View Issue Details

This bug affects 1 person(s).
 14
IDProjectCategoryView StatusLast Update
03572Bug reportsStatisticspublic2009-12-11 12:17
Reportermiamiman Assigned Tolemeur  
PrioritynormalSeverityminor 
Status closedResolutionfixed 
Product Version1.85+ 
Fixed in Version1.87 
Summary03572: Statistics not equal to Pie Charts
Description

Ok, I am back with this, because the bug was not reproducible on the bug forums, and was closed.

So after updating to 1.85 and seeing the error again, I hereby submit the following.

  1. Please see the following three [3] step survey.
    limesurvey.encuestadatos.com/index.php?sid=75679&lang=en

  2. If you choose YES on question #1, the survey takes you to question #2, onto question #3, and finish.

  3. However, if you choose NO on question #1, the branching logic takes you directly to question #3, and finish.

So far, so good!

However, I am [again] informing the community that the STATISTICS generated for the "completed questions only", do NOT match those generated by PIE CHARTS. The percentages are calculated against different sets of data.

Of course, the STATISTICS presented should be those of the PIE CHARTS.

Please see image below of the results. Hopefully this time, it will be clea

TagsNo tags attached.
Attached Files
limesurvey_survey_75679.csv (3,907 bytes)   
# LimeSurvey Survey Dump
# DBVersion 138
# This is a dumped survey from the LimeSurvey Script
# http://www.limesurvey.org/
# Do not change this header!

#
# SURVEYS TABLE
#
"sid","owner_id","admin","active","expires","adminemail","private","faxto","format","template","language","additional_languages","datestamp","usecookie","notification","allowregister","allowsave","autonumber_start","autoredirect","allowprev","printanswers","ipaddr","refurl","datecreated","listpublic","htmlemail","tokenanswerspersistence","usecaptcha","bounce_email","startdate","publicstatistics","publicgraphs","assessments","usetokens","attributedescriptions"
"75679","1","Max Jaramillo","Y","","maxjaramillo@latinpressinc.com","Y","","G","default","en","","N","N","0","N","N","0","Y","N","Y","N","N","2009-08-14","Y","Y","N","D","maxjaramillo@latinpressinc.com","","Y","Y","N","N",""

#
# GROUPS TABLE
#
"gid","sid","group_name","group_order","description","language"
"57","75679","Group Beginning","0","","en"
"58","75679","Group Middle","1","","en"
"59","75679","Group Last","2","","en"

#
# QUESTIONS TABLE
#
"qid","sid","gid","type","title","question","preg","help","other","mandatory","lid","lid1","question_order","language"
"248","75679","57","Y","1","Do you like Swiss cheese?","","","N","Y","0","0","0","en"
"249","75679","58","Y","2","Do you like Fondue?","","","N","Y","0","0","0","en"
"250","75679","59","Y","3","Have you been to Switzerland?","","","N","Y","0","0","0","en"

#
# ANSWERS TABLE
#

#
# CONDITIONS TABLE
#
"cid","qid","scenario","cqid","cfieldname","method","value"
"239","249","1","248","75679X57X248","==","Y"

