This research paper investigates whether varying the difficulty sequence of exam questions — easy-to-hard, random, or hard-to-easy — can alleviate test anxiety and improve student performance. Drawing on prior literature from Towle and Merrill (1975), Munz and Smouse (1968), and Tobias (1985), the authors hypothesized that an easy-to-hard question sequence would benefit high-anxiety students. Forty-two adult participants completed Spielberger's Test Anxiety Inventory and one of three exam versions. A 2×3 factorial ANOVA revealed a significant main effect of question difficulty sequence on performance, but no significant effect of test anxiety level alone and no interaction between the two variables. The findings partially contradict the original hypothesis and prior assumptions about anxiety-reducing exam structures.
Tests in school are designed to be instruments that can provide educators with an opportunity to assess the skills and knowledge of a student on a particular subject. Students who experience a great deal of anxiety during tests may not be able to perform as well on their examinations as they might were they not suffering from test-related anxiety. "In evaluative situations, highly test-anxious individuals perform more poorly than low test-anxious persons," note Jerry L. Deffenbacher and Susan L. Hazaleus of the Colorado State University Department of Psychology. "Their performance, however, is not a simple function of ability but varies with evaluative stress." They further observe that when stress levels are lower, students perform well, but when stress levels are higher, students tend to have more difficulty even with common questions.
Over time, instructors have been alerted to the difficulties experienced by students with high test anxiety. They were advised to create tests that progress from easy questions to more difficult questions, in that order (Towle & Merrill, 1975). The rationale for this type of exam structure is straightforward: students who experience high levels of test anxiety are less likely to become discouraged if an examination starts with simpler, more manageable questions. Under this testing model, questions would grow progressively more difficult as the examination continued. The inspiration for a graduated test sequence came from the idea that students are likely to perform better when not subjected to conditions that provoke anxiety. Beginning an exam with difficult questions has the potential to discourage students and increase their level of test-related anxiety. Starting with easier questions may bolster a student's confidence and help alleviate some of the anxiety that difficult early questions can produce.
After reviewing previous hypotheses and studies, we concluded that it would be natural for students to feel more at ease with an examination that begins with easier questions. As questions progress in difficulty, it was our expectation that students would become more confident in their ability to perform well, thus experiencing less test-related anxiety than usual. The hypothesis for this experiment is therefore that students who receive an easy-to-difficult question sequence will perform better on an examination compared to those who receive a different question-difficulty structure.
Test anxiety and question sequence is an issue that has been studied for many years. "Effects of Anxiety Type and Item-Difficulty Sequencing on Mathematics Test Performance" by Nelson J. Towle and Paul F. Merrill was published in the Journal of Educational Measurement's Winter Edition in 1975. It examined the impact of question sequencing on people with test-related anxiety. In their report, they suggest that, "Although the [previous] studies have produced no evidence to support the present convention of ordering items in an easy-to-hard sequence, several of the cited studies found interactions between item-difficulty order and anxiety type" (Towle & Merrill, 1975).
Towle and Merrill divided students into four groups based on anxiety type. Facilitators comprised approximately 25% of the sample population. Their classification was drawn from Munz and Smouse (1968), who defined four anxiety types using scores from the Achievement Anxiety Test (AAT):
"Facilitators were those students, making up about 25% of the total sample, whose facilitating anxiety scale scores were considerably higher than their debilitating anxiety scale scores. Debilitators were defined as those students, about 25% of the sample, whose debilitating anxiety scores were considerably higher than their facilitating anxiety scores. For all remaining subjects, the two scale scores were summed and ranked. The subjects scoring above the median of the summed scores were defined as high-affecteds, and the subjects in the lower half of the distribution were defined as non-affecteds. Munz and Smouse (1968) discovered a significant anxiety-type by item-difficulty sequence interaction effect on performance scores on a final exam in an introductory psychology course. On a random item-difficulty sequenced form, facilitators and high-affecteds scored significantly higher than the debilitators and non-affecteds. On an easy-to-hard form, facilitators scored significantly higher than the other three anxiety types. There were no significant differences among anxiety types on a hard-to-easy item-difficulty sequenced final examination" (Towle & Merrill, 1975).
Their hypothesis was that if students were more or less aroused than normal during an exam, their functioning would be impaired. They used an easy-to-difficult question sequence similar to the one employed in the present study.
A related study was published in the Journal of Educational Psychology in 1968. "Interaction Effects of Item-Difficulty Sequence and Achievement-Anxiety Reaction on Academic Performance" by David C. Munz and Albert D. Smouse of the University of Oklahoma was designed to "test the assumption that individual reaction to test taking mediates the effect of item-difficulty sequence on performance" (Munz & Smouse, 1968).
