| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
REVIEW ARTICLE |
Eastern Illinois University, Department of Psychology, Charleston, IL.
Address correspondence and reprint requests to Dr. Jeffrey R. Stowell, Eastern Illinois University, Department of Psychology, 1151 Physical Sciences, 600 Lincoln Avenue, Charleston, IL 61920. E-mail: jrstowell{at}eiu.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
Key Words: test anxiety, • perceived stress, • psychoneuroendocrinology.
Abbreviations: DC = difficulty coefficient.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
|
LIMITATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
Examination Difficulty
The types of examinations referred to are more challenging than other types of examinations because they are usually longer, cover more material, and carry an increased risk of failure in terms of a student’s course grade. Choosing difficult examinations maximizes the opportunity of finding statistically significant effects of examination stress, but an overrepresentation of these types of examinations in the literature may exaggerate differences between examination and nonexamination periods. Even presumably difficult examinations may not result in significant increases in perceived stress for some students (13, 20).
Confounds in Within-Subject Comparisons
The following potential confounds may occur in within-subject comparisons between examination and nonexamination periods and should be measured to allow statistical adjustment in data analyses; some will also be relevant in between-subject designs. These include unknown stressors occurring on nonexamination days (2, 4) , coincidence in timing between end-of-semester examinations, and other stressful events (eg, preparing to graduate, moving, or seeking employment), changes in health practices (smoking, alcohol consumption, sleep, exercise, nutrition, medication use, etc.) (21), and sex differences in endocrine function (22–24).
Timing of Repeated Measurements
When considering physiological changes associated with the examination, timing is very important. Catecholamines have short-acting effects, with a half-life of less than 10 minutes, while the effects of steroid hormones may last for hours (25). Cortisol levels may take at least 10 days to return to normal after examinations (5). Further complexity arises with the timing of endocrine measures that exhibit diurnal rhythms because the increase and decrease of a hormone may be caused by internal rhythms rather than external stimuli.
|
RECOMMENDATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
Objective Index of Examination Difficulty
In addition to using a subjective measure of examination difficulty, it is helpful to report an objective index of test difficulty. One option would be the test mean and standard deviation. Preferably, a difficulty coefficient would be reported, such as one proposed by Bruce (29). Based on the assumption that increasing test difficulty results in increased variability of test scores and a lower mean, the difficulty coefficient (DC) is equal to (100 multiplied by SD of scores at or above the mean) divided by the mean test score. The variability of scores above the mean is used because the variability below the mean can be affected by other factors besides test difficulty (lack of studying or motivation, missed examination, etc.). A higher DC indicates a more difficult test. If test difficulty was so high that everyone performed very poorly, the variability in scores could theoretically decrease, lowering the DC. Despite this limitation, having a fairly robust and objective index would allow comparison of test difficulty across studies, which may help explain some of the contradictory findings.
Experimental Design
The choice of experimental design depends on whether the researcher is more concerned about individual differences over time, group differences, or both. In the latter case, a mixed design would be best. The control group would be participants who were not undergoing examination stress, but otherwise would have the same level of perceived stress as the experimental group during nonexamination periods. One suggestion would be to use a control group that is performing routine schoolwork. This specific type of control group has been rarely used (15) but would help determine whether or not the observed changes are the result of a generalized stress response or some specific feature of the academic examinations that is different from other schoolwork.
One advantage of using a longitudinal study is that participants serve as their own control, reducing variability in their scores. While many studies have used within subject designs, surprisingly few have focused on what is happening during the examination itself. Collecting physiological data after the examination may be too late to capture certain changes. By including multiple measurements, one could determine whether the changes associated with the examination were caused by the anticipatory component of the examination, completing the examination, or something after the examination.
Depending on the methods used to measure physiological changes, multiple measurements during the examination may be disruptive. However, with techniques currently available, these interruptions can be minimal, if not unnoticed. For example, research personnel can unobtrusively collect blood for immune measures via tubing connected to indwelling catheters (30) or measure hormone levels in saliva with wristwatch-prompted timing using the salivette method (31). Repeated administration of self-report surveys is more likely to be disruptive during an examination but is usually not as costly.
Analyses of Interindividual Variability
During the same examination, there can be enormous variability among individuals in both perceived stress and physiological reactions (4, 15, 18, 32). Physiological reactions may differ among individuals despite a lack of differences in perceived stress. This variability increases the difficulty of finding generalized patterns but allows for important analyses of individual differences. Analyzing data in terms of individual differences may reveal what makes some individuals more vulnerable to the effects of examination stress than others (19, 33). For example, in one prospective study, examinations were associated with increased perceived stress but not increased cortisol levels. When participants were categorized depending on the extent of increase of perceived stress, there was a significant increase in cortisol levels of those who were stress-reactive (6). This type of approach was likely first used with people who were test anxious by the Russian neuropsychologist Alexander Luria when he explored the personality traits that made some students "reactive-stable" and others "reactive-labile" to regular course examinations (34).
According to the "reactivity hypothesis" (35), separate individuals may report similar levels of perceived stress to an acute laboratory stressor such as mental arithmetic or public speaking tasks, but have different physiological responses because of genetics or learning history. Thus, in some individuals there is an uncoupling of psychological and physiological processes. People who are more physiologically reactive may be at greater risk for susceptibility to disease (35). Thus, one recommendation for examination stress analyses is to identify those who are most reactive and try to determine the factors that may account for their uniqueness.
Statistical Recommendations
When identifying examination reactors and nonreactors, the selection of the criterion for categorization should be justified. The criteria used for a cutoff score should not be arbitrary or based on some level at which significance finally appears. Furthermore, while reclassification of participants as high/low reactors may create significant differences, a simple median split reduces statistical power (36). A more powerful approach when working with continuous data is to use regression analyses. Depending on the design and analysis, measures of stress or anxiety could be part of the interaction term(s) or be an independent factor (covariate). When an interaction occurs between two continuous variables such as stress and coping (37), statistical procedures recommend by Aiken and West (38) can be used to describe and graph the interaction.
Exploration of statistical interactions may reveal the role that age, sex, social support, coping, optimism, and other factors have on reactions to examinations. While two- and three-way interactions are more difficult to explain than simple effects and they require a greater number of participants, they probably more accurately represent the real world. More complex statistical treatments, such as multilevel modeling may also provide unique explanations of findings (39).
Clinical Significance
Even though physiological changes may be associated with examination stress, there may not be any clinically significant change in health (33). For example, changes in immune function during final examination time were found in asthmatic adolescents, but there was no significant change in lung function (40). Use of clinical populations will help clarify the extent of the impact that examination stress has health. Based on previous research, this would be particularly important for those who already have dysregulated immune function such as the elderly, HIV-positive individuals, or those with autoimmune disorders.
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication December 3, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONLIMITATIONSRECOMMENDATIONSSUMMARYACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. A. Bosch, E. E. J. de Geus, C. Ring, A. V. Nieuw Amerongen, and J. R. Stowell ACADEMIC EXAMINATIONS AND IMMUNITY: ACADEMIC STRESS OR EXAMINATION STRESS? * Response Psychosom Med, July 1, 2004; 66(4): 625 - 627. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
