EFTA01113830

BASIC AND APPLIED SOCIAL PSYCHOLOGY. 28(1). 1-16 Copyright 0 2006. Lawrence Erlbaum Associates. Inc. Improved Self-Control: The Benefits of a Regular Program of Academic Study Megan Oaten and Ken Cheng Macquarie University, Sydney Academic examination stress impairs regulatory behavior by consuming self-control strength (Oaten & Chang, 2005). In this study, we tested whether a study intervention program. a form of repeated practice of self-control, could improve regulatory strength and dampen the debili- tating effects of exam stress. We assessed 2 cohorts at baseline and again at the commencement of exams. Without any intervention, we replicated our previous findings of deteriorations in regulatory behaviors at exam time. Students receiving the study program, however, showed sig- nificant improvement in self-regulatory capacity as shown by an enhanced performance on a visual tracking task following a thought-suppression task. During examinations, these partici- pants also reported significant decreases in smoking, alcohol. and caffeine consumption and an increase in healthy eating, emotional control. maintenance of household chores, attendance to commitments, monitoring of spending, and an improvement in study habits. Hence, the study program not only overcame deficits caused by exam stress but actually led to improvements in self-control even during exam time. Self-regulation or self-control (terms used interchangeably Heatherton, & Tice, 1994; Muraven, Tice, & Baumeister, here) can be defined as the capacity to enact control over 1998). The resource model considers self-control to operate one's behavior. Self-control is needed to override dominant like a muscle. Any act of self-control tires this muscle, leav- behaviors that may be self-destructive, irrational, or undesir- ing less available strength for subsequent self-control tasks. able in the long term. Examples of typical self-control prob- This muscle is considered to fatigue easily, as all acts of lems include not exercising enough, engaging in unsafe sex- self-control have been argued to draw on a common resource ual practices, abusing drugs and alcohol, overspending, and or regulatory strength that is of limited capacity and is there- not sticking to study schedules. fore readily depleted. This aspect of the model is well estab- Our goals in this study were to (a) replicate the finding lished, with evidence to suggest that in the short term, peo- that real world stress, specifically academic examinations, ple's capacity for self-control diminishes following exertion consume self-control strength and consequently produce im- much like muscular action. For example, when individuals pairments in a number of unrelated regulatory behaviors were asked to engage in tasks involving self-regulation, the (Oaten & Cheng, 2005a), and (b) test whether the repeated ability to self-regulate in subsequent activities significantly practice of self-control (a study intervention program) could declined (Muraven et al., 1998; Vohs & Heatherton, 2000; improve regulatory strength and make students less vulnera- Vohs & Schmeichel, 2003). This effect of depletion has been ble to the debilitating effects of periods of high academic reported across a variety of tasks in physical, intellectual, and demand. emotional domains. RESOURCE MODEL OF SELF-CONTROL ACADEMIC STRESS AND SELF-CONTROL FAILURE A recent model suggests a lack of self-regulatory resources as one reason why self-regulation might fail (Baumeister, Failures of self-control may be related to experienced stress. A disturbing trend in student health is the reported increase in Correspondence should be addressed to Megan Oaten. Department of student stress internationally (Sax, 1997; Cotton, Dollard, & Psychology. Macquarie University. Sydney. New South Wales. Austra- de Jonge, 2002). Students report experiencing academic lia 2109. E-mail: stress at predictable times each semester, with the greatest EFTA01113830 2 OATEN AND CHENG sources of academic stress resulting from studying for and cost"ofcontrollingstress such that this cost is reflected in are- taking exams, grade competition, and the large amount of duced capacity to regulate task performance following an course content to master in a small amount of time (Archer & external stressor (unpleasant electric shock or unpredictable Lamnin, 1985; Britton & Tesser, 1991; Kohen & Fraser, noise). Glass et al.'s (1969) findings that performance is im- 1986). Examination periods have been used to investigate a paired followingstressors have beenreplicatedmany times us- number of stress responses. A finding that surfaces in these ing measures of frustration tolerance (Glass & Singer, 1972), studies is that many forms of self-regulation break down proofreading (Gardner, 1978; Glass & Singer, 1972), and the when people are managing stress. For example, West and Stroop Task (Glass & Singer, 1972). These tasks all required Lennox (1992) reported that smoking level among students the individual to override a dominant response, thus requiring was higher immediately preceding exams than at a more neu- self-control (Muraven & Baumeister, 2000). It seems that the tral period. Cartwright et al. (2003) revealed that greater aca- work required to control stress leaves the individual less able to demic stress was associated with more fatty food intake, less regulate behavior successfully. Poorer self-control is a conse- fruit and vegetable intake, more snacking, and a reduced like- quence of previous attempts to regulate stress. lihood of daily breakfast consumption. Recent longitudinal research has found that academic examination stress was as- SELF-REGULATORY IMPROVEMENT sociated with increases in cigarette smoking and decreases in physical activity (Steptoe, Wardle, Pollard, Canaan, & Thus, artificial laboratory tasks of self-regulation and having Davies, 19%). to deal with the stress of examination both lead to poorer In a previous study (Oaten & Cheng, 2005a), we tested self-control. These findings support one important aspect of whether at stressful times (during examination periods) peo- the resource model: depletion. In addition, the resource ple fail at self-regulation in domains in which control has