#
# LABELSETS TABLE
#

#
# LABELS TABLE
#

#
# QUESTION_ATTRIBUTES TABLE
#

#
# ASSESSMENTS TABLE
#

#
# SURVEYS_LANGUAGESETTINGS TABLE
#
"surveyls_survey_id","surveyls_language","surveyls_title","surveyls_description","surveyls_welcometext","surveyls_urldescription","surveyls_email_invite_subj","surveyls_email_invite","surveyls_email_remind_subj","surveyls_email_remind","surveyls_email_register_subj","surveyls_email_register","surveyls_email_confirm_subj","surveyls_email_confirm","surveyls_url","surveyls_endtext","surveyls_dateformat"
"75679","en","Test - Branching Conditions","test","welcome test","","Invitation to participate in survey","Dear {FIRSTNAME},\n\nYou have been invited to participate in a survey.\n\nThe survey is titled:\n""{SURVEYNAME}""\n\n""{SURVEYDESCRIPTION}""\n\nTo participate, please click on the link below.\n\nSincerely,\n\n{ADMINNAME} ({ADMINEMAIL})\n\n----------------------------------------------\nClick here to do the survey:\n{SURVEYURL}","Reminder to participate in survey","Dear {FIRSTNAME},\n\nRecently we invited you to participate in a survey.\n\nWe note that you have not yet completed the survey, and wish to remind you that the survey is still available should you wish to take part.\n\nThe survey is titled:\n""{SURVEYNAME}""\n\n""{SURVEYDESCRIPTION}""\n\nTo participate, please click on the link below.\n\nSincerely,\n\n{ADMINNAME} ({ADMINEMAIL})\n\n----------------------------------------------\nClick here to do the survey:\n{SURVEYURL}","Survey registration confirmation","Dear {FIRSTNAME},\n\nYou, or someone using your email address, have registered to participate in an online survey titled {SURVEYNAME}.\n\nTo complete this survey, click on the following URL:\n\n{SURVEYURL}\n\nIf you have any questions about this survey, or if you did not register to participate and believe this email is in error, please contact {ADMINNAME} at {ADMINEMAIL}.","Confirmation of completed survey","Dear {FIRSTNAME},\n\nThis email is to confirm that you have completed the survey titled {SURVEYNAME} and your response has been saved. Thank you for participating.\n\nIf you have any further questions about this email, please contact {ADMINNAME} at {ADMINEMAIL}.\n\nSincerely,\n\n{ADMINNAME}","","","1"

#
# QUOTA TABLE
#

#
# QUOTA_MEMBERS TABLE
#

limesurvey_survey_75679.csv (3,907 bytes)   
Bug heat14
Complete LimeSurvey version number (& build)
I will donate to the project if issue is resolved
Browser
Database type & versionlatest
Server OS (if known)linux
Webserver software & version (if known)latest
PHP Versionlatest

Users monitoring this issue

c_schmitz, miamiman

Activities

user372

2009-08-17 13:55

  ~09149

Please attach a simple sample survey (survey structure .csv-file), where we can reproduce the issue - Thx!

miamiman

miamiman

2009-08-17 14:54

reporter   ~09151

I have uploaded the csv file for the above survey.

Please note that the statistics are incorrect when pulled against "Completed Records Only" option.

Good luck.

Mazi

Mazi

2009-09-02 18:25

updater   ~09346

Jason, can you take care of this? Otherwise leave a note.

user372

2009-09-05 15:00

  ~09402

miamiman already donated 50$ for a quick fix for this problem.
@ Lemeur: Jason Cleeland is on vacation at the moment - can you have a look at it?

lemeur

lemeur

2009-09-05 15:41

developer   ~09407

@miamiman,
Hi, as ElMatador69 said, our statistics specialist is on vacation now.

I'm myself getting back from holiday and have serious bugs waiting for me so I'll have to concentrate on this first.

As far as I can see on your screenshots, the issue is about what total is used to compute the percentage.

There are 4 answers:

  • 2 with Yes answers for Question1 and 2 with No as answer for this Question
  • Question2 is conditionnal (only displayed when Question1's answer is Yes)
    ==> It seems that LS uses 2 computations for the percentage:
    • in the table the percentage is re'lative to the total number of answers (4) weither or not the participants were presented the question
    • in the graph, the computation is relative to the total number of answers for this question (2)

I can't decide if solution 1 or 2 is the good one, cause both are interresting and reveal different information...

miamiman

miamiman

2009-09-05 18:43

reporter   ~09409

I agree that it is NOT neccesarily an error.. but the discrepancy is annoying and time consuming to make the final presentation of the data.

Regarding which data to be presented, I would argue strongly that it needs to be the data of the sub-set. That is, once you've made a branching question, the results graphed and mapped should be "the completed only" of the sub-set.

lemeur

lemeur

2009-09-08 22:16

developer   ~09474

The solution for this is clearly not easy because there is no easy way to differentiate in the response if a question cas displayed but received no answer, or wasn't displayed and thus received no answer.

The clean fix would be a change in LS1 so that not displayed answers receive a NULL value instead of a blank value.
note for later: see deletenonvalue in save.php.

miamiman

miamiman

2009-09-14 18:01

reporter   ~09551

Sorry for delay here in responding.

The easiest solution would be for the statistics to mirror the graphical data.

That's all that's really required.

user372

2009-09-25 08:27

  ~09651

@ Lemeur: how will we proceed at this problem? miamiman is waiting for fix.
@ c_schmitz; which solution do you prefer?

c_schmitz

c_schmitz

2009-09-25 14:39

administrator   ~09657

We have talked about this. THe fix is scheduled for >1.86

miamiman

miamiman

2009-10-02 21:04

reporter   ~09704

Hello fellows!

