[tomxor]

That's a total no-go unfortunately since WCAG-AA accessibility is a non-optional requirement in edtech.

The solution in general is in improving the quality of the content, using a more sophisticated format of questions that requires work rather than mere knowledge (which is also far better for formative assessment, but most platforms are focused on summative, particularly in the US)... independent multi-choice is always easily gameable in some way - if there is no better format of question available, the best that can be done for multi-choice is to have a massive pool from which you randomly draw a different subset of questions for each student, and limit the number of attempts to make it impossible to fully scrape... even then, a smart group of students may pool and share their feedback as they progress.

The tricky thing is that you want to encourage such behaviour, helping each other learn, and although in some people's eyes this is purely cheating, it's not dissimilar in spirit.

[expensive_news]

Bad multiple choice sets reminded me of a history class my friend took in summer school. All the questions were multiple choice, and the teacher was extremely lazy and decided that the answer bank for every question would just be randomly drawn from the population of all answers. The end result is that most the questions end up like :

What year was the Declaration of Independence signed?

a) Martian Luther King Jr. b) The Spanish-American War c) 1776 d) The New Deal

Needless to say everyone aced the test.

[Zak]

Aside from being easy to cheat on, multiple choice tests are bad measures of knowledge/ability.

I'm good at them. I can often infer the desired answer from the phrasing of the question and answer without actually knowing enough about the topic to answer correctly in a free response format. I can almost always eliminate a wrong answer or two that way even if I can't necessarily pick the winner, improving my odds.

Some people are bad at them, especially when the test demands the "best" of several defensible options.

In either case, the test results in an unfair and inaccurate estimate of the evaluatee's performance.