pre- model makes a second prediction: Self-control should also viously been successful (e.g., diet). We found that students at become stronger with repeated practice, and such strengthen- exam time reported breakdowns in regulatory behavior that ing may provide a strategy to counter regulatory failure. were not found in a control group. We found this effect inboth a Previous research has found that the repeated practice of laboratory task (Stroop Test; Stroop, 1935) and on a range of self-control was followed by increments in self-control per- self-reported day-to-day behaviors. Performance on the formance (Muraven, Baumeister, & Tice, 1999; Oaten & Stroop Test deteriorated following thought suppression, a Cheng, 2005b; Oaten, Cheng, & Baumeister, 2003). In the form of regulatory activity, during the examination period. study with the longest duration, the uptake and maintenance Outside of the examination period, no such effect due to of an exercise program over a 2-month period produced sig- thought suppression was evident. Exam time also proved det- nificant improvements in a wide range of regulatory behav- rimental to a number of other self-control operations. During iors (Oaten & Cheng, 2005b). Improvements were found in a the examination period, students reported an increase in smok- laboratory task (visual tracking under distraction, which is ing and caffeine consumption; a decrease in healthy dietary used in this study as well) and on many self-reported every- habits, emotional control, frequency and duration of physical day behaviors. The laboratory measure and the self-reported activity, maintenance of household chores and self-care hab- behaviors bore no resemblance to the exercise program other its, attendance to commitments, and monitoring of spending; than that they all involved self-regulation. In particular, indi- and deterioration of sleep patterns and study habits. viduals who participated in the exercise program demon- In light of the resource model of self-control, our interpre- strated better self-regulation in other spheres: related (e.g., tation of the link between exam stress and self-control failure they engaged in more healthy behaviors), unrelated (e.g., is that managing stress requires self-regulation and thus de- missed fewer appointments), and laboratory based (visual pletes limited regulatory resources. An important part of the tracking task WM. body's defenses for coping with stress is the "fight-or-flight" There are two ways in which self-control strength could response. The fight-or-flight response prepares people for be improved. These are consistent with the ways in which physical, emotional, and mental action and is considered es- muscular strength can be increased: power (an increase in the sential for survival (Selye, 1956). These fight-or-flight re- simple baseline capacity) and stamina (a reduction in vulner- sponses, however, can be counterproductive when dealing ability to fatigue). Muraven et al. (1999), Oaten et al. (2003), with the stresses of modem life such as academic examina- and Oaten and Cheng (2005b) found evidence for increased tions (Zillman, 1983).People therefore require self-regulation stamina. The self-regulatory training appears to make people to override these natural responses to substitute other, quite less vulnerable to the effects of resource depletion. unnatural responses (such as studying harder) in their place. Stress regulation may also involve processes that demand inhibition, such as ignoring sensations, overriding negative THIS RESEARCH thoughts, and suppressing emotions (Wegner & Pennebaker, 1993) as well as regulating attention (Hockey, 1984). Glass, In this study, we examined how students fared in the exami- Singer, and Friedman (1%9) found that there is a "psychic nation period after they had been partaking in a regular study EFTA01113831 IMPROVED SELF-CONTROL 3 TABLE 1 Timeline for Study Program Semester I Semester Break Semester 2 Thu Baseline Exams Control Baseline Control Follow-up Baseline Exams Cohort SP SP Cohort 2 WL WL C C SP SP Note. SP = intervention phase (study program): WL = no-intervention phase (waiting list control): C = control phase (non-stressful testing sessions. program. In the experimental design, two cohorts partici- ing tasks that required some form of regulatory exertion— pated in the study intervention program (Table () at different in particular, a thought-regulation task (Oaten & Cheng, times of the academic year. Cohort 1 entered the study inter- 2005b) or emotion regulation (Oaten, Chau, & Cheng, vention program directly; they were tested twice across Se- 2005)—but was unaffected when following tasks that did mester 1 (baseline, exams). Cohort 2 was tested across a time not require self-control (watching humorous videos; Oaten span that included parallel testing sessions to Cohort I during et al., 2005). Thus, this task is sensitive to depletion manip- Semester 1 (waiting-list control). Cohort 2 then entered a ulations but not to nondepleting intervening tasks. In this control phase that included two assessments of self-regula- study, we administered the VTT twice at each session, and tory behavior (baseline, follow-up) during the semester in between V11' testings, participants were told to control break, which provided a neutral period of academic demand. their thoughts by not thinking about a white bear. This is a The control phase tests whether any obtained findings were standard manipulation of regulatory depletion used in past the result of repeated testing and provides measures of retest research (Muraven et al., 1998). Our (Oaten & Cheng, reliability. Finally, Cohort 2 entered the study intervention 2005b) previous research has found that performance on program in Semester 2. the VTT is highly sensitive to an intervening thought-sup- Cigarette smoking, alcohol consumption, and caffeine pression task, performance being worse after 5 min of consumption are some of the behaviors included in this thought suppression. A program of regular physical exer- study. Cigarettes, alcohol, and caffeine are the most widely cise, however, alleviated the adverse effect of the used psychoactive substances in the world (Nehlig, 1999). thought-suppression task