In any assessment, but especially certification exams, it is extremely important that we are fair to all test takers. Fairness is often considered in relation to individuals who have special educational needs but is actually much broader. A test may be unfair because the level of English is far beyond that of the test taker when it doesn’t need to be. It could be unfair because it requires knowledge of a specific geographic location or activity. All of these can cause what we call construct irrelevant variance. We want to avoid this as much as possible, so keep in mind the following as you write.

Please also take a look at this blog post which gives a great overview of how to handle bias and stereotyping.

Our test takers come from an international audience. We use US English when writing our items but also think about the language level. We are not testing English language ability. Keep sentences short, in the active voice, and use simple vocabulary. There are a range of useful online tools that can help you out. Take a look at Hemmingway App to get you started. You should target US grade 8 or below.

Some topics may introduce bias by causing distress or offense to test takers and must be avoided. This includes (but is not limited to):

Commonly in teaching materials we are told to ensure that we balance the use of gender. For DataCamp assessments, we ask that you avoid the use of gender in the context of items entirely.

Why? Our tests are adaptive. This means that we have no control over which combination of items our test takers see. We can have carefully crafted a balanced set of items only for the test taker to see a subset that appears to be imbalanced.

When it comes to gender in data, ask yourself if it is really necessary to test the concept. If you do feel it needs to be used, ensure that you provide a balance and do not conform to gender stereotypes e.g. all female names have stereotypically female roles.

Expect to be asked to justify the decision to include gender or other personal characteristics.

Any topic that is culturally specific must be avoided. Something that requires knowledge of a culture can be a big cause of bias due to the fact that not all test takers will be aware of it.

It often seems that sporting data fills data-related training courses, probably because it is often freely available. While it may be ok to use in teaching materials we ask that you do not use it in test items. This may seem like a very harsh policy but sport can have regional differences that lead to bias. A simple example is Football. In most of the world, it is not the same as Football in the US. I may lose lots of time on an item being distracted by not having understood this difference, but it wouldn’t tell me anything about my data skills.

This can be one of the hardest to ensure we stick to, but make sure that you are not testing general knowledge or giving an unfair advantage to individuals with specific knowledge.

One of the most common examples of this is geography. It is common to use US-centred data and assume a general knowledge of US geography. Avoid this as much as possible. And remember, our audience is international, so try to avoid repeated use of US, Canada, UK etc. and include examples that represent the huge diversity of our test takers.