as desired Regards Jens Lauritsen

1 1

Epicurve comments
by epidata-list＠lists.umanitoba.ca 13 May '07

13 May '07

I fully agree. I think there is a need of a date-time variable at least in Analysis. That can allow users to group cases in different periods of time (hours or days) with the complexity of reprograming as we have to do now. We also need the posibility of labelling the epicurve at two diferent levels in the time axis so that we can clearly see days an hours for example. This is more or less solved in EpiInfo for Windows (the grouping part). Jens, maybe you can have a look at the specifications for the epicurve I sent you earlier this year. Another question regarding some of the graphs is the need of doing the graph by a second variable for example bar agegroup by sex or line month by year. Thanks a lot Jens for all this job and thanks every body for their comments Pedro Arias > I agree. However, the question of x-axis increment is not fully defined. > There are two aspects to this. The usual option /xinc= will only affect > the labelling on the chart, as with other graphs. However, the interval > used to define boxes is something that should also be an option with > histogram. This is particularly important for epicurves. This means that > we can avoid awkward recoding of variables (especially tricky with > dates) to have an epicurve that groups cases by 4-hour intervals, for > example, and also labels the x-axis every 12 hours: > > epicurve onsettime case /xinc=12 /interval=4 > > The same would be true of a histogram by age, where we wanted age > grouped by 5 years. I would prefer > > his age /interval=5 /xinc=10 > > or > > define age5 ## > age5 = age div 5 > his age5 /interval=5 > > Right now, his and epicurve are constrained to intervals of 1, which > means that labelling is not always correct. > > Jamie Hockin > Public Health Agency of Canada > > Jens wrote: >> After constructive comments from Egil Arnesen in Tromsø, Norway and >> searching in various textbooks I have decided that for v2 the >> following two definitions will be implemented: >> >> histogram: >> A graph, where there is one bar for each value present in the data. >> The height of the bar corresponds to the count of records in the data. >> (possibly later the percentage of data). Each bar starts at the >> x-value which was in the data (e.g. age 45 is shown from 45-46.) And >> the width of one bar is the same as the unit of the x variable. >> The X-axis is numerical and can be scaled with minimum, maximum and >> increase.E.g. from 25 to 100 with tickmarks by 5 (25 30 35 ......95 >> 100) >> >> Bar: >> A graph, where there is one bar for each value represented in the >> data. The X-axis is categorical and the bar is centered around the >> tick mark, which would show the value label (or the value) given by >> the x value. >> >> If anyone has comments to this please send to this list. >> >> The principle of the histogram will be implemented also for EpiCurve >> (but has not been done yet). The mentioned implementation of bar and >> histogram should work from build 112 uploaded shortly before this >> mail. (Minor problem with the build 112 uploaded yesterday). > > > > ------------------------------ > > ________________________________________ > EpiData-list(a)lists.umanitoba.ca > http://lists.umanitoba.ca/mailman/listinfo/epidata-list > > > End of EpiData-list Digest, Vol 43, Issue 5 > *******************************************

1 0

EpiData Analysis - Definition of histogram
by epidata-list＠lists.umanitoba.ca 12 May '07

12 May '07

After constructive comments from Egil Arnesen in Tromsø, Norway and searching in various textbooks I have decided that for v2 the following two definitions will be implemented: histogram: A graph, where there is one bar for each value present in the data. The height of the bar corresponds to the count of records in the data. (possibly later the percentage of data). Each bar starts at the x-value which was in the data (e.g. age 45 is shown from 45-46.) And the width of one bar is the same as the unit of the x variable. The X-axis is numerical and can be scaled with minimum, maximum and increase.E.g. from 25 to 100 with tickmarks by 5 (25 30 35 ......95 100) Bar: A graph, where there is one bar for each value represented in the data. The X-axis is categorical and the bar is centered around the tick mark, which would show the value label (or the value) given by the x value. If anyone has comments to this please send to this list. The principle of the histogram will be implemented also for EpiCurve (but has not been done yet). The mentioned implementation of bar and histogram should work from build 112 uploaded shortly before this mail. (Minor problem with the build 112 uploaded yesterday). Kind regards Jens Lauritsen EpiData Association

1 1

EpiData Analysis - next test release ready
by epidata-list＠lists.umanitoba.ca 11 May '07

