The relationship of academic cramming to flow experience – Statistical Data Included

Thomas M. Brinthaupt

Research has neglected to examine the experiential aspects of academic cramming. In the present study, we assessed the relationship between cramming and Csikszentrnihalyi’s (1990, 1997) flow state. We expected that experiencing such a state would be more likely for students who typically cram than for non-crammers. One hundred sixty-one undergraduates participated in the study. Following a simulation of a cramming session, they completed a measure of flow experienced during the task. Results indicated that students who normally cram performed better on the test and reported higher flow scores than the non-crammers. Implications for research on flow and study habits are presented.


Many educators probably have a negative view of the efficacy and wisdom of academic procrastination and cramming. At the same time, it is safe to say that many college students have either a need or preference for academic procrastination and cramming. For example, surveys of procrastination and cramming show that most students do both at least on occasion (e.g., Ferrari, Johnson, & McCown, 1995; Hill, Hill, Chabot, & Barrall, 1978; Solomon & Rothblum, 1984; Vacha & McBride, 1993). Some researchers claim that academic procrastination and cramming are part of an adaptive study and performance strategy (e.g., Crewe, 1969; W. Sommer, 1990), whereas others argue that academic crammers suffer from a lack of both motivation and self-regulation (Tuckman, 1991, 1998). As R. Sommer (1968) put it, cramming is “a technique as widely condemned by educators as it is widely used by students” (p. 104).

There remains considerable debate about the relative costs and benefits of both academic procrastination and cramming. The intent of our study was to examine their experiential aspects, something that has been neglected by researchers. In particular, we proposed that a major positive effect of cramming is that students may feel something akin to the “flow state” discussed by Csikszentmihalyi (1990, 1997).

Procrastination and cramming

Lay (1986) defined procrastination as “the tendency to postpone that which is necessary to reach some goal” (p. 475). Most research has focused on academic procrastination (such as the delays in completing writing assignments, staying caught up on reading assignments, and preparing for exams) and neurotic indecision (i.e., postponement of major life decisions or other forms of self-defeating behaviors) (see Milgram, Sroloff, & Rosenbaum, 1988). However, procrastination also applies to a wide variety of everyday goal-directed behaviors, including paying bills, doing the dishes, and making dental appointments (Lay, 1986, 1992; Milgram et al., 1988).

One of the major results of procrastinating in the academic realm is the need for cramming. R. Sommer (1968) defined cramming as “a heavy burst of studying immediately before an exam which followed a long period of neglect and reliance on memorization rather than understanding” (p. 105). As Vacha and McBride (1993) noted, Sommer’s definition contains two dimensions — the heavy bursting (what we shall call “cramming”) and the neglect or procrastination.

In his classic research on cramming, R. Sommer (1968) argued that there are many reasons for cramming, including such factors as the difficulty or interestingness of a course and the type of exams given. In a series of studies, Sommer found that almost all students (even successful ones) reported at least some cramming for exams, that most students reported cramming more in college than in high school, and that most students did not begin serious study for final exams until the week before finals began. On the negative side, students who crammed for finals reported increased disruptions in their normal eating and sleeping routines and increased stress and other physical symptoms (e.g., nervousness, headaches, eyestrain). Among the positives reported by student crammers were increased concentration on the materials and better memory for them.

Research has explored some of the possible costs and benefits associated with procrastination and cramming suggested by R. Sommer (1968). For example, Solomon and Rothblum (1984) found that procrastinators reported more depressed affect and irrational cognitions than non-procrastinators. However, there was no relationship between students’ procrastination scores and their course grades. Ferrari (1992) showed that procrastination scores were positively related to academic behavior delays but unrelated to exam scores. In two longitudinal studies of college student procrastination, Tice and Baumeister (1997) found that, compared to non-procrastinators, procrastinators experienced less stress and fewer daily physical symptoms early in the semester, but they showed increases in stress and declines in health as the end of the semester approached. In addition, procrastinators received lower grades than non-procrastinators on both written papers and exams. Based on these results, Tice and Baumeister argued that there are short-term health benefits to procrastination but long-term costs.

Are there other benefits of procrastination (or cramming) beyond the short-term ones noted by Tice and Baumeister (1997)? Crewe (1969) suggested that crammers might benefit from a recency effect in the learning of material. Vacha and McBride (1993) noted that some benefits of cramming might include freeing up time for other activities (e.g., leisure or catching up in other classes), relieving the monotony or boredom of studying, and privately rebelling or protesting against the demands of one’s professors. Experience with cramming might also be beneficial when a student’s time management falters. Practice at cramming could prove useful for those situations where one has been unable (for whatever reason) to adequately prepare for an exam or assignment (see Vacha & McBride, 1993).

