The Test Anxiety Test, Redundant?

For some students, test anxiety is as much of a part of their academic career as ramen or sweatpants. Especially at the university level, examinations can have higher stakes than say an elementary spelling test (although those were not easy either). Recently, as part of my test construction assignment, I gave myself the interesting task of not only operationalizing test anxiety, but of also creating a scale to measure it.

Test anxiety as a construct is already rather well researched. Spielberger’s Test Anxiety Inventory is currently the most used and well- known scale for measuring test anxiety. The Test Anxiety Inventory consists of 20 items with 4 possible response options ranging from almost never to almost always (Spielberger, 1980). Several other scales exist as well, most very similar to the model Spielberger has set.

Although originally very optimistic, creating a new scale was more challenging than I thought it would be. My goal was to focus on increasing practicality while maintaining enough face validity that students would respect my scale and fill it out seriously as well as ensure that I had a decent amount of reliability. I also really had to stress how important honesty was in filling out the scale because topics like anxiety can be very personal. Many questions raced through my brain in the early stages of my scale’s conception.

Q: How can I make students take my test seriously?

A: As previously mentioned, this can be a face validity issue. Face validity is the extent to which an item or a test is subjectively viewed as covering the concept it is supposed to be measuring. No deception involved in this scale, pretty much every single item had the word “test” and/ or “anxiety” in it and with the title being “Test Anxiety Scale”, I was being pretty transparent about what my scale was attempting to measure. I hoped that this would show participants that I knew what I was trying to measure and I wanted them to know it too.

Q: Do these items make sense?

A: My answer to this was hopefully. I based my scale off of Spielberger’s, which is widely used for its reliability and validity. For practicality reasons, I reduced the number of items, but kept the same ratio of the two variables Spielberger measured: worry and emotionality (1980). Other items that I found similar were combined. I used a random order to ensure that not all items assessing worry were at the top and emotionality at the bottom so that when it would come time to verify my split- half reliability, there would be no issues.

Q: How can I make a one- time measure of test anxiety as reliable and representative of a student’s stress anxiety profile as possible?

A: This was tough because participants filled out my survey during prime midterm season. One can only assume that stress levels were much higher than usual. At first concerned with how reliable my results would be due to the fact that students were in a high- stress period, I finally calmed myself down because this was actually the ideal situation for them to be in. Because I was not looking for overall anxiety levels, I actually needed midterms to be as salient as possible in students’ minds. As seen in other retrospective research, we sometimes forget how bad or good a situation actually was. Once midterms are over, I usually look back and think to myself that it was not that bad (although at the time, the stress was terrible).

Overall, creating a test anxiety scale was ironically pretty anxiety- inducing. I did not want to activate any superfluous test anxiety by merely bringing up the topic and triggering students but I also needed to access those feelings.