11 May '07

Build 112 of analysis has been uploaded for testing. Users are encouraged to try the following: * Install routine * new table command * epicurve dotplot cdfplot * Spc routines * documentation * reading and saving files _Known issues and comments_ * Default sorting of non-integer reals in tables * Histogram error (rightmost bar not shown) * Notice /cc changed to /nm in tables * recode v1 to v2 by x gives partial errors (e.g. 10-1 09) * possibly help file principle partially correct writing word in command prompt via F1 link to help file _Recently solved_ * All fields saved correctly (?) * EpiCurve Final * Added Upper Case fields * New graphs: Histogram refined (xinc xmin and xmax working) * Clarified help file (F1) * (other bugs removed) * test of date correctness * added protection against reserved terms(e.g. date) The addition of upper case fields had consequences in many areas. Previously analysis did not compare strings correctly. Before release as version 2 the following will be added: Writing chk files, Translation of menu's, Dialog for the new tables module, fixing bugs Kind regards Jens Lauritsen EpiData Association

1 0

Confidence intervals during analysis
by epidata-list＠lists.umanitoba.ca 07 May '07

07 May '07

A very belated thanks Jens for your response! Looking forward to new developments. Regards Juanette John Natural Resources & the Environment - CSIR PO Box 395 Pretoria 0001 South Africa CSIR: "our future through Science" http://www.csir.co.za -- This message is subject to the CSIR's copyright, terms and conditions and e-mail legal notice. Views expressed herein do not necessarily represent the views of the CSIR. CSIR E-mail Legal Notice http://mail.csir.co.za/CSIR_eMail_Legal_Notice.html CSIR Copyright, Terms and Conditions http://mail.csir.co.za/CSIR_Copyright.html For electronic copies of the CSIR Copyright, Terms and Conditions and the CSIR Legal Notice send a blank message with REQUEST LEGAL in the subject line to CallCentre(a)csir.co.za. This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.

1 0

Réponse automatique d'absence du bureau : [EpiData-list] problem starting Epidata
by epidata-list＠lists.umanitoba.ca 02 May '07

02 May '07

Bonjour, Je suis actuellement absent du bureau.Je serai de retour le 21 mai 2007 Je répondrai à cette date à votre message A bientôt Hello, I am currently out of the office and will be back on May the 21st 2007 I will reply to your message then. All the best Robert

1 0

problem starting Epidata
by epidata-list＠lists.umanitoba.ca 02 May '07

02 May '07

Hi all, I'm new to Epidata and have worked with it for about a week now. It's running on various PCs with Windows XP, connected in a network. Now, however, a problem has occured, that the application refuses to run on any of the computers connected to the network. Upon start it looks like the app will start as normal, but then will remain as a minimized application in the taskbar. I've tried to start it both by using the shortcut on the desktop and by clicking the datafile I am working with directly, however nothing really worked. Does anybody have a clue what might be the source of the problem? And if I cannot avoid to reinstall the software, do I need to backup my data? Sorry for my poor language proficiency and the relatively simple-looking problem I have. Thanks in advance. Thorsten _______________________________________________________________ SMS schreiben mit WEB.DE FreeMail - einfach, schnell und kostenguenstig. Jetzt gleich testen! http://f.web.de/?mc=021192

1 0

Getting values from other records (was: create reduced dataset with EpiData Analysis): long to wide conversion
by epidata-list＠lists.umanitoba.ca 30 Apr '07