on the VTT. We were therefore Despite differing levels of social acceptability, these behav- interested in finding out whether a study intervention pro- iors are all considered addictive (Stepney, 1996) and there- gram would have similar effects. We predicted similar per- fore require some level of regulatory management formance on the VTT before thought suppression in all (Mumford, Neill, & Holtzman, 1988). The other regulatory conditions. After thought suppression, however, perfor- behaviors of interest are diet, physical activity, self-care hab- mance on the V11' should be most impaired in participants its such as household chores, emotional control, study habits, tested at exam time without intervention (waiting-list con- spending habits, and time management. If managing the trol), next most impaired in participants tested during stress of examinations does deplete regulatory resources, and nonstressful times (control), and least impaired in partici- the repeated practice of self-control does improve regulatory pants who had partaken the study intervention program capacity, then we would expect (a) maintenance or even im- (study program). provement in regulatory behavior at exam time for those peo- ple participating in the intervention phase (study program), (b) impairment in regulatory behavior for those people in the METHOD no-intervention phase (waiting-list control) during exam time, and (c) no change in regulatory behavior across the Participants control phase (nonstressful testing sessions). We were also interested in finding out whether academic A total of 45 Macquarie University undergraduates (7 men stress affects self-control performance on a standard labora- and 38 women) recruited from introductory psychology tory task. We used visual tracking under distraction, which courses participated in return for partial fulfillment of a requires participants to perform a computerized VTT while course requirement. The age of participants ranged from 18 a distracter video is played simultaneously at a loud vol- to 51 years, with a mean age of 23 years. ume. The VTT requires participants to track the movement We randomly assigned participants to one of two cohorts of multiple independent targets displayed on a computer (Cohorts 1 and 2). Cohort 1 = 28; 4 men and 24 women) monitor (Pylyshyn & Storm, 1988; Scholl, Pylyshyn, & entered the study intervention phase directly and was indi- Feldman, 2001). The participant must ignore the distracter vidually tested in 2- to 30-min sessions separated by 8-week video content and attend only to the VIT. In a recent set of interim periods. Cohort 2 = 17; 3 men and 14 women) first studies, VTT performance deteriorated only when follow- entered the no-intervention phase (wait-list control) and then EFTA01113832 4 OATEN AND CHENG provided general controls (control phase) before proceeding making participants aware of their own concrete progress, to the study intervention phase and were individually tested which was required to maintain their long-term engagement in 6- to 30-min sessions separated by 8-week interim periods. with the program (Schunk, 1995; Zimmerman, 1989). Study schedule. The study schedule provided a tem- Design poral plan for studying in the lead up to examinations. The Table 1 shows the schedule of testing for the two cohorts. Co- study schedule specified all of the available dates and times hort 1 entered the intervention phase (study program) di- during that specific semester (taking into consideration uni- rectly. We obtained baseline measures for Cohort I in Week 5 versity contact hours and any specified work commitments), of Semester 1, the commencement of the study program, and along with a "suggested" study task designated to a specific then again during the exam period for that semester. Cohort 2 date(s). We administered the study schedule so as to enable entered the no-intervention phase (waiting-list control) in Se- participants to detect and react to any discrepancies resulting mester I with no study program. Parallel to Cohort 1, we ob- from the comparison of their current level of study and final tained baseline measures for Cohort 2 in Week 5 and then study goal state over the course of the semester. Students again during the exam period. Cohort 2 entered the interven- were expected to (a) gradually increase awareness to these tion phase (study program) in Semester 2. We again obtained suggested versus enacted discrepancies and (b) learn to mod- baseline measures in Week 5, at the commencement of the ify their behaviors so as to reduce incongruities, thus enhanc- study program, and then during the exam period. Cohort 2 ing self-regulation and improving performance. also provided general controls by participating in two testing sessions (baseline, follow-up) occurring during nonstressful Study register and study diary. These tools provided times. This served as within-subjects and between-subject opportunities for students to monitor themselves and to gen- control for the effects of the study program. All testing ses- erate the feedback necessary for self-regulation. Self-moni- sions were uniform. Experimental sessions were separated toring refers to the activities involved in observing and re- by 8-week periods. cording one's own behavior (Mace, Belfiore, & Shea, 1989). We tailored study programs to suit each participant's stu- Feedback is generated by a perceived discrepancy between dent workload and included the provision of a study register the outcome state (in this case, the study goal) and the current (log of hours spent studying, which was submitted to us in state regarding the task. This feedback fosters attempts to re- testing sessions), study diary (which was also submitted in duce any disparity by changing plans, tactics, or strategies; experimental sessions), artificial early deadlines, and a study modifying aspects of their goals; or even abandoning the task schedule for the examination period. We give more details (Ruder & Winne, 1995). Participants' utilization of these following. tools was expected to reveal their planning process and their We analyzed each experimental phase (intervention, awareness of various cues while monitoring. no-intervention, and control) separately