In an ambitious effort to examine the relationship of procrastination and cramming to academic performance, Vacha and McBride (1993) had students complete weekly diaries of their study habits throughout a semester. Based on differences in the timing of studying (i.e., regularly versus showing a “burst” of activity immediately prior to the due date) and the quantity of studying (i.e., not neglecting versus neglecting one’s studying), they identified four categories of study patterns. These categories were labeled the “ideal” students (do not neglect their studies and do not burst), the “confident” students (neglect their studies and do not burst), the “zealous” students (do not neglect their studies but do burst), and the “crammers” (neglect their studies and do burst). Vacha and McBride found that crammers performed just as well as the other groups. In addition, crammers reported a higher number of hours of study time than the ideal or confident groups, with those hours tending to be more compressed. They were also more likely to cram in courses that required a great deal of writing. In summary, there is reason to think that cramming may be an effective study strategy, at least under certain circumstances.

Cramming and Flow

Although the costs and benefits of procrastination and cramming on performance are important, their experiential aspects are also likely to be related to student academic behaviors and preferences. One such aspect may be whether or not procrastination and cramming can bring about a “flow” state. According to Csikszentmihalyi (1990, 1997), characteristics of the flow state include a challenging activity that requires skills, the merging of action and awareness, clear goals and feedback, concentration on the task at hand, the loss of self-consciousness, and the transformation of time. Perhaps the most important determinant for flow or “optimal experience” is the balance between the challenge of the task or situation and one’s skills. According to Csikszentmihalyi (1997), an imbalance of challenge and skills is reflected in one’s feelings of either boredom (one’s skills exceed the challenge) or anxiety (the challenge exceeds one’s skills). When challenge and skills are matched, flow is more likely to occur.

Researchers have documented flow experiences across a wide variety of settings and tasks, including artistic endeavors, work, religious rituals, listening to music, and various physical and athletic performances (e.g., Cooper, 1998; Csikszentmihalyi, 1990, 1997; Kimiecik & Stein, 1992). However, no research to our knowledge has examined the possibility that flow might be related to certain kinds of studying behaviors.

On the one hand, there is reason to think that procrastination and cramming should be associated with a reduced likelihood of flow experience. In their Studies of the conditions associated with flow in reading, McQuillan and Conde (1996) found that self-selected texts read for pleasure were most likely to produce flow experiences (compared to, for example, assigned texts read for a college course). One of the reasons why students procrastinate (and therefore have to cram) is that the course material may be boring, uninteresting, or unengaging for them (Blunt & Pychyl, 2000). Consistent with this possibility, Vodanovich and Rupp (1999) found that procrastination scores were positively correlated with boredom proneness. Thus one of the conditions for flow (that the task is engrossing, enjoyable, and engaging) may be absent when students engage in cramming for an exam or written assignment. In addition, flow seems to be more closely tied to the experiential (i.e., intrinsically interesting) aspects of reading than to its more functional aspects such as reading in order to acquire information for later use (McQuillan & Conde, 1996).

On the other hand, one of the most attractive aspects of cramming may be that it promotes flow-like experiences. Researchers have found that people are more likely to experience flow when reading than with any other activity (Massimini, Csikszentmihalyi, & Delle Fave, 1998). In its milder forms, cramming a couple of hours for an exam increases the likelihood that students experience a challenge to their study skills, merge their action and awareness, monitor their goals and feedback, and experience deep concentration, decreased self-consciousness, and alterations in their perception of time. Pulling an “all-nighter” during mid-term or finals week might lead to an even greater likelihood of flow, at least for some students.

W. Sommer (1990) argued that adept students (those who have figured out how to “ace the system”) are those who have successfully adapted to the demands of heavy course work with maximum efficiency. These students cycle through a pattern of calculated procrastination, preparatory anxiety, climactic cramming, nick-of-time decision-making, and finally a victory to be celebrated. According to Sommer, this academic ritual takes on the characteristics of “a peak performance of personal best, an addictive high” (p. 6). in addition, he notes that the high-pressure cramming cycle fits very well into the demands of many professional and business settings. Similarly, Ferrari (1992) found that procrastination scores were positively associated with sensation-seeking scores. It may be that, when they must complete a task, procrastinators experience a “rush” of excitement and stimulation from the pressure created by their delays and postponements of it. Lay, Edwards, Parker, and Endler (1989) assessed students’ self-perceptions of threat, harm, challenge, and gain one week before, one day before, and five days after an exam. They found that immediately prior to the exam, procrastinators reported increased perceptions of challenge (i.e., confident, eager, hopeful) and gain (i.e., exhilarated, pleased, happy, relieved) compared to one week before. Although Lay et al. interpreted this pattern as “unrealistic, last-minute bravado” (p. 205), it could also be interpreted as an increase in the conditions that foster flow.