30 Apr '07

Hans wishes to get from SET12.REC with variables: ID AGE SEX VAR1 VAR2 - where each person (ID) have four records which we each call one visit to NEWSET.REC with variables: ID AGE SEX VAR11 VAR21 VAR12 VAR22 VAR13 VAR23 VAR14 VAR24 where each person (ID) have one record This conversion can be named "From long to wide format" and is a conversion we often wish to work on: In the example it is quite easy, since it seems that there are exactly four visits per person. One way of doing this is: (principle must be tested on actual data) read set12.rec freq id // here I would look to make sure that all id's actually had four records * We assume that the order of visits in set12 is chronological, but wish to have an identifier: gen i visit = 1 // integer variable with value 1 for all if id = id[_n-1] then visit = visit[_n-1] + 1 // now the value of visit is correct 1 2 3 4 * the [_n] indicates recnumber and by subtracting one "_n-1" we get the previous record browse // here I would look to control that the calculation was correct * generate the 8 variables: (you could also use define followed by let ) gen var11 = -1 ..... etc until gen var24 = -1 * the reason for the value -1 is that we wish to know if the variable is changed later. Sort id visit * now we can move the values to the correct variable: (shown for the var11..var14) if visit = 1 then var11 = var1 if visit = 1 then var12 = var1[_n+1] if visit = 1 then var13 = var1[_n+2] if visit = 1 then var14 = var1[_n+3] * and similarly for the var2x variables browse // to see visually that it works * all values have now been copied from four records to the first record for each person select visit = 1 labeldata "My combined file in wide format" savedata newfile The principle above can be used for many conversions. E.q. for first and follow up visit: * assume bmi indicates body mass index. And we wish to test for bmi change by sex: gen bmidif = . if (visit = 1) and (id = id[_n+1]) then bmidif = bmi[_n+1] - bmi means bmidif sex // for testing in version 2 add /t The next question could be "How to get from wide to long format" - The reverse of what Hans wanted. This would not be quite as easy, but something like generate 400 // an empty file first, e.g. 100 observations with four records gen i s = _n // instead of _n recnumber can be used gen i id = 1 if (s mod 4) = 0 and (recnumber > 4) then id = id[_n-4] + 1 * plus some more logic to get the id numbers right. * then the data could be added by "relate" and subsequently moved to the correct variables regards Jens Lauritsen

1 2

Getting values from other records - long to wide conversion - continued
by epidata-list＠lists.umanitoba.ca 30 Apr '07

30 Apr '07

Marcel extended the example by asking: Is there a way to make sure that the data set is sorted to be sure that the assumption: "* We assume that the order of visits in set12 is chronological, holds" ? I recommend having a time in the "visits record", preferably: e.g. datevisit, but also a sequential number in patient follow up studies, e.g.visit 1=start 2=4monts 3=12 months 4=2years etc. In real empirical data it usually happens that one or more visits are either missed or taking place at a point in time not exactly being on time. E.g. how to designate a visit 5 monts after initial treatment in a study where visits are supposed to be at 4 and 6 months. Sequence often indicates certain tests (e.g. blood samples) which vary by visit, e.g. more expensive ones less often, such as x-rays. By using as well the date and the sequence we can know exactly what are the data, which variables are missing and what is the time difference from "supposed time of visit". Therefore careful planning obviously remedies this by having the possibility to sort as well on sequence and date. sort id visit In the control of such data I would always: read rawdata *suppose we need 4 visits per person, numbered 1 2 3 4 * 1+2+3+4=10 aggregate studyid /sum="visit" /close gen i ok=(sumvisit=10) list studyid visit if ok <> 1 * or the same in "epi6" mode: read rawdata aggregate .... as above... define ok # let ok = 0 if (sumvisit = 10) then ok = 1 select if ok <> 1 list studyid select regards Jens Lauritsen

1 0

Create reduced dataset with EpiData Analysis
by epidata-list＠lists.umanitoba.ca 28 Apr '07

28 Apr '07

I have a parent SET1.REC, containing a key unique identifier ID and AGE and SEX, related to child SET2.REC, containing VAR1 and VAR2 In analysis, I create SET12.REC that has now as many records as SET2 (ID, AGE, and SEX) repeated for each record up to as many times as there are values for VAR1 and VAR2. If SET1 actually represents the individuals identified by ID, and SET2, e.g., visits for ID, and I wish now to make a NEW.REC containing only as many records as there are IDs, but making new variables VAR11, VAR21, VAR12, VAR22, and VAR13, VAR23, VAR14, VAR24 (if there are up to 4 visits for each ID, HOW would one thus write an EpiData Analysis command set to get from: SET12.REC ID AGE SEX VAR1 VAR2 to NEWSET.REC ID AGE SEX VAR11 VAR21 VAR12 VAR22 VAR13 VAR23 VAR14 VAR24 (where I know there is a maximum of 4 visits per ID) with only ID number of records? Thank you for any input you might have, Hans -- Hans L Rieder, MD, MPH Jetzikofenstr. 12 3038 Kirchlindach Switzerland Tel: +41 31 829 4577 Mob: +41 79 321 9122 Web: http://www.tbrieder.org

1 0