using a more conser- vative alpha value of .01 for all statistical tests due to re- Manipulation Checks peated analysis of the same participants. We employed the study register and study diaries as manipu- lation checks to ensure that participants were adhering to the Study Program study program. Participants were instructed to bring both their student time- table (i.e., a schedule of class contact hours) and assessment Study register. Average study time was assessed by timetable (due dates for coursework assessments) to the ini- having participants complete a study register (a log of the tial testing session. We discussed with the participants any time spent studying) throughout the no-intervention (wait- work commitments that needed to be incorporated into the ing-list control) and intervention (study program) phases. For study program. analyses, study time was defined as the total number of hours, on average, that participants studied per week. Artificial early deadlines. Self-imposed deadlines are a popular strategy used by many in attempts to curb procrasti- Study diaries. To assess ease of uptake and mainte- nation (Tice & Baumeister, 1997). In fact, recent research nance of the study program, we employed the use of study di- suggested that external deadlines are more effective than aries. As part of their diary logs, participants were asked the self-imposed deadlines in boosting task performance (Ariely following questions: "What level of difficulty, if any, have & Wertenbroch, 2002). We therefore imposed early artificial you experienced complying with the program?"; "Do you deadlines on participants' assessment schedules. The artifi- feel your study habits are improving with the program'?"; and cial deadlines required the breaking down of the distant goal "Do you wish to comment on the program generally?". Par- into several proximal, specific, clear, achievable goals, thus ticipants were instructed to record their progress in the dia- EFTA01113833 IMPROVED SELF-CONTROL 5 ries provided and to return them to the experimenter at each mat. We estimated current cigarette smoking as the number experimental session. of cigarettes smoked over the past 24 hr. We assessed current alcohol consumption using a 7-day recall procedure in which quantity of alcoholic beverage was recorded. We also as- Psychosocial Self-Reports sessed caffeine consumption using a 7-day recall procedure, The GeneralHealth Questionnaire (GHQ; Goldberg, with quantity being the measure of interest. 1972). We assessed emotional distress in all sessions using the 28-item version of the GHQ. This measure assesses Dietary habits. We assessed dietary habits by ques- symptoms of emotional distress in four areas: anxiety/insom- tioning participants about food choice (e.g., "In the last nia, somatic symptoms, social and cognitive dysfunction, week, how successfully did you maintain a healthy diet?") and depression. The questionnaire referred to respondents' and dietary restraint (e.g., "In the last week, how often did experiences over the past week and was coded using a you eat junk food?) over the past week. Response sets were method that assigns weights of 0, 1, 2 and 3 to each answer recorded on a 5-point scale ranging from 0 (never) to 4 (more option. The GHQ has a high degree of internal consistency, than once per day). We derived 2 measures for analysis: junk with a reported Cronbach alpha of .87, and retest reliability food and healthy eating. was reported as .88 (Goldberg, 1972). Physical activity. We measured exercise by question- Perceived Stress Scale (ASS; Cohen, Kamarck, & ing participants about the frequency and duration of physical Mermelstein, 1983). We measured perceived stress in all activity sessions over the past week. Response sets were re- sessions using the 10-item version of the PSS. We used the corded on a 5-point scale ranging from 0 (never) to 4 (more PSS to assess the degree to which situations in life are ap- than once per day). We derived 2 measures for analysis: the praised as stressful. Each item (e.g., "In the last week, how number of episodes of physical activity and the total duration often have you felt that things were going your way?") was of physical activity sessions. assessed on a 5-point scale ranging from 0 (never) to 4 (very often), with higher scores indicating greater stress. The PSS General regulatory behavior. We measured various has been shown to be very useful to assess perceived stress, everyday behaviors that involve self-control (e.g., "In the last with an overall Cronbach alpha of .87, and retest reliability week, how often did you go out with friends instead of study- was reported as .85 (Cohen et al., 1983). This measure has ing?"). We aimed to include those behaviors that do not serve also been used in studies of academic examination stress a stress-relieving function. We recorded response sets on a (Steptoe et al., 1996; Oaten & Cheng, 2005a). 5-point scale ranging from 0 (never) to 4 (more than once per day). We derived nine measures for analysis: self-care habits General Self-Efficacy Scale (GSES; Jerusalem & (laundry habits, leaving dishes in the sink), time management Schwarzer, 1992). We measured self-efficacy in all ses- (keeping appointments and procrastination), study habits sions using the 10-item version of the GSES. Each item (e.g., (spending time with friends instead of studying and watching "It is easy for me to stick to my aims and accomplish my television instead of studying), spending habits (spending goals") was assessed on a 5-point scale ranging from 0 (not at without thinking and overspending), and emotional control all tnte) to 4 (very true), with higher scores indicating higher (loss of temper). perceived self-efficacy. The scale has been used in numerous research projects in which it has typically yielded internal Visual Tracking Under Distraction consistencies between a = .76 and .91. Its stability is satisfac- tory, with retest reliability reported as .75 (Jerusalem & We gave a laboratory task of self-control twice in each test Schwarzer, 1992). session. Participants performed a VTI' while a distracter video played at the same time in the forefront of the partici- pant. We instructed the participant