In addition to its possible flow-like characteristics, the cramming experience is likely to be reinforcing to the extent that it is associated with academic (and other forms of) success. For some successful students, cramming, may be the most important part of their studying (see Vacha & McBride, 1993). As W. Sommer (1990) put it, when students cram, “minds mired in torpor, anhedonia, and unnameable anguishes suddenly transmogrify into dazzling think-machines” (p. 6). How might such a transformation occur?

In Csikszentmihalyi’s (1990, 1997) terms, when students procrastinate in their studies, they are either intentionally or unintentionally increasing the level of challenge they are facing. One’s skills at cramming (e.g., being able to stay awake and alert, managing one’s limited time, memorizing and integrating large amounts of material) presumably remain relatively constant. However, crammers may wait for that time when the level of challenge matches their skills so that their experience is more enjoyable and engaging. Indeed, prior to a looming deadline, the course materials are more likely to be seen as boring (one’s skills exceed the challenge). However, if one waits too long to begin studying, anxiety will be the result (because the challenge exceeds one’s skills). In essence, one of the ways that students may increase the attractiveness or interestingness of otherwise uninteresting or unengaging assigned course materials or assignments is by procrastinating and then cramming. Thus, a major benefit of cramming may be that it provides students with an opportunity to demonstrate their study skills under challenging circumstances.

The present study was designed to test this possibility by examining the relationship between cramming and flow. In this study, we created a simulated cramming and testing situation for college students. Our major prediction was that self-identified crammers (i.e., Vacha & McBride’s (1993) “zealous” and “crammer” students) would be more likely to report flow-like experiences in the simulation than would self-identified non-crammers (“ideal” and “confident” students). In addition, we explored the relationship of typical study habits and academic procrastination to performance on, and the experience of flow during, the cramming simulation.



Participants were 167 undergraduates (75 males, 92 females (1)) enrolled in Introductory Psychology classes at a large (18,000+) southeastern US university. Students participated in order to earn course credit. They volunteered for the study by placing their names on a sign-up sheet which described the study as dealing with academic cramming and procrastination. They participated in groups of 10-20. Data were collected approximately two weeks before the end of the semester. Six of the participants were dropped due to missing data, leaving a final N of 161. There were two major parts to the study: (a) the assessment of student study habits and general flow experiences and (b) the cramming simulation.

Study Habits & General Flow Assessment

In order to assess student study habits, we instructed participants to “consider all your current and recent courses. We are most interested in your general study habits and patterns (rather than how you study in a particular course). In other words, make the following ratings by considering your overall study habits, across all of your courses.” First, participants provided their own definition of academic cramming. We then provided them with R. Sommer’s (1968) definition of cramming (i.e., “a period of neglect of study followed by a heavy burst of studying immediately before an exam”) and instructed participants to use this definition to rate the following items. The study habits items, all based on 5-point Likert scales, included frequency of cramming for exams during the current semester (1 = never, 2 = rarely, 3 = sometimes, 4 = often, 5 = always), general preference to not cram or to cram for exams (1 = definitely not cram, 2 = probably not cram, 3 = neutral, 4 = probably cram, 5 = definitely cram), how often one crammed by choice (preferring to wait to study even if one didn’t have to) and by necessity (being forced to because of work, lack of time, or other external circumstances) (1 = never, 5 = always), how often one gave oneself enough time to study (1 = never, 5 = always), overall level of satisfaction with study habits (1 = very dissatisfied, 2 = somewhat dissatisfied, 3 = neutral, 4 = some what satisfied, 5 = very satisfied), and frequency of engaging in last-minute or last-second cramming just before an exam is passed out in class (1 = never, 5 = always). Finally, participants classified themselves into one of the four categories (ideal student, confident student, zealous student, and crammer) described by Vacha and McBride (1993).

The participants also completed the 23-item Academic Procrastination State Inventory developed by Schouwenburg (1995). This scale focuses on students’ tendencies to postpone academic tasks and to engage in alternative activities. It has three subscales measuring fear of failure (e.g., “had panicky feelings while studying”), lack of motivation (e.g., “found the subject matter boring”), and procrastination (e.g., “gave up studying early in order to do more pleasant things”). Participants rated how frequently during the last week, using 5-point Likert scales (1 = not at all, 2 = rarely, 3 = sometimes, 4 = most of the time, 5 = always), they had experienced these behaviors or thoughts. Scale scores can range from 23-115, with higher scores indicating a greater frequency of academic procrastination. Schouwenburg (1995) reports data supporting the reliability and validity of this measure.

