Considering factors other than task type when creating test questions

What task you'd like to set candidates in your test question is not the only consideration you will have when thinking about how to create it.  

For each question, there is likely to be a range of considerations, but you will almost certainly want to consider the following two general factors when designing your questions, and thus choosing which interactions you would like to use: 

Amount of Guidance given

The GUIDANCE spectrum refers to the level of assistance a candidate receives in the test item when trying to reach an appropriate response. It ranges from “guidance given“ at one end to “no guidance given“ at the other. 

Interactions at the first end of the spectrum provide a very limited number of options from which the candidate chooses a response, thus giving considerable guidance. The most "guided" response is a true/false question, where the candidate knows that one of the two possible responses supplied is the correct one. Around the middle of the spectrum are interactions which present a significantly greater number of options as possible responses (one of which is the desired response), giving some, but more limited, guidance. At the other end of the spectrum no options are provided at all; the candidate has to draw on his or her own knowledge entirely.

This spectrum can be used to take into account the level of knowledge of the candidates; the less guidance given, the more difficult the test is.

Amount of Language used

The LANGUAGE spectrum refers to the level, or amount, of language used in the test item. It ranges from "language-based" to "non-language-based". 

Interactions at the first end of the spectrum are based purely (or almost purely) on text. This includes the prompt, any stimulus provided, and the options indicating possible responses, if supplied. Around the middle of the spectrum there might be interactions involving a mixture of images and limited text. Because the prompt, at least, is usually expressed in words, there are almost no test questions which contain absolutely no language at all, but the interactions towards the other end of the spectrum are all based on an image or a graphic. 

This spectrum can be used to take into account the language levels of the test-takers. If the language used in the test is not their first language, for example, or if they have reading issues, the interactions at the non-language-based end of the spectrum can be used to ensure that it is the content rather than language ability which is being assessed.