to ignore the distracter Behavioral Self-Reports video content and attend only to the Vff. The VTT requires We designed a questionnaire to assess cigarette smoking, al- participants to visually track the movement of multiple tar- cohol and caffeine consumption, physical activity, dietary gets displayed on a computer monitor (see Figure 1). The habits, and other regulatory behavior. We administered the distracter video included excerpts from a comedy routine by questionnaire in both sessions. The test—retest reliability of Eddie Murphy (Murphy, Tieken, & Wachs, 1983). The use of the questionnaire is reported in the Results. the VTT to assess self-regulatory capacity has been validated in previous research (Oaten and Cheng, 2005b; Oaten, et al., Chemical consumption. We assessed cigarette smok- 2005), and we selected it for that reason. ing, caffeine consumption, and alcohol consumption by the Stimuli were displayed on an I-Mac* computer equipped use of open-ended questions presented in a questionnaire for- with a 15-in. monitor set to a resolution of 800 x 600 pixels EFTA01113834 6 OATEN AND CHENG • • NI MI IN II NI • • U Step 1 Step 2 Step 3 FIGURE 1 A representation of a visual tracking task experimental sequence. Participants view items on computer monitor. In the target identification phase (Step I). six cubes appear on the screen. and three of them Hash briefly to indicate that they are the targets: then all squares move randomly (Step 2). The task of the participant is to select the three targets once they have stopped moving by placing the cursor on them andclicking with the mouse (Step 3). and a refresh rate of 95 Hz. Participants were seated 54 cm sure of self-regulatory performance by administering a sec- away from the monitor. We controlled and measured the VT1' ond VT1'. using Psyscript (Version 4; Bates & D'Oliviero, 2000). Each V11' consisted of 16 trials. At the beginning of each trial, six black squares (20 x 20 mm) were presented in a horizontal Procedure line. After 2 sec, three target items were highlighted with Testing procedure was uniform across sessions. Participants small blinking probes (disappearing and reappearing for five first signed experimental consent forms and we then admin- flashes). Then all items moved in random trajectories for S istered in order a VTT, the thought suppression task, and then sec. After all of the objects stopped moving, the participant a second VT!'. We then obtained measures of emotional dis- had to indicate the three target items using the mouse. The fi- tress, perceived stress, perceived self-efficacy, and general nal mouse click caused the display to disappear, and the par- regulatory behaviors. We conducted data collection between ticipant initiated the next trial with a key press. Tuesday and Friday of each week so that all smoking infor- Forty-eight sets of trajectories (along with target selec- mation related to a weekday. tions) were generated and stored offline. Participants com- pleted a practice trial for which the data were not collected and then completed the experimental trials in a randomized RESULTS order (different for each participant). Overall, 9 (24%) women and 2 (28%) men smoked at some point throughout the testing session; 17 (45%) women and 4 Thought Suppression Task (57%) men consumed caffeine; and 21 (55%) women and 4 (57%) men consumed alcohol. The numbers that engaged in Following the first assessment of self-regulatory perfor- regular physical activity included 32 (84%) women and 7 mance, we administered a thought suppression task to ma- (100%) men. There was no significant difference between nipulate regulatory exertion. The procedure, developed by genders in the proportions carrying out these behaviors and Wegner, Schneider, Carter, and White (1987), requires the no baseline differences between the exam-stress and control participant not to think about a white bear. This task has been groups. We restricted analyses of each behavior to those indi- used previously to manipulate self-regulatory depletion viduals who engaged in these activities rather than the entire (Muraven et al., 1999, 1998). We told participants that over sample. the course of the experiment, they would be asked to perform a cognitive task (thought suppression). We instructed partici- pants to write down all their thoughts on a piece of paper for S Manipulation Checks min, one thought per line, so that we could "see how you use words in naturally occurring sentences" (Muraven et al., Study register. The study register (log of hours spent 1998). We then administered the experimental manipulation. studying) indicated that participants did adhere to the study We instructed participants to list any thoughts that came to program. Figure 2 summarizes the mean hours spent study- mind with the caution that they should avoid thinking about a ing. Cohort 2 was the only cohort to participate in the no-in- white bear. We told participants that whenever they thought tervention phase (waiting-list control) and was therefore the of a white bear, they were to write that thought down. We em- only cohort included in the following analyses. The reported phasized that it was critical to change their thoughts immedi- average number of hours spent studying were entered into a ately and to try not to think of a white bear again. Following session (baseline, exams) repeated measures analysis of vari- the thought suppression task, we recorded a follow-up mea- ance (ANOVA). The ANOVA showed no effect of session EFTA01113835 IMPROVED SELF-CONTROL 7 Study Habits Visual Tracking Task Hours studying per week 25 40 20 35 5 30 10 - 25 20 5- 0 I IS 10 study time S 0 ■baseline: no intervention CI CXIIIIIS: no intervention baseline: no exams: no baseline: mama lIbaseline: intervention Sextons: intervention intervention intervention intervention intervention FIGURE 2 Reported average number of hours spent studying per la pre Moil& suppression epees thought inippressien week (mean ± standard error) across the testing sessions. FIGURE 3 Error rate on the visual tracking task (meant standard error) measured before and after the thought suppression task across across the no•intervention phase. Both cohorts participated sessions. in the