Participants also rated their experiences of flow across a wide variety of domains or circumstances. After reading a short paragraph defining flow experiences, they described any of their own past circumstances or activities that led to flow. Following this description, they made an overall rating of how often they experience flow, using a 5-point scale (1 = never, 2 = seldom; infrequently enough to be considered a rare exception, 3 = occasionally; often enough to be considered a familiar experience, 4 = frequently; often enough to be considered a common experience, 5 = very frequently; often enough to be considered a routine experience).

Cramming Session Procedure

The cramming procedure consisted of the administration of (a) instructions and textbook reading material, (b) a multiple-choice test on this material, (c) a brief questionnaire about the study materials and test, and (d) an assessment of flow experienced during the cramming and testing.

The reading material used in the cramming task came from a popular text on research methods in psychology (Bordens & Abbott, 1999a). We chose a 34-page chapter entitled “Using Survey Research,” which consisted of a detailed overview of questionnaire design, questionnaire administration, and survey sampling techniques. The text and chapter were chosen because they were unlikely to be familiar to participants taking an introductory-level course in psychology.

Instructions to participants were as follows:

We would like you to “study” or “cram” a chapter from a Psychology

textbook. You will have approximately 10 minutes to study this chapter,

after which you will be given a multiple-choice test on the material.

Please note that we are simply trying to simulate what it’s like to “cram”

material like this. Try to devote your full attention to “cramming” this

material and also try to do as well as you can on the test. If you want to

mark or highlight material on the reading selection, feel free to do so.

Try to use the same study techniques you might use if you were cramming

this material for one of your actual courses.

Once everyone understood the instructions, the researcher started a stop-watch and instructed participants to begin studying the materials. As each minute passed, the researcher indicated the amount of time remaining on a chalkboard and also announced the time remaining. After the time expired, the reading material was collected and participants completed a 24-item multiple-choice test. This test was obtained from the published study guide for the text (Bordens & Abbott, 1999b). The number of items answered correctly on this test served as the performance measure.

Immediately after the test, participants used 5-point Liken scales (1 = not at all, 2 = a little bit, 3 = moderately, 4 = a fair bit, 5 = very much) to rate the extent to which they had previously been exposed to the material, thought the material was difficult, and thought the experiment was similar to an actual cramming session. Then they completed the Flow State Scale (FSS) developed by Jackson and Marsh (1996). This scale is a 36-item measure of the flow state as described by Csikszentmihalyi (1990, 1997). It is designed to be administered immediately following a task or activity. Respondents rated the FSS items in relation to their experience in the cramming and testing session they just completed. The nine sub-scales of the FSS include challenge-skill balance (e.g., “my abilities matched the high challenge of the situation”), action-awareness merging (e.g., “I performed automatically”), clear goals (e.g., “I knew clearly what I wanted to do”), unambiguous feedback (e.g., “I was aware of how well I was performing”), concentration on the task at hand (e.g., “my attention was focused entirely on what I was doing”), sense of control (e.g., “I had a feeling of total control”), loss of self-consciousness (e.g., “I was not worried about my performance during the event”), transformation of time (e.g., “it felt like time stopped while I was performing”), and autotelic experience (e.g., “I found the experience extremely rewarding”). Students rated the items with 5-point Likert scales (1 = strongly disagree, 2 = disagree, 3 = neither agree nor disagree, 4 = agree, 5 = strongly agree). Scores on the FSS can range from 36-180, with higher scores indicating higher flow experiences. Jackson and Marsh (1996) report that the FSS possesses adequate reliability and validity.

Order of presentation of the cramming/testing session, procrastination scale, study habits assessment, and general flow measure was counterbalanced across groups of participants (i.e., all participants tested in the same group received the same order). Following completion of all materials, participants were thanked and debriefed.


Table 1 presents the descriptive statistics for the major variables used in the study. As suggested by these data, students reported cramming more often during the current semester than they preferred to cram, t(160) = 15.32, p < .001. They also indicated that they crammed more often by necessity than by choice, t(160) = 2.53, p < .02. In addition, last-minute cramming was frequently reported by the students. Scores on the Academic Procrastination State Inventory and the Flow State Scale were normally distributed. Not surprisingly, students answered only around 45% of the multiple-choice items correctly. The distribution of test scores was also normally distributed. Finally, the respondents' mean frequency of general flow experience was above the scale mid-point (corresponding to "occasionally; often enough to be considered a familiar experience").

