Jump to other news and events
Purdue signature
 

Test Construction and Grading

Back to Teaching Topics

Good quality instruction not only includes the use of effective learning activities, but assessing the degree to which students have attained learning outcomes. Although we typically think of designing tests and assigning grades as occurring after course goals, learning outcomes, and learning activities have been specified, these aspects of instruction are closely linked; so it's helpful to consider them all together in the early stages of planning your course.

Writing good tests is not easy. It takes a great deal of practice. The following are some tips and general guidelines you should keep in mind when constructing tests.

  • Test what you intend to test
    "Nothing in the content or structure of [a test] item should prevent an informed student from responding correctly. Similarly, nothing in the content or structure of [a test] item should enable an uninformed student to select the correct answer". This quote by Gronlund (1998) strikes at the core of test validity, to test what you claim to test.
    Some threats to test validity include unclear test directions, confusing and ambiguous test items, overly complicated vocabulary and sentence structure, cheating, and unreliable grading (Gay, L.R., & Airasian, P. (2000). Asking an outside person to read your test for clarity prior to giving it, or asking your students to let you know which questions were unclear to them can provide you with valuable feedback for improving your test items. If you have your test scored by the Data Processing Center in Stewart Center, they can provide you with an item analysis of the exam which is another way to identify problematic test items.
  • Align the test with the learning outcome
    Alignment of the test with the learning outcome is a key factor for good testing. Achievement of some outcomes might be better tested with an objectively scored test (e.g., multiple choice, matching, T/F) whereas testing the achievement of other outcomes requires the use of a subjectively scored test (e.g., short answer, essay, project, and performance).
    For example, the outcome Students can explain differences between presidential and congressional legislative power, requires students to provide the answer because they have to explain the difference. Therefore, an essay would be an appropriate way of testing students' achievement of the outcome because an essay requires students to provide rather than to select an answer. However, if the outcome stated Students can recognize differences between presidential and congressional legislative power, multiple choice questions could be used for testing. Selecting the correct answer from a set of possible options allows students to demonstrate that they can recognize differences between the presidential and congressional legislative power.
  • Be fair
    You need to give students a fair chance to succeed on their tests. Among other things, fairness can be facilitated through transparency of expectations for performance, adequate test preparation opportunities, valid tests, transparency of grading criteria and approach, and reliable and unbiased grading. Students should know what is expected of them in order to obtain a given letter grade and their grades should be a reflection of their performance and not a reflection of extraneous and irrelevant factors.
  • Triangulate
    Despite your best efforts, there are factors beyond your control that can contribute to invalid test results. For example, a student might perform below his/her skills on a given test due to serious personal problems such as financial pressures or health challenges. Or a student might underperform due to having a proverbial bad day.
    At the same time, a student might perform better than his/her actual skills due to the sheer luck of correctly guessing some of the answers. To decrease the likelihood of test performance being a reflection of irrelevant factors, it is important to provide students with multiple opportunities for demonstrating their performance level. The likelihood of eliciting students' 'true' performance level can also be increased by providing students with diverse opportunities for demonstrating their skills. For example, some students might generally do better on exams whereas other students might generally do better on projects. Including both, exams and projects in your course will accommodate those differences between students.
  • Unleash the learning power of testing
    While the primary purpose of testing is typically for students to demonstrate their skills and knowledge, testing can also be a powerful tool for facilitating student learning. For example, when returning the graded tests to students, you can spend some time on reviewing some of the most common mistakes students made on the tests. Providing students with specific feedback on written assignments is another way of facilitating student learning. Students shouldn't only know the letter grade for their work but also the reasons why they received that grade. This provides them with information that they can utilize for building on their strengths and for improving their weaknesses.

[ References ]   [ Links ]   [ Words of Wisdom ]