intervention phase (study program) and we included them in the analyses. The reported average number of hours across sessions, with less depletion during the examination spent studying were entered into a session (baseline, exams) period following participation in the study program. These repeated measures ANOVA. The ANOVA found a significant impressions were confirmed by a Session (Baseline, Exams) main effect for session, F(1, 44) = 24.58, p< .001. These re- x Time (before thought suppression vs. after thought sup- sults suggest that although on average, participants' spent 11 pression) repeated measures ANOVA. With the ANOVA, we hr per week studying, study time increased to an average of found significant main effects for time, F(I, 44) = 2395.40, p 22 hr per week during the intervention phase (study pro- < .001, indicating a general tendency toward depletion fol- gram). lowing a previous self-regulatory act; a significant main ef- fect for session, F(I, 44) = 79.96,p < .001, suggesting that vi- Study diaries. All study diaries were returned to us as sual tracking performance improved across sessions; and a instructed. An inspection of the diaries indicated that all par- significant Time x Session interaction, F(1, 44) = 359.98, p< ticipants recorded progress on the study program as in- .001. The pattern of results indicates that the study program structed. Accordingly, the diary content suggested a roughly improved regulatory stamina, increasing resistance to the de- equal expenditure of effort from all participants. bilitating effects of a manipulation of regulatory depletion (a Entries from the study diaries indicate that the study pro- thought suppression task). gram required ongoing regulatory effort. For example, some participant comments include the following: "My studying is Behavioral sett-reports. Figures 4 through 10 (black improving but it is a constant struggle ... especially when ev- and striped bars) show the reported changes in regulatory be- eryone is watching TV ... I want to join them so bad"; "In or- haviors across the intervention phase (study program). Both der to stick to the program I have to get out of bed an hour cohorts participated in the intervention phase and were in- earlier so I can get the study hours in ... some mornings it is cluded in the analyses. We entered the data in Figures 4 so hard to get up ... I'd much prefer to lie in"; and "Studying through 10 into a repeated measures ANOVA, with Session at uni isn't so bad as everyone is pretty much doing the same (Baseline, Exams) as the within-subjects variable. We re- thing ... but when I get home and my flatmates are heading stricted analyses of each behavior to those individuals who out to the pub ... it is so hard not to go with them ... so far engaged in these activities rather than the entire sample. Ta- I've managed to stay strong and stick to the planned study- ble 2 summarizes the main effects of session. ing:' The comments suggest that the academic study pro- As predicted, people seemed better able to control their gram required self-control. behavior during the exam period following the intervention phase (study program). In fact, all of the behaviors showed changes in the predicted direction. Figure 4 shows a reported Study Intervention Phase decrease in chemical consumption during examinations for V77: Figure 3 summarizes (striped bars) performance those people in the study program. Smoking decreased by a on the VTT across the intervention phase (study program). mean of 7 cigarettes per day, caffeine consumption decreased Both cohorts participated in the intervention phase and were on average by 2 cups per week, and alcohol decreased on av- included in the analyses. The thought suppression task erage by 2 drinks per week. Figure 5 shows changes in di- caused deterioration in performance at baseline (depletion). etary trends across sessions. Dietary patterns improved for This effect of depletion, however, appeared to attenuate those participants in the study program, with decreased junk EFTA01113836 Consumption Patterns Self-can Habib 2 leaving dishes leafing laundry alcohol cigarettes caffeine Ohneline: no Lawn...ono° Denims! no tatelvention Obasebne: intervention intenention lebaseline no nenration Oceans no inienen000 • buekne ententotieo •e :Nis': unentonon FIGURE 7 Self-care habits across sessions (mean ± standard er- FIGURE 4 Number of cigarettes (over 24 hr). cups of caffeine. ror). Frequency of behaviors were coded as follows: 0 = never: I = and standard units of alcohol (over 7 days) across sessions (mean ± once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than standard error). We restricted analyses of each behavior to those indi- once per day. viduals who engaged in these activities rather than the entiresample. Dietary Intake Study Habits 1. 4 .2 8 2 3 ° 2 Ti healthy eating junkfood tv instead of study friends instead of study bkiehne: DO annum. 0 MANS Intenential Inbasdinc: intervention Meurer interment Obaschse. no MIcrveatina Delinens. no intenention ▪ internxilien Monson: intervention FIGURE 5 Dietary intake across sessions (mean ± standard error). Frequency of behaviors were coded as follows: 0 = never I = once FIGURE 8 Study habits across sessions (mean 3 standard error). per week:2 = 2 to 3 times per week: 3 =daily:4 = more than once per Frequency of behaviors were coded as follows: 0 = never: I = once day. per week: 2 =2 to 3 times per week: 3 =daily:4 = more than once per day. Physical Activity Impulse Control exercise duration impulse over-spending emotional Welding ontrol Obaselme: no intervention Cl CLAMS: no intervention • baseline: intervention •cuens: intervention inbasdiiic no interretti00 0 Wens: no inlervential IIbother isavennoti Omani. intervention FIGURE 6 Frequency and duration ofphysical activity acrass ses- FIGURE 9 Impulse control across sessions (mean ± standard er- sions (mean ± standard error). Frequency of behaviors were coded as ror). Frequency of behaviors were coded as follows: 0 = never: I = follows: 0= never: I = once per week: 2 =2 to 3 times per week: 3 = once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than daily: 4 = more than once per day. once per day. 8 EFTA01113837 IMPROVED SELF-CONTROL 9 time NIonagement TABLE 2 Regulatory Behavior: Intervention Phase 4 (Study Program) Iv 3 Behavior df F Consumption 2 Cigarettes' 1. 