Show logic file
From LimeSurvey Manual
Revision as of 13:43, 6 November 2018 by Cdorin
An important option that helps you create and maintain surveys is Show logic file
Throughout development and testing of the survey, and before activating it, it is very important to validate the survey logic. This is especially true because you can write complex relevance, tailoring, and validation equations, and need to be sure that nothing will break when you run the survey.
This feature lets you quickly validate the accuracy of your survey, group(s), or question(s). It can be accessed from the top bar menu options located under the survey-related settings. It is available via the Tools menu:
As you can observe above, you can run this option four times, for each language used within your survey.
The Survey Logic File shows everything that you specified for each question (e.g. name, text, help, conditions/relevance, validation rules, defaults, subquestions, answers) in a convenient tabular format. It highlights any errors, and lets you click on the question and group IDs (or variables used within equations) to open new browser tabs to edit those questions or groups. This makes it easy to quickly edit any errors and refresh the logic file to confirm the accuracy of the survey before activating it.
The display is also designed to be be readable by researchers and study sponsors so that they can validate the accuracy of the survey design and logic. Show logic file update the cache for all expression for active survey.
It includes the following columns:
- # - shows the Group and Question sequence counts, starting from 0.
- Name [ID] - shows the variable name for the group/question/subquestion, using qcode naming convention (these are the names by which you can refer to your variables in Expression Manager compliant equations). ID is the question id (QID), or group id (GID). This field also shows the question type (e.g. Multiple choice [M])).
- Relevance [Validation] (Default) - shows the following:
- Relevance - the syntax-highlighted relevance equation for the question or group. If it is always true, the value will be 1.
- Validation - Expression manager automatically generates the validation equation based upon the selected question attributes (e.g. min/max number of answers, min/max/equals sum values, min/max individual values, regular expression validation). This section shows the generated validation equation so that you can detect if there are any errors (e.g. undefined variables).
- Question-level validation shows the equation needed to verify the above-described question attributes
- Sub-Question-level validation shows the equation needed to implement array_filter, array_filter_exclude, and exclusive_option.
- Default - if the question has a default value, it is shown here, syntax-highlighted (since the default could be an expression).
- Text [Help] (Tip) - shows the following
- Text - the text of the group, question, subquestion, or answer. They are syntax-highlighted to show any embedded tailoring, thus letting you verify that you have declared all of the variables you plan to use in the tailoring.
- Help - this shows the help text for the question, also syntax-highlighted.
- Tip - this shows the internally generated validation tip, based upon the question attributes. This same tip is used in all survey styles, plus in the printable survey and data entry screens.
- Question Attributes - this shows a table of all relevant question attributes for this question. Attributes that might be equations are syntax-highlighted so that you can validate their accuracy.
Rows are color coded as follows:
- Groups - are shown with a light grey background
- Questions - are shown with a light green background
- Sub-Questions - are shown with a pale yellow background
- Answers - are shown with a plain white background
Answers have an additional attribute in the Relevance column
- Value - this is the default internal value used by calculations. If you are using Assessments, this will be the assessment value. Otherwise, this will be the same as the Answer name.
At the top of the page, there is a summary message. If all is well, it will say "No syntax errors detected in this survey", or "This group" or "This question", "by itself, does not contain any syntax errors". Otherwise, it will say "X questions have syntax errors that need to be corrected".
Each question that has syntax errors will be color-coded red in the leftmost column (e.g. Q-1). The following errors are common:
- Bad variable name - if your question code has non alpha-numeric values, or is used more than once in the survey, it is an error. However, to support legacy instruments, this is tolerated for now. Such questions will be shown as with red qcode names, boxed in purple, but the question itself is not flagged as an error.
- Undefined variable - if you have not defined all of your variables, or are using a mis-typed array_filter (or different sets of answer options for array_filter), then some of your validation questions will show errors. Undefined variables are shown in red text, boxed with a red line.
- Bad syntax - as you start to use relevance equations, you may use too many or too few parentheses. Such syntax problems are highlighted and boxed in red. If you hover the mouse over any such red-boxed text, a tool-tip will describe the error.
Here are many good examples of using syntax highligting.
Here is an example of a survey that has errors:
The following types of warnings and errors are present:
The variable name will be color-coded in red and surrounded by a red line. If you hover your mouse over the variable name, it will say "Undefined variable".