Examination of the correlations among the study habits measures indicated several relations that were in the expected direction. For example, frequency of cramming during the current semester was positively correlated with general preference for cramming, r(159) = .43, p< .001, cramming by choice, r(159) = .54, p< .001, cramming by necessity, r(159) = .21, p < .01, and last-minute cramming, r(159) = .42, p < .001. Cramming frequency was also negatively correlated with giving oneself enough time to study, r (159) = -.34, p < .001, and overall satisfaction with study habits, r(159) = -.20, p < .01. Finally, students who reported a greater frequency of general flow (on the single overall flow item) also reported more satisfaction with their study habits, r(159) = .23, p < .005.

Cramming Simulation Check

The student ratings of the cramming simulation indicated that we were moderately successful in approximating a common student experience (see the bottom of Table 1). As expected, students rated their familiarity with the text material as significantly below the scale midpoint, t(160) = 15.90, p < .001, and the difficulty of the material as significantly above the scale midpoint, t(160) = 6.22, p < .001. However, they rated the similarity of our simulation to their own cramming experiences as significantly below the scale midpoint, t(160) = 6.85,p < .001. Thus, students were relatively unfamiliar with the reading material, they perceived it as moderately difficult, and they rated the simulation as between "a little bit similar" and "moderately similar" to actual cramming experiences.

Study Habits and Flow

Table 2 presents the mean flow state and test scores for the four study habit types. As the table shows, the majority of the students (83%) categorized themselves as cramming, and 59% of those who reported cramming also reported that they procrastinate (the “crammer” group). Two-way analyses of variance (cramming status X study neglect status) indicated a significant main effect of cramming status on flow state score, F(1,157) = 4.28, p < .05, as well as a marginally significant effect of cramming status on test score, F(1, 157) = 3.48, p < .07. No other effects were significant. Thus, students who normally cram reported greater flow during the simulation and tended to perform better on the test than the non-crammers. These tendencies occurred regardless of whether or not the students typically neglected their studies.

Because the cramming measures (the five study habits measures pertaining to cramming in Table 1) were strongly correlated with each other, we summed these items to create a cramming index. The index showed acceptable reliability (alpha = .69). Scores on the cramming index were unrelated to test scores, r(159) = .13, p > .15. Academic procrastination scores tended to be negatively related to test scores, but not significantly so,r(159) = -.12, p > .13. However, procrastination scores were negatively correlated with flow scores, r(159) = -.16, p < .05.

Students scoring higher on the cramming index also reported greater overall procrastination (r(159) = .25, p < .001), particularly in terms of the lack of motivation (r(159) = .31, p < .001) and procrastination (r(159) = .32, p < .001) factors. As we predicted, those who typically cram in their studies also reported higher flow scores on the cramming simulation, r(159) = .21,p < .01. In particular, cramming index scores were positively related to the challenge-skill balance (r(159) = .24, p < .01), action-awareness merging (r(159) = .22, p < .01) and unambiguous feedback (r(159) = .17, P < .05) factors of the FSS.

Our final series of analyses involved a comparison of those students who reported typically cramming by choice with those who typically crammed by necessity. Those who were categorized as choice-crammers were students whose rating of the cramming-by-choice measure was higher than their rating of the cramming-by-necessity measure. The necessity-crammers were those with the opposite pattern of response. Approximately 75% of participants fell into one of these categories, with 44 choice-crammers and 76 necessity-crammers. Choice-crammers (M = 11.91, S.D. = 3.03) scored significantly higher than necessity-crammers (M = 10.51, S.D. = 3.18) on the simulation test score, t(118) = 2.36, p < .02. The two groups did not differ in their total flow scores.

We considered the correlations among the major measures separately for the two cramming groups. Surprisingly, for the choice-crammers, the correlations between the cramming index and test score, cramming and flow score, and test score and flow score failed to reach significance. However, for the necessity-crammers, cramming was marginally correlated with test score, r(74) = .20, p < .08, cramming was significantly related to flow score, r(74) = .30, p < .01, and test and flow scores were positively correlated, r(74) = .29, p < .01. Finally, whereas cramming and procrastination total scores were uncorrelated for the choice-crammers (r = .03), they were more strongly correlated for the necessity-crammers, r(74) = .31, p < .01.


In the present study, we tested the hypothesis that one of the positive effects of academic cramming is that it increases the likelihood of flow-like experiences. To our knowledge, this is the first direct examination of flow in the academic studying realm. Participants’ behavior during and ratings of the cramming and testing simulation supported our cramming-as-flow hypothesis. In particular, self-reported crammers performed better on the task and reported a greater amount of flow-like experiences while they worked on it. On the other hand, self-reported procrastinators were less likely to experience flow during the task than were non-procrastinators. Thus, even though procrastination and cramming were positively correlated, their relationships to flow were opposite in our simulation. This pattern of results is consistent with other research that shows that procrastination and cramming should be distinguished (e.g., Vacha & McBride, 1993).