10 135.87 <.001 1 AlcohoP 1.24 28.47 <.001 Caffeine 1.20 43.33 < .001 0 Physical activity procrastination missing appointments Frequency" 1.38 67.86 <.001 Duration' 68.14 <.001 libawItne no intern:mon Clautor m i;oinnntio• I Diet( IIII bowline IIIICIN(11001. ffilexamg Savant. Junk food 1.44 103.53 <.001 Healthy habits 78.22 <.001 FIGURE 10 Time management across sessions (meant tandard Self-cam habitst error). Frequency of behaviors were coded as follows: 0 = never. 1= Leaving dishes in sink 1.44 29.75 <.001 once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than Leaving laundry 29.33 <.001 once per day. General regulator, TV instead of study 1.44 47.43 <.001 Friends instead of study 47.42 <.001 food consumption and an increase in healthy eating habits Impulse spending 45.34 < .001 during the examination period. Figure 6 shows the same pat- Overspending 70.82 < .001 tern for physical activity. During the exam period, the fre- Emotional control 57.00 < .001 quency and duration of physical activity increased for those Procrastination 43.90 < .001 Missing appointments 47.42 < .001 participants in the study program. Figures 7 through 10 show improvements in general regu- Note. Analyses restricted to participants who engaged in these behav- latory habits in the lead up to examinations. Following inter- iors. vention, participants reported an increase in attendance to art=11.bn=25. 4tt=21."n=39."n=39. 1N=45. household chores (leaving the dishes in the sink less often and doing the laundry more often), emotional control, and a We also compared the no-intervention phase (wait- decrease in impulse spending, overspending, watching tele- ing-list control) with the intervention phase (study pro- vision instead of studying, spending time with friends instead gram) across cohorts. The two cohorts were compared at of studying, failures to attend to commitments, and procrasti- the same time of year; they were randomly assigned to con- nation. ditions (see Table 1). We conducted a mixed analysis with session and time serving as within-subjects variables and cohort as the between-subject variable. In the ANOVA, we No-Intervention Phase (Wading-List Control) compared Cohorts I and 2, with Session (Baseline, Exams) x lime (before thought suppression vs. after thought sup- VT7: Figure 3 (black bars) summarizes performance on pression) x Cohort (Cohort I [intervention phase] vs. Co- the VTT across the no-intervention phase (waiting-list con- hort 2 ]no-intervention phase]) as factors. The ANOVA trol). Cohort 2 was the only cohort to participate in the no-in- found a significant main effect for time, F(l, 43) = tervention phase and was therefore the only cohort included 5016.22, p < .001. There was also a significant Time x Co- in the following analyses. The thought suppression task hort interaction, F(1, 43) = 295.56, p < .00I, indicating that caused deterioration in performance at baseline (depletion). the rates of depletion differed across the cohorts; a signifi- This effect of depletion, however, appeared to worsen at cant Session x Cohort interaction, F(1, 43) = 110.30, p < exam time for those not participating in the study program. .001, indicating that overall visual tracking performance These impressions were confirmed by a Session (Baseline, differed across groups; and a significant Session x Time x Exams) x Time (before thought suppression vs. after thought Cohort interaction, F(I, 43) = 406.64, p < .001. These find- suppression) repeated measures ANOVA. With the ANOVA, ings suggest that during the examination period, partici- we found significant main effects for time, F(I, 16) = pants in the intervention phase (study program) showed a 3136.52, p < .001 and session, F(l, 16) = 155.82, p < .001, pattern of performance consistent with improved stamina, this time suggesting that visual tracking performance wors- whereas participants not in the study program appeared ened across sessions and importantly, a significant Time x more susceptible to the depleting effects of a prior regula- Session interaction, F(I, 16) = 252.12, p < .001. The pattern tory exertion (a thought suppression task). of results indicates that participants not in the study program were more vulnerable to the debilitating effects of a manipu- lation of regulatory depletion (a thought suppression task) Behavioral self-reports. Figures 4 through 10 (grey during the examination period. and white bars) show the reported changes in regulatory be- EFTA01113838 10 OATEN AND CHENG haviors across the no-intervention phase (waiting-list con- spending, spending time with friends instead of studying, trol). Cohort 2 was the only cohort to participate in the no-in- watching television instead of study, failures to attend to tervention phase and was therefore the only cohort included commitments, and procrastination. in the following analyses. We entered the data in Figures 4 As with the VTT, we conducted mixed ANOVAs to com- through 10 into a repeated measures ANOVA, with session as pare the no-intervention phase (waiting-list control) with the the within-subjects factor. Table 3 summarizes the main ef- intervention phase (study program) within a single statistical fects of session. test. Again, the two cohorts were randomly assigned and As predicted, people not in the study program (no-inter- compared at the same points in the semester (see Table 1). vention phase) appeared less able to control their regulatory We entered each dependant variable in Figures 4 through 10 behavior during the examination period. In fact, all of the re- into the following analyses: a Session (Baseline, Exams) x ported behaviors show changes in the predicted direction. Cohort (Cohort 1 [intervention phase] vs. Cohort 2 [no-inter- Figure 4 shows a reported increase in cigarette smoking, caf- vention phase]) repeated measures ANOVA. Table 4 summa- feine, and alcohol consumption during the examination pe- rizes the inferential statistics. Consistent with the within-sub- riod for those people not participating in the study program. jects analyses, significant Cohort x Session interactions Cigarettes