Status of Units, Relevance, Validity and Missing Values

a) Unit nonresponse

Fully completed (status 6) or partially completed surveys (status 5) are identified as VALID SURVEYS (units) throughout the data analysis. These surveys contain at least some partial answers of the respondents, while blank surveys and survey that have only been viewed and not answered are excluded.

All other survey statuses (e.g. 0, 1, 2, 3, and 4) and surveys that don't contain responses, are identified as INVALID SURVEY (units) and are therefore excluded from the analysis by default, although they can be important in the analysis of non-response bias and the survey completion process.

Therefore, only valid survey units (fully and partially completed, status 5 and 6) are included in the data analysis by default. The 1KA tool also offers the option of including other statuses (in addition to status 5 and 6) in the analysis. To do this, navigate to the 'Filtering' option in the 'DATA' tab and click on the button 'Statuses'. The tool has elaborated statuses of units which are presented with the following values of responses:

• E-mail – not sent: the e-mail was not sent to this person – '0';
• E-mail – non-response: the respondent did not respond to the questionnaire, which was sent by e-mail – '1';
• E-mail – error: an error occurred when sending the questionnaire to the e-mail address – '2';
• Click on the introduction: the respondent only clicked on the introduction (invitation to the survey) and left the survey immediately after. Either the person did not intend to complete the survey or the introduction page was not convincing enough – '3';
• Entered first page: the respondent only clicked on the introduction page and then on the second page of the survey. The person did not answer the first question and did not continue with the next page of questions – '4';
• Partially completed: the respondent partially completed the survey, i.e. responded to at least one or several questions and afterwards left the survey, resulting in a break-off. Thus the person did not click on the END of the survey button – '5';
• Partially blank: the respondent left the survey without any answers (blank survey). In this case we have a blank survey that resulted in a break-off – '5l';
• Completed survey: the respondent completed the survey (clicked on the ‘END of the survey’ button) and answered at least one question or possibly all of the questions – '6';
• Entirely blank: the respondent skipped all the questions without answering and clicked on the ‘END the survey’ button – '6l';
• Unknown status of survey: survey has a certain value – 'null'.  

If the respondent only clicked through the survey and did not answer any of the questions (i.e. viewer or lurker), the survey will be marked by a survey status that is specific to the 5 and 6 statuses. If the respondent came to the end of the survey, but did not answer any of the questions, this would be marked as survey blank (marked by 6I). In the mentioned case, all respondents' answers would be assigned value '-5'. If the respondent went through only a part of the survey (making the survey partially answered with the status 5),  the ‘not answered’ questions would be assigned the value -5, and the answers after leaving the survey would be assigned '-3', and the survey is marked blank and partially (not) completed (5I).

b) Item nonresponse

If the respondent does not answer certain questions, the following values are automatically generated:

• '-1' – the respondent completed the survey, but did not answer a specific question;
• '-2' – the respondent did not answer a specific question because of an inserted condition, e.g. the question was intended only for women; men would therefore skip it;
• '-3' – all questions that remained after the respondent left the survey;
• '-4' – all questions that were added after the respondent had already completed the survey;
• '-5' – all questions when a respondent did not answer any of the questions but only viewed the whole questionnaire or left in the middle of the survey.

c) Valid and missing responses

Missing values (as described in section b) have negative values in the database and are excluded from the analysis by default (but can also be included in the analysis). Missing values can be viewed in the Frequencies – Descriptive statistics under MISSING VALUES. Some analysis (e.g. Averages) do not take other missing responses into account. Namely, you can offer respondents the following options:

• Do not know (-99): the person did not know the answer to the question;
• Refusal (-98): the person refused to answer the question;
• Invalid (-97): the person replied with an invalid answer.

Answers Do not know (-99), Refusal (-98) and Invalid (-97), together with the other missing values (indicated by negative values, e.g. -1, -2, -3, -4, -5), are invalid answers in the statistical analysis. The analysis only takes VALID VALUES (i.e. greater than zero) into account.

The importance of marking all of the aforementioned negative values is very important because such values are not taken into account in the certain analysis (e.g. Average). If you want to change the default missing values settings, you can change them in the 'EDIT' - 'Settings' - 'Missing values' tab.

You can change the typical response values of categorical variables (default values are 1, 2, 3, 4, 5 ...) before collecting data in the advanced question editor (an icon for advanced editing of individual question categories).

d) Recoding questions

If you want to change the value of collected answers, do so by editing the question by clicking on the 'Edit values' checkbox in the 'Basic' tab. Example: You have questions on a scale of 1 to 5, and you marked the answer 'Do not know' with 8. You can subsequently mark this value as missing value 'Do not know'. This is then treated in the same way as 'Do not know' (-99), so it is not taken into account when calculating the average.