Why should the tendency to cram be positively related to flow, whereas procrastination is negatively related to it? As we proposed in the introduction to this paper, crammers may wait for that time when the level of challenge matches their skills so that their experience is more enjoyable and engaging. We are proposing that students may increase the attractiveness or interestingness of academic materials or assignments by cramming, regardless of whether or not they also procrastinate. The results are consistent with the proposal that cramming “feels good” to students and provides them with an opportunity to demonstrate their study skills under challenging circumstances. Our simulation presented students who cram with the opportunity to feel good.

Procrastination, on the other hand, may be inversely related to flow experiences because postponing required tasks and engaging in alternative activities provide exactly the enjoyment that one’s studies do not. We think that having procrastinators complete the cramming simulation amounted to asking them to engage in the kind of activity they prefer to avoid. With regard to everyday procrastination, Milgram et al. (1988) noted that “the procrastinator may gain enough in subjective consequences (relaxed pace, spontaneity, freedom from the yoke of conformity) to compensate for what one loses in objective consequences (being late or failing to perform certain life routines)” (p. 202). Our data suggest that these subjective consequences may only be beneficial so long as one is able to avoid the required task. Once no further delays are possible, the task becomes a burden and stressor, unlikely to be associated with the occurrence of flow.

Of course, crammers may have performed better than the non-crammers on the simulation simply because they are better practiced at this type of situation. If one is more accustomed to these situations (by choice or necessity), flow may be more likely to occur. With regard to scores on the FSS, students’ cramming tendencies were positively related to their perception of having the skills to meet the challenge of the simulation, their ability to perform automatically on the task, and their awareness of how well they were doing on it. This pattern of results is quite consistent with where and how we would expect cramming experience to impact flow experiences.

The differences between students who cram by choice and those who cram by necessity are intriguing. The choice-crammers scored higher on the test than the necessity-crammers. This suggests that those who cram by choice may be more skilled at cramming than those who cram because they are forced to. At the same time, test and flow scores were more strongly related to cramming for the necessity-crammers than the choice-crammers. This suggests that increases in cramming for students who are forced to cram are associated with some performance and experiential benefits that choice-crammers do not receive. It might also reflect the fact that our cramming simulation was not a choice situation for participants and was closer to the circumstances that necessity-crammers are accustomed to.

Limitations of the study

One major limitation of our study is that the simulation was only moderately similar to true cramming situations. There were no real incentives to do well on the task, and there was very little realistic time-or performance-pressure for the participants. It is probably safe to say that the reading material and test were not very meaningful to the students. For one thing, the outcome would not affect them in any manner, except for some possible self-presentational concerns. In addition, the technical nature of the reading materials made it unlikely to be of interest to most of the students. Undoubtedly, these aspects of the study made it less likely that many kinds of flow-related experiences occurred. Nonetheless, the strength of the simulation was that it actually approximated some study situations for at least some students. After all, lack of interest in the materials is a common student complaint for some courses (e.g., R. Sommer, 1968; Vacha & McBride, 1993).

The assessment of flow experiences was also limited by the nature of the study. Participants had only 10 minutes to “cram” the reading materials and took another 10-15 minutes to complete the test. Most actual cramming and testing situations probably average in the hours rather than minutes. Thus, the opportunity for the experience of flow was quite restricted in our study. Of course, there are no time restrictions on when or how one experiences flow (see Csikszentmihalyi, 1997), and our participants showed a good deal of variability in their responses to the FSS in relation to the cramming simulation.

Another limitation was that we assessed flow as it was experienced in the combined cramming and testing phases of the simulation. When it comes to academic tasks, it may be more useful to distinguish between flow experienced while studying and flow experienced during a test. When cramming for an exam, the challenge is to incorporate a large amount of material in an efficient manner. To the extent that one is progressing effectively with this task, flow is more likely to occur. Alternatively, it may be that flow is more likely during the testing than the cramming, to the extent that there is a stronger “right” and “wrong” aspect to the testing phase and it is more structured than the cramming phase. In addition, the evaluative aspect may have provided participants with clearer and less ambiguous feedback about their performance, something that is important to the experience of flow. Clearly, obtaining separate measures of flow for the cramming and testing phases of an actual cramming situation would be interesting. It may be that flow is most likely to occur in the academic domain when both aspects are present. After all, the absence of an imminent evaluation might mean that there would be insufficient motivation or challenge to bring about flow during the cramming phase.