Obsolete variable names
Starting in 1.92, the preferred naming of question codes is [a-Z][0-9a-Z_]* - meaning it must start with a letter, and can contain letters, numbers, or underscores. However, for backwards compatibility, we are are only flagging invalid variable names via a warning for now. We expect to give the option to disallow those old variable names in the future, since many of the new features (along with proper SPSS / R export) require unique and properly formatted question codes.
These warnings appear as a pink line surrounding the variable name in the Name [ID] column. If you hover the mouse over the variable name, you will see this message:
Clicking on the link for the variable name opens up a new browser tab in which you can fix the error. You can have as many tabs open as you want, so you can simply click on all of the errors, fix and save them all, then re-run the Show Logic File report to confirm that you have fixed all of the errors.
Speeding editing and validation
All of the syntax-highlighted text has embedded tool-tips, and is clickable.
- Tool Tips
- Functions - hovering the mouse lets you see the purpose and syntax definition of the function
- Variable Names - hovering the mouse lets you see the position (group and question sequence), question text, and allowable answers for the question
- Variable Names - clicking on the variable name opens a new window that lets you edit the question. This makes it easy to navigate and verify logic - simply keep clicking on variable names of relevance or validation criteria for the question to see where they come from and how they are used.
The following examples are taken from the Expression Manager Sample Surveys. You can find screen-shots of running surveys, explanations, and downloads on that page.
Body Mass Index
Here are screen shots of this example.
This is the Question-Reorder view of the Body Mass Index calculation. You can see the relevance equations (which are after the variable name, within square brackets - like  for weight, and [!is_empty(height) && !is_empty(weight)]) for BMI. You can also see the tailored text of the questions.
The logic file provides additional information, including question type, and lists of subquestions and answer options, plus all question attributes.
Note that the (VALIDATION: ) section is automatically generated for you from the question attributes you have selected. This lets you see what validation logic is being applied, but you never have to specify or edit that validation logic directly.
This has a good example of nested if() statements to generate the weight_status.
Here are screen shots of this example.
This example shows the subquestion validation logic that is automatically generated when you use array_filter and array_filter_exclude. This example also shows how you can substitute the tailored "Other" value (the answer for Q02_other is Q01_other).
Q05 in this example shows simultaneous use of array_filter and array_filter_exclude on Q01 and Q02, respectively. This is part of how this example demonstrates cascading array_filter capabilities. Note that one of the main reasons for showing the question and subquestion level VALIDATION criteria is to help ensure you have not made any typos in specifying the array_filter or array_filter_exclude variable names (or in case you use different variable names for your list of filtered subquestions). If you have such typos, you all of the invalid variable names will appear in red indicating that they are undefined, letting you quickly fix the problem.
This example demonstrates dynamic, cascading relevance logic to control display of question visibility. You can download this example here.
All questions starting with Q-20 (age) have a relevance equation. Q-22 (yearsMarried) also has a validation equation which uses the max_num_value_n advanced question attribute, but passes it an equation (age-5) rather than a fixed number.
Also note that questions kid1-kid3 are mandatory. However, if the question is not relevant, the mandatory requirement is ignored (which makes sense, since why would you force someone to answer an irrelevant question). These questions can be considered "conditionally mandatory".
This example shows how group-level relevance appears in the logic file. Here are screen shots of the working demo.
As you can see, the group-level relevance equation (cohabs > 1 && p1_rel != "") appear in the grey Person 2 row for G-2.
You may also notice that all of the questions are mandatory. However, if the group is irrelevant, so are all of its questions, so those questions are only truly mandatory if the group is relevant.
Additionally, note that the TIP messages are also automatically created for you. They are organized by value range (min/max), sum value range (min/max/equals), number of answers (min/max), etc. Sometimes you want to validate an answer range but don't want to display what might appear to be silly validation tips to the user. In such cases, you can use the hide_tip advanced question option (as in this case, to avoid telling the user that the age must be between 0 and 115 unless they try to enter a bad value).
Comma as radix (decimal) separator
This demo shows that although LimeSurvey 1.92 now fully supports use of comma as the radix (decimal) separator at run-time, you must still use a decimal as the radix separator at design-time (e.g. when specifying min/max values in advanced question attributes). The working example is found here.
Also, remember, the VALIDATION logic is created for you automatically from the advanced question attributes. The equations can look overwhelming, but you don't need to worry about them.