In his 1985 article in the Educational Psychologist entitled "Test Anxiety: Interference, Defective Skills and Cognitive Capacity," Sigmund Tobias investigates whether test anxiety affects students' ability to perform to the best of their ability on an exam. "It has long been assumed that test anxiety interferes with students' recall of prior learning on examinations," states Tobias. "This so-called interference model recently has been challenged by an alternative deficit explanation advanced by a number of researchers. The deficit hypothesis assumes that the lower test scores obtained by test-anxious students are attributable to inadequate study habits or to deficient test-taking skills rather than to interference by anxiety." The conclusion of his study was that students' test performance can be negatively impacted by anxiety itself, and not only by a lack of knowledge or poor study habits.
Forty-two adults from the state of Wisconsin (24 male, 18 female) ranging in age from 19 to 23 (M = 23.05, SD = 5.52) participated in the study. Participants were 97.6% Caucasian and 2.4% Hispanic. Participation was voluntary with no monetary compensation.
A two (test anxiety: high, low) by three (question difficulty sequence: easy-to-hard, random, hard-to-easy) between-subjects design was used. The dependent variable was test performance.
The study used Spielberger's Test Anxiety Inventory (TAI) to assess test anxiety among participants. The TAI is a 20-item inventory producing scores for total test anxiety and its two main components: worry and emotionality. It is a self-report, four-point scale (1 = almost never, 4 = almost always) on which participants rate how often they experience particular symptoms of test anxiety. Both eight-item subscales range from 8 to 32, with the total test anxiety scale ranging from 20 to 80. The TAI subscale and total scale correlate highly with other well-established test anxiety measures (Spielberger, 1980). The TAI and Sarason's Test Anxiety Scale (TAS) have a strong correlation of .82.
Three forms of the exam were prepared. Each form consisted of the same twelve open-ended questions and differed only in the sequence of question difficulty (easy-to-hard, random, hard-to-easy). All questions were selected at random from McGraw-Hill's SAT preparation book, and the multiple-choice format was removed to make the questions open-ended. McGraw-Hill's SAT book uses question analysis from previous exams to rate question difficulty on a five-point scale. Four questions rated as 1 were selected as easy, four questions rated as 3 were selected as moderately difficult, and four questions rated as 5 were selected as hard, for a total of twelve exam questions.
A questionnaire consisting of seven items addressed basic demographics including age, gender, and ethnicity, as well as confound checks. The confound checks included participants' math ability (math GPA), the number of math courses taken in the previous two years, and participants' assumptions about the true purpose of the experiment.
The experimenter conducted the study at several locations in Madison, WI and Green Bay, WI over the course of two weeks. Before the experiment began, participants were told that the experiment concerned differences in math ability across majors, in order to prevent bias. Each participant was given a consent form and reminded that they could stop participation at any time, leave questions blank, and that there was no compensation for participating.
Before taking the exam, each participant completed Spielberger's 20-item TAI. A median split was then performed to assign participants to either the high test anxiety group or the low test anxiety group. The 50% of participants who scored above the median were assigned to the high test anxiety group; the other 50%, those who scored below the median on the TAI, were assigned to the low test anxiety group.
Upon completing the TAI, participants were randomly assigned to one of the three question-sequence conditions and given the corresponding exam. Participants were allowed to use a calculator and were given a time limit of eighteen minutes to complete the exam. After the exam, participants completed the short demographic and confound-check questionnaire.
To determine whether participants' academic major was a confounding variable, a one-way ANOVA with an LSD post-hoc test was performed to examine whether major had an effect on exam performance.
A 2×3 factorial ANOVA was performed to analyze the main effects of test anxiety and question difficulty sequence, as well as their interaction, on exam performance. An LSD test was conducted for the three-level independent variable (question difficulty sequence) to determine whether cell means differed across the easy-to-hard, random, and hard-to-easy conditions. Three contrast t-tests were conducted to examine the interaction between each question sequence condition and test anxiety level (high vs. low). The hypothesis predicted that participants with low test anxiety would outperform participants with high test anxiety on exams with random or hard-to-easy question sequences, but that there would be no difference in performance between the two anxiety levels on the easy-to-hard sequence.
"ANOVA findings on sequence and anxiety effects"
"Interpretation of unexpected null interaction results"
"Cited works in APA format"
You’re 70% through this paper. Sign up to read the remaining 3 sections.
Sign Up Now — Instant Access Already a member? Log inAlways verify citation format against your institution’s current style guide requirements.