Future research and implications

Given its limitations, it is clear that our study was only a partial test of the cramming-as-flow hypothesis. What would be a better test of the hypothesis? As we just mentioned, assessing flow experiences separately for cramming and testing would be a useful next step for future research. Similarly, one could assess students’ amount of cramming (and flow) while studying for one of their exams and then relate these to their performance on (and flow during) the exam. Of course, the strongest test of the cramming-as-flow hypothesis would be to somehow manipulate cramming for an actual course exam. One way to do this might be to assign a class a large amount of extra reading material immediately before a scheduled exam and then assess students’ experiences with studying for the exam immediately before they take it. The problem with such an approach (in addition to likely student revolt) is that an important aspect of flow is that one engages in an activity by choice and for the intrinsic enjoyment of it. Forcing students to cram for an exam creates an excessive external demand that would likely reduce the chances of experiencing flow. Flow might, however, occur in this situation for those students who regularly cram and therefore possess the skills to meet such a challenge.

There are many different reasons why students procrastinate and cram. For example, students may differ in whether they do so intentionally or unintentionally, occasionally or all the time, in some courses but not in others, and so on. Examining these differences more closely, particularly variations among students who cram by choice and those who cram by necessity, might provide a better understanding of the cramming and flow relationship. In addition, exploring other trait characteristics that might be associated with a preference for cramming would be another useful avenue for future research. For example, we would expect that cramming and sensation-seeking will be positively correlated. Ferrari (1992) found that procrastination scores were positively related to sensation-seeking scores. It may be that crammers experience a “rush” of excitement and stimulation from the pressure created by an imminent deadline and that this contributes to their experience of flow.

In some ways, we might consider the mid-term and final exam periods to be unique times in the lives of college students. Most colleges and universities specify periods during the semester when these will occur and these periods have a “tradition” all their own. For many students, these two weeks are anticipated to be (and actually experienced) as stressful, disruptive, and overwhelming. There are likely to be additional “avoidance” aspects to these times. At the same time, mid-terms and finals also have certain “approach” characteristics. Students are provided with (often multiple) performance incentives, they are faced with strong challenges to their abilities to study, manage their time, and accomplish multiple objectives in a short period of time and, we would argue, there is ample opportunity to experience “flow” over the course of these weeks of “hell.” As R. Sommer (1968) put it, “students would not cram if they didn’t get anything from it, particularly since the activity itself is burdensome, time-consuming, not particularly pleasurable, and is actively discouraged by those in authority” (p. 109). The findings of the present study suggest that experiencing a flow-like state during cramming may be one of the things (at least some) students get from it and one of the ways that they bring some pleasure to their academic requirements.

Author Note

We wish to thank Bill Compton and Ginny Poole Brinthaupt for their comments on an earlier version of this manuscript. Address correspondence to Tom Brinthaupt, P.O. Box X034, Department of Psychology, Middle Tennessee State University, Murfreesboro, TN, 37132, e-mail: Chul Shin is now at the Department of Psychology, University of North Carolina, Chapel Hill.

Table 1

Descriptive statistics for the major measures

Variable Mean S.D.

Study Habits

Cramming frequency 3.53 0.92

Preference for cramming 2.24 1.07

Cramming by choice 2.92 1.09

Cramming by necessity 3.21 1.03

Enough study time for exams 2.99 0.82

Satisfaction with study habits 2.82 1.11

Last-minute cramming 3.58 1.12

Academic Procrastination State Inventory 67.39 12.46

Flow State Scale 106.18 16.23

Reading materials test score 10.78 3.18

Frequency of flow experiences 3.14 0.99

Cramming simulation

Familiarity with text material 1.87 0.90

Difficulty of text material 3.53 1.09

Similarity to cramming experiences 2.38 1.15

Note. N = 161. All items except the procrastination and flow scales

and the test score were based on 5-point scales (1 = never/not at all,

5 = always/very much).

Table 2

Mean flowstate and test scores by self-reported study habit types

Study Habit Types Flow State Score Test Score


Ideal (n=14) 100.21 9.86

Confident (n=13) 100.54 9.54


Zealous (n=55) 108.05 10.71

Crammer (n=79) 106.86 11.19

Note. N=161. Test score refers to number correct on the

multiple-choice test.


(1) Due to an oversight, we failed to collect gender information on our materials. Thus, we were unable to examine possible gender differences. The numbers reported are based on the sign-up sheets and a record of who attended each session.


Blunt, A. K., & Pychyl, T. A. (2000). Task aversiveness and procrastination:

A multi-dimensional approach to task aversiveness across stages of personal projects. Personality and Individual Differences, 28, 153-167.

Bordens, K. S., & Abbott, B. B. (1999a). Research design and methods: A process approach (4th ed.). Mountain View, CA: Mayfield Publishing Co.

Bordens, K. S., & Abbott, B. B. (1999b). Study guide to accompany “Research design and methods: A process approach (4th ed.).” Mountain View, CA: Mayfield Publishing Co.

