Considerations in Test Construction
No matter how intuitive psychological tests seem, measuring a construct is a little more than throwing a questionnaire at participants. The construction of a test is a process that requires constant revision and consideration of outside factors that may affect the data you collect. Even constructs that seem easily understood are more complex than it looks upon first glance.
For example, trust is a construct that forms at least a small part of our everyday interactions with countless people that flit in and out of our lives. Considering how pivotal this can be in most people’s lives, interpersonal trust is a construct that intrigues me greatly, and so I was motivated to test it.
Rotter (1967) defines interpersonal trust as “an expectancy held by an individual or a group that the word, promise, verbal or written statement of another individual or group can be relied upon”; this expectancy is the basis of the trust, despite the risk of personal vulnerability that Johnson-George and Swap (1982) mention is inherent in the interaction. These definitions show the value of testing the construct as well as establish a foundation upon which to operationalize it. Most people have many in their lives that they trust; though here, what varies most are the levels of trust that one places in one person versus another. Based on my own experiences, I posited that people would be inclined to trust closer others, like a best friend, a sibling, or a significant other, more than they would removed and authoritarian others, like a parent, a boss, or a salesperson.
However, testing sounds a lot easier than it is. In operationalizing interpersonal trust and writing the questions for my test, I soon realized that I had to have an equal amount of questions for each of my hypothesized trust relationships; otherwise, the scale might be unbalanced in the direction of one of the relationships. In addition to this, it was essential to me to have reverse-scored questions, as I believe that answering every question the same way lends itself to participants answering every question the same way because they believe that one answer is the “correct” one; this, when combined with the need for equal amounts of questions for each relationship and the administration time limit of 3 minutes, led to every trust relationship having two questions each, with one of the two being reverse-scored. I also had to keep in mind that not everyone has a sibling, a significant other, or a boss, and include contingency options for if that was the case for one of my respondents. Finally, picking a scale to use for my participants to respond with came along with deliberation on how it would affect my data (would the ‘neutral’ option in my 5-level Likert scale skew my responses in that direction?), but ultimately, I had to choose what would fit within the time limit but was still descriptive enough for what I considered necessary.
All of these factors changed the way the test was constructed and presented, and the final product was much shorter and less comprehensive than I had wanted it to be. Regardless, the experience captures the constant process of revisions that constructing a test entails, as well as the consideration of outside factors that might have affected the data that I collected. In this way, actually constructing a test gave me an insight into the reality of psychological tests that simply reading a textbook would never have given me.
References
Johnson-George, C. and Swap, W.C. (1982). Measurement of specific interpersonal trust: Construction and validation of a scale to assess trust in a specific other. Journal of Personality and Social Psychology, 43(6), 1306–1317. DOI: http://dx.doi.org/10.1037/0022-3514.43.6.1306
Rotter, J.B. (1967). A new scale for the measurement of interpersonal trust. Journal of Personality, 35(4), 651–665. DOI: 10.1111/j.1467–6494.1967.tb01454.x