I just downloaded the most recent build of 1.86 and the inconsistency between the graphs and percentages on "completed only" is still there for condition/branching questions.

Really would love to have this issue resolved?

Let me know when you have a chance.

THanks.

c_schmitz

c_schmitz

2009-10-02 23:08

administrator   ~09705

1.86 means "greater than 1.86" - so that's why it is not in there.
Atm it looks like it might not be even on 1.87 but that is not decided yet.

lemeur

lemeur

2009-10-25 18:16

developer   ~09857

A first version of the full patch is implemented in rev 7779.

I've fixed a part of the issue by making sure that columns corresponding to hidden questions are always filled with NULL values.

Second part of the fix is be done in statistic_functions, though since I'm not an expert of this module I can't assure this won't have other collateral effects.

Matador, can you test ?

Mazi

Mazi

2009-10-27 14:19

updater   ~09871

lemeur, did you change the code at both, admin and public statistics?

lemeur

lemeur

2009-10-27 14:43

developer   ~09873

I've modified the statistics_function file... not sure if this is used in the public_stats feature ?
Anyway user stats was and seems to be still broken....

miamiman

miamiman

2009-12-04 16:59

reporter   ~10455

Can I ask if this issue has/will be taken care of in the current beta release?

c_schmitz

c_schmitz

2009-12-07 01:57

administrator   ~10498

The issue will be fixed in the upcoming 1.87RC5 release.

Issue History

Date Modified Username Field Change
2009-08-15 18:49 miamiman New Issue
2009-08-15 18:49 miamiman Status new => assigned
2009-08-15 18:49 miamiman Assigned To => user372
2009-08-15 18:49 miamiman File Added: statistical_errors-branchingquestions.png
2009-08-15 18:49 miamiman Database & DB-Version => latest
2009-08-15 18:49 miamiman Operating System (Server) => linux
2009-08-15 18:49 miamiman Webserver => latest
2009-08-15 18:49 miamiman PHP Version => latest
2009-08-15 18:50 miamiman Issue Monitored: miamiman
2009-08-17 13:55 user372 Note Added: 09149
2009-08-17 13:55 user372 Status assigned => feedback
2009-08-17 14:53 miamiman File Added: limesurvey_survey_75679.csv
2009-08-17 14:54 miamiman Note Added: 09151
2009-08-19 01:30 c_schmitz Status feedback => assigned
2009-08-19 01:30 c_schmitz Assigned To user372 => c_schmitz
2009-08-19 14:06 c_schmitz Assigned To c_schmitz => jcleeland
2009-09-02 18:25 Mazi Note Added: 09346
2009-09-05 15:00 user372 Note Added: 09402
2009-09-05 15:00 user372 Assigned To jcleeland => lemeur
2009-09-05 15:41 lemeur Note Added: 09407
2009-09-05 18:43 miamiman Note Added: 09409
2009-09-08 22:14 c_schmitz Project Bug reports => @2@
2009-09-08 22:16 lemeur Note Added: 09474
2009-09-14 18:01 miamiman Note Added: 09551
2009-09-25 08:27 user372 Note Added: 09651
2009-09-25 08:27 user372 Issue Monitored: c_schmitz
2009-09-25 14:39 c_schmitz Note Added: 09657
2009-10-02 21:04 miamiman Note Added: 09704
2009-10-02 23:08 c_schmitz Note Added: 09705
2009-10-25 18:16 lemeur Note Added: 09857
2009-10-25 18:16 lemeur Status assigned => feedback
2009-10-27 14:19 Mazi Note Added: 09871
2009-10-27 14:43 lemeur Note Added: 09873
2009-12-04 16:59 miamiman Note Added: 10455
2009-12-07 01:57 c_schmitz Note Added: 10498
2009-12-07 01:57 c_schmitz Status feedback => resolved
2009-12-07 01:57 c_schmitz Fixed in Version => 1.87
2009-12-07 01:57 c_schmitz Resolution open => fixed
2009-12-07 01:57 c_schmitz Project @2@ => Bug reports
2009-12-11 12:17 c_schmitz Status resolved => closed
2021-08-04 14:27 guest Bug heat 10 => 14