Cooper, A. (1998). Playing in the zone: Exploring the spiritual dimensions of sports. Boston: Shambala.

Crewe, J. C. (1969). The effect of study strategies on the retention of college material. Journal of Reading Behavior, 1, 45-52.

Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York: Harper Perennial.

Csikszentmihalyi, M. (1997). Finding flow: The psychology of engagement with everyday life. New York: Harper Perennial.

Ferrari, J. R. (1992). Psychometric validation of two procrastination inventories for adults: Arousal and avoidance measures. Journal of Psychopathology and Behavioral Assessment, 14, 97-110.

Ferrari, J. R., Johnson, J. L., & McCown, W. G. (Eds.) (1995). Procrastination and task avoidance: Theory, research, and treatment. New York: Plenum Press.

Hill, M. B., Hill, D. A., Chabot, A. E., & Barrall, J. F. (1978). A survey of college faculty and students procrastination. College Student Journal 12, 256-262.

Jackson, S. A., & Marsh, H. W. (1996). Development and validation of a scale to measure optimal experience: The Flow State Scale. Journal of Sport & Exercise Psychology, 18, 17-35.

Kimiecik, J. C., & Stein, G. L. (1992). Examining flow in sports contexts: Conceptual issues and methodological concerns. Journal of Applied Sport Psychology, 4, 144-160.

Lay, C. H. (1986). At last, my research article on procrastination. Journal of Research in Personality, 20, 474-495.

Lay, C. H. (1992). Trait procrastination and perception of person-task characteristics. Journal of Social Behaviour and Personality, 7, 483-494.

Lay, C. H., Edwards, J. M., Parker, J. D. A., & Endler, N. S. (1989). An assessment of appraisal, anxiety, coping, and procrastination during an examination period. European Journal of Personality, 3, 195-208.

Massimini, F., Csikszentmihalyi, M., & Delle Fave, A. (1988). Flow and biocultural evolution. In M. Csikszentmihalyi & I. Csikszentmihalyi (Eds.), Optimal experience: Psychological studies of flow in consciousness (pp. 60-84). Cambridge, MA: Cambridge University Press.

McQuillan, J., & Conde, G. (1996). The conditions of flow in reading: Two studies of optimal experience. Reading Psychology, 17, 109-135.

Milgram, N., Sroloff, B., & Rosenbaum, M. (1988). The procrastination of everyday life. Journal of Research in Personality, 22, 197-212.

Schouwenburg, H. C. (1995). Academic procrastination: Theoretical notions, measurement, and research. In J. R. Ferrari, J. L. Johnson, & W. G. McCown (Eds.), Procrastination and task avoidance: Theory, research, and treatment (pp. 71-96). New York: Plenum Press.

Solomon, L. J., & Rothblum, E. D. (1984). Academic procrastination: Frequency and cognitive-behavioral correlates. Journal of Counseling Psychology, 31,503-509.

Sommer, R. (1968). The social psychology of cramming. Personnel and Guidance Journal, 9, 104-109.

Sommer, W. G. (1990). Procrastination and cramming: How adept students ace the system. Journal of American College Health, 39, 5-10.

Tice, D. E., & Baumeister, R. E. (1997). Longitudinal study of procrastination, performance, stress, and health: The costs and benefits of dawdling. Psychological Science, 8, 454-458.

Tuckman, B. W. (1991). The development and concurrent validity of the Procrastination Scale. Educational and Psychological Measurement, 51, 473-480.

Tuckman, B. W. (1998). Using tests as an incentive to motivate procrastinators to study. Journal of Experimental Education, 66, 141-147.

Vacha, E. F., & McBride, M. J. (1993). Cramming: A barrier to student success, a way to beat the system or an effective learning strategy? College Student Journal, 27, 2-11.

Vodanovich, S. J., & Rupp, D. E. (1999). Are procrastinators prone to boredom? Social Behavior and Personality, 27, 11-16.



Middle Tennessee State University

COPYRIGHT 2001 Project Innovation (Alabama)

COPYRIGHT 2001 Gale Group

You May Also Like

Campus chaplains: cult training and perceptions

Campus chaplains: cult training and perceptions Russell K. Elleven This article examines the perception of 43 college chaplains acr…

College Student Use Of The Internet For Mate Selection

College Student Use Of The Internet For Mate Selection – Statistical Data Included David Knox One-hundred-ninety-one never married …

Predictors of academic achievement and retention among college freshmen: a longitudinal study

Predictors of academic achievement and retention among college freshmen: a longitudinal study M. Scott DeBerard This research exami…

Familial interactions that impact a daughter’s style of life

The father-daughter relationship: familial interactions that impact a daughter’s style of life Rose Merlino Perkins This research e…