Some variables do not appear in my data set & the finished-variable and the missing-variable are often not accurate.

Question

My English questionnaire includes the variables H114–H112, H114–H132, H134–H149, H158, and H151–H157, but they do not appear in the dataset I downloaded. I do not have any problems with the other variables or the German version of my questionnaire.

Furthermore, the finished-variable and the missing-variable are often not accurate. Maybe this problem is linked to the first one.

Thanks in advance!

Answer 1

Usually, you should have the same variable in all languages, see Multilingual Surveys.

If you're missing variables, please follow this manual to narrow the issue down: Problem Solutions for Data Retrieval

Furthermore, the finished-variable and the missing-variable are often not accurate.

The variable FINISHED will tell you, if the person reached the last page (this may include a filter via goToPage('end')) or if you sent them away via redirect().

The MISSING variable counts the items asked and checks how many were answered, and how many not. Using the online view of the data (see the second link, above) you can exactly see which variables were counted (red/green) to calculate the MISSING value.

Screenshot from the project "view dataset"

Comments

Okay, please open the question H114 in the list of questions, and check the tab "Variables". Did you rename the variable, there?

If that does not explain the variables' whereabouts, I would like to create an administrator login and take a look the the issue myself. Would you be fine with that?

---

You mean the list of  of questions on the first page of SosciSurvey? What do you mean with ''the tab Variables'' - are you talking about the output?  

I have never renamed a variable.

I would be fine with you creating an administrator login - thank you very much for helping me! Should I give you access or activate anything for you?

---

When you select the question in the list of questions for editing, there's a tab "variables" under "Detail Settings". But I confirm that you have not renamed the variable H114, there :)

However, when I go to "collected data" -> "view dataset", I see the variable H114, and there's also data from your test case there. I'll edit my answer above and add the screenshot. Can you please use the "search" feature in your browser to look for H114 in that view and confirm that you can see the variable, as well.

---

Okay, after looking around a bit more in your project, I think I got bad news for you: https://support.soscisurvey.de/?qa=60160/variables-appear-finished-variable-missing-variable-accurate&show=60285#a60285

---

ok, now I can see them - I had to choose ''all questionnaires''. Sorry for that...

However the variables still doesn't appear in the output, when I download the collected data. I do not have the same or another problem with the other english questions.

Answer 2

Thanks for your permission to take a look into the project. I think I know the cause of your trouble.

The questions H114 etc. are located on page 17.

That would be fine, but there's a PHP filter on page 16.

And this filter says: Either go back to page 13 (under 6 years) or skip page 17 and directly go to page 18.

That means: No matter what your respondents did, they never saw page 17 with the H114 etc. questions. As the questions were never answered by the respondents, there cannot be any data for them in the dataset. By default, SoSci Survey will not show empty variables in the download.

That's a pity, but definitely something that should have beeen observed in the technical test.

Comments

that's how it should be - there are three versions of the FSSM-questionnare: FSSM 3bis6, FSSM 6bis11, FSSM 12bis18.
..and two versions of the CBCL-questionnaire: CBCLU6 & CBCL

After having filled in the FSSM Intro the participants should fill in a version of the FSSM and a Version of the CBCL:

If the participant goes to the kindergarten, then he/she has to fill in the FSSM 3bis6.
If the participant is between 6 and 11, he/she has to fill in the FSSM 6bis11.
If the participant is older than 11, he/she has to fill in the FSSM 12bis18.

Here the code (page 10):
if ($born > $limit4a || $born < $limit12a){
    goToPage('Screenout');
}elseif ($kiga_school == 7){ // Wenn Kindergarten ausgewählt wurde
    goToPage('FSSM 3bis6');
}elseif ($born > $limit11a && $kiga_school != 7){ //6-11 Jährige
 goToPage('FSSM 6bis11');
}else{
 goToPage('FSSM 12bis18'); //12 Jährige
}

After having filled in a FSSM-Version the participants under 6 years should fill in the CBCLU6 and the others (older than 6 years) the CBCL. Afterwarts they get to the KONT-Page (last page of the survey)

Here the code for this (page 12 or 16):

$born = strtotime(value('D104_01'));  // Zeitstempel des Geburtstags

$limit6a = strtotime('-6 years');  // Heute vor 6 Jahren

if ($born > $limit6a){
    goToPage('CBCLU6'); // Unter 6 Jahren
} else {
    goToPage('CBCL'); // Über 6 Jahren
}

I hope you understand what I mean - if not, please let me know

---

Well, try to fill out a questionnaire in a way to see question H114 ... if you can do it, you'll have data in the respective variables for that test case. If you do not have any such data in the previous cases, either nobody gave the proper answers or the filters wre incorrect. You can always look into the debug information to track how your filters work.

Given your test case 8915, I'm quite sure that answers from the respondents are stored correctly when the question is shown. To the most likely explanation for you to have no data in the respective variables is, that the questions have not been shown.

---

Thank you very much for your help! I thought that all the variables should appear in the output , regardless of whether they were shown/answered or not. It seems that i had not a participant older than 11 for the english questionnaire.  

Thanks again! I really appreciate it!

---

I have also question about the FINISHED- and Missing-variables. If a participant has not finished the survey, are the unanswered questions (those that have not yet appeared because of the dropout) declared as missings?

The missings are simply the red cells and the green cells are used to calculate the relative number of missings, right? Can I somehow determine afterwards which missing data are counted as missings and which are not?

---

The MISSING measure only counts those questions that have been presented to the respondent, and only if the page presenting them has been sent via "next" or "back". You'll ee that not-seen-variables remain white in the "view dataset" view.

> The missings are simply the red cells and the green cells are used to calculate the relative number of missings, right?

That is correct, yes. Grey and white are complete ignored.

> Can I somehow determine afterwards which missing data are counted as missings and which are not?

At the moment, SoSci Survey reports this metadata only as colors in the online data view, and by the variables MISSING and MISSREL. The detailed data contains as may variables as there are in the dataset, therefore, this infomation would usually be a little overload.

---

The cell of a question that can be skipped (because I set it that way when I created the question) is also marked red (categorised as missing) if it is not answered, isn't it?

For my survey that should not be a missing - these questions should be optional. That's why I was asking for a way to determine by myself which missing data are counted as missings and which are not?


Thanks in advance!

---

> The cell of a question that can be skipped (because I set it that way when I created the question) is also marked red (categorised as missing) if it is not answered, isn't it?

This is correct. The MISSREL indicator is weighted for such variables, while MISSING will simply count.