Skip logic and scoring

Two features turn a flat list of questions into something that behaves. Skip logic decides which questions a person is asked at all; scoring turns the answers they give into a number.

Showing a question only when it applies

Skip logic belongs to the question you want to hide. Open that question in the builder and give it one or more conditions. The question then appears only when those conditions are met, and stays out of the way when they are not.

In the builder, a condition looks back: it tests the answer to a question that comes earlier in the form. There is no separate logic screen: you set the logic where you set everything else about the question. A form imported from XLSForm may carry a rule that reads a later question; the import names every such rule, and so does the Design tab, so that you can check it.

A section can carry a condition too, set under Show this section only when in the section's settings. It uses the same conditions a question does, every question in the section follows it as well as its own, and the questions it tests must come before the section.

The conditions you can write

Each condition names an earlier question, an operator and, for most operators, a value to compare against. These are the operators:

• equals
• not equal
• contains text
• is blank
• is answered
• greater than
• greater or equal
• less than
• less or equal
• contains option
• does not contain option

Add more than one condition and you choose how they combine. With and, which is what you get unless you change it, every condition must be true before the question appears. With or, any one of them is enough.

Scoring a form

A form uses one scoring method, chosen on the form itself:

• No Scoring, which leaves the answers as answers.
• Knowledge Test, where you supply an answer key and the platform marks each response against it.
• Rating Average, for a satisfaction survey where the useful figure is the mean of a scale such as 1 to 5.
• Rating Sum, where the useful figure is the total across the scored questions.

The form also carries the maximum of its rating scale, which is 5 unless you change it, and a pass threshold, which is 70 percent unless you change it.

Each question is then set up in the builder. Mark it as scored, set the correct answer, and set the points it is worth, which is 1 unless you say otherwise. For satisfaction scoring you also mark which answer counts as a positive one, and whether the question is counted in the score at all.

You do not calculate anything yourself. When a response becomes complete the platform works out the score, the maximum that was available and the percentage, and stores all three against that response.

Domains and reverse questions

In Rating Sum and Rating Average, a choice question is scored by its chosen answer read as a number: answers named 0, 1, 2 and so on count as that number, yes counts as 1 and no as 0. The answer is scaled from the highest choice to the points you set on the question, so a question with choices 0, 1 and 2 worth 6 points adds 0, 3 or 6. A question whose choices are not numbered cannot be scored, and the import report names it. A number question counts its value as typed, out of its maximum, times its points.

A form can split its score into domains, such as governance or finance in an organisational capacity assessment. Each scored question names the domain it counts towards, and every complete response then stores a score for each domain beside the form score. The domain scores are shown on the response and added as columns to the responses export, and an indicator that averages a form score can read one domain instead of the total.

Some questions are scored in reverse, where a higher answer is better, such as the protective factors beside the risk factors in a best interest assessment. A reverse question counts only in its own domain: its domain score is reported beside the form score and is never added to it or taken away from it.

In an XLSForm workbook the cl::domain and cl::reverse columns set these, and a commonlynk_domains sheet gives each domain its label in each language. If the workbook states the most the form can score, the import adds up what the scored questions can score and tells you if the two disagree, rather than storing a number that could be wrong.

Comparing a pre test with a post test

Two forms can be paired: one marked as the pre test, the other as the post test. Ask the same people the same questions before and after your activity, and the platform does the comparison for you.

When a post test response is completed, the platform looks for that same person's most recent completed pre test and works out three figures:

• The simple gain, which is the post test score minus the pre test score.
• The normalised gain, which is the share of the ground still left to gain that the person actually gained.
• The relative gain, which is the increase expressed as a percentage of the pre test score.

Two of the three have limits worth knowing. The normalised gain has no value when the pre test was already 100 percent, because there was nothing left to gain, and the relative gain has none when the pre test score was zero, because an increase cannot be expressed as a share of nothing. All of them appear on the response detail page for the post test response.