Can we design reliable tests?

When I started reading this chapter, one of the preview questions called my attention: Do you think it is a good idea to give students lots of times to complete a test?

Regarding my experience in designing tests, I always had a bad time when designating time to the tests. Sometimes I tried to state the time depending on how much time I would spend answering the test, because I thought that if they studied enough, they wouldn’t need much more time. But, is this reliable? Is this a fair way to give time to the tests? I always asked myself that.

Other times, I just gave my students a random time span to take the test.  Then, I would frequently check their progress. If I saw they needed more time (or they asked me for more time) I would enlarge the time limit.  For me, it felt better this way. Nevertheless, there were times of disorganization because most of the students would ask for extra time, and as I was flexible before, I should be flexible with all of them.  According to Genesse & Upshur (1996), judging each skill by real contexts would be a good strategy to allocate the correct time to a test.

Regarding instructions, I think this is one of the reasons why many tests fail their purposes. Sometimes we don’t know how to give instructions and we don’t know exactly how much information should we include in the instructions of the test. For example, we don’t usually include the purpose of the test, the students’ response we expect, or in writing tests, what exactly we are going to assess (like adding a rubric), and so on (more aspects are included in this chapter). In my case, I haven’t told my students the purpose of the test, or the objectives, or those things.  As I have worked with kids only, I consider they wouldn’t give it much importance to this. They just want to answer the test as soon as possible and get a good grade, of course. I would like to know, do you usually give your students all this important information and instructions before starting to implement your tests? Or you really don’t? Do our students really care about knowing all this information?

As said by Genesse & Upshur (1996), tests that are “carefully constructed, edited, tried out, and revised” are more reliable. A test that is prepared the night before its implementation will certainly be lacking of reliability. Sometimes we don’t give tests design the importance it requires, so we should start doing it if we want successful assessment.

*This entry is about Genesse & Upshur’s Classroom-based Evaluation in Second Language Education, chapter 11. 
**Image retrieved from http://minds-on-math.blogspot.com/2012/10/timed-tests-and-development-of-math.html