increased at a mean rate of 13 cigarettes per day, indicate that during the examination period, self-regulation caffeine consumption increased at a mean rate of 4 cups per in all variables improved for those participants in the inter- week, and alcohol increased at a mean rate of 4 drinks per week. TABLE 4 Figure 5 shows changes in dietary trends across the no-in- Regulatory Behavior: Cohort 1 (Intervention tervention phase, with a reported increase in junk food in- Phase) Versus Cohort 2 (No•Intervention Phase) take, and a decrease in healthy eating habits. Figure 6 shows a Behavior df similar pattern for physical activity, with the reported fre- quency and duration of physical activity of participants not in Consumption the study program decreasing during the examinations. Cigarette0 1.9 16.03 <.001 x Cohort 145.24 <.001 Figures 7 through 10 show deficits in general regulatory Alcohol') 1.23 5.95 <.001 habits for those not in the study program in the lead up to ex- x Cohort 37.97 <.001 aminations. Participants reported a decrease in household Caffeine, 119 4.23 <.001 chores (laundry, leaving the dishes in the sink) and emotional x Cohort 45.58 <.001 control and an increase in spending without thinking, over- Physical activity Frequency° 1.37 6.46 <.001 x Cohort 60.28 <.001 Duration° 12.04 <.001 TABLE 3 x Cohort 44.32 <.001 Regulatory Behavior: No•Intervention Phase Diet' (Waiting•List Control) Junk food 1.43 7.22 <.001 Behavior df F p x Cohort 131.65 <.001 Healthy habits 9.43 <.001 Consumption x Cohort 131.64 <.001 Cigarettes• 1.5 106.50 <.001 Self-care habits' Alcohol') 1. I 0 19.55 <.031 Leaving dishes in sink 1.43 6.29 <.001 Caffeine•' 1.8 23.10 <.031 x Cohort 31.01 <.001 Physical activity° Doing laundry 6.32 <.001 Frequency 1. 16 35.86 <.00I x Cohort <.001 Duration 12.78 <.031 General regulatory behavior Diet° TV instead of study 1.43 5.12 .010 Junk food I. 16 47.80 <.001 x Cohort 4155 <.001 Healthy habits 27.20 <.001 Friends instead of study 11.12 <.001 Self-care habits,' x Cohort 4155 <.001 Leaving dishes in sink 1.16 13.19 <.031 Impulse spending 11.47 <.001 Leaving laundry 13.18 <.031 x Cohort 47.18 <.001 General regulatory behavior° Overspending 12.82 <.001 TV instead of study 19.43 x Cohort 41.75 <.001 Friends instead of study 19.42 <.001 Emotional control 4.67 .030 Impulse spending 16.10 <.001 x Cohort 4652 <.001 Overspending 12.24 <.001 Procrastination 15.25 <.001 Emotional control 34.00 <.001 x Cohort 15.25 <.001 Procrastination 19.43 <.001 Missing appointments 12.97 <.001 Missing appointments 15.61 <.001 x Cohort 15.29 <.001 Note. Analyses restricted to participants who engaged in these behav- Note. Analyses restricted to participants who engaged in these behav- iors. iors. an= 6. 6n= 11. 0n= 9. dN= 17. an =II. bpi= 25. to =21. tin= 39. 'N= 45. EFTA01113839 IMPROVED SELF-CONTROL 11 TABLE 5 TABLE 8 Regulatory Behavior Control Phase Mean Relationship Between VTT and Standard Error and Behavioral Self-Reports Baseline Follow-Up Behavior Difference VTT Differrnee Behavior df SE M SE R Consumption Cigarettes' A6* Emotional responses' Alcohols Perceived stress scale 19.1 0.5 19.2 0.8 .98° Caffeine' .440 General health questionnaire 18.8 0.4 18.6 0.3 .94. Physical activity General self-efficacy scale 193 0.7 19.3 0.8 .96° Frequency' .490 Consumptionb Duration' .490 Cigarettesb 3.1 1.0 3.0 10 .96° Mete Alcohol, 2.1 0.6 2.1 0.6 .94. Junk food Caffeined 6.0 0.5 6.2 05 .91* Healthy habits Physical activity° Self-care habits difference Frequency 21 0.3 2.3 0.2 .97* Leaving dishes in sink Duration 1.2 0.3 1.2 0.2 .97* Doing laundry .48* Diet° Regulatoryt Junk food 14.2 0.2 14.2 0.2 .97* TV instead of study .48* Healthy habits 3.2 0.3 3.2 0.2 .97* Friends instead of study .58* Self-care habits' Impulse spending ASt• Leaving dishes in sink 2.9 0.3 2.9 03 .92* Overspending .490 Leaving laundry 2.8 0.3 2.7 0.2 .93* Emotional control .560 General regulatory* Procrastination .52.0 TV instead of study 2.7 0.3 2.7 0.2 .96° Missing appointments .520 Friends instead of study 2.6 0.3 2.7 0.2 .96° Impulse spending 2.4 0.2 2.5 0.3 .90* Hole. Analyses restricted to those participants who engaged in these Overspending 2.4 0.3 2.5 03 .80* behaviors. tillT = visual tracking task. Emotional control 2.9 0.2 2.9 0.2 .97. an =II. = 25. 4n = 21.41= 39. 4N = 45. Procrastination 2.6 0.3 2.7 0.2 .98* "p is significant at the .01 level, two-tailed. Missing appointments 23 0.3 2.5 0.3 .94• £N= 17. ha =6. = 11. 4=9. 'p is significant at the .01 level, two-tailed. the reported regulatory behaviors at baseline, (b) the reported regulatory behaviors during exams, and (c) the degree of vention phase (study program), whereas regulatory behavior change in reported regulatory behavior across the interven- worsened for those participants not in the study program tion phase (study program). Both cohorts participated in the (no-intervention phase). exercise phase and we included both in the analyses. We measured the degree of change using differences across ses- Control Phase sions: intervention Session 2 (exams) score minus interven- tion Session I (baseline) score. We correlated the difference Cohort 2 was the only cohort to participate in the control scores of VTT performance (Pearson) with the differences phase (testing during nonstressful times) and was therefore scores of general regulatory behavior (Table 6). There were the only cohort included in the following analyses. Table 5 no significant correlations between the degree of change in reports the regulatory behavior (mean ± standard error) dur- VET performance and regulatory behavior reported at base- ing the control phase. There were no significant effects for line or during the exams. any of the regulatory behaviors (laboratory or self-reported) The correlations between degree of change in V11' per- across the control sessions, indicating that regulatory behav- formance and the degree of change in reported regulatory be- ior remained stable during the control phase (Table 5). havior across the intervention phase (study program), how- Test—retest reliability of the general regulatory question- ever, were all significant, suggesting that changes in the VTT naire was calculated using the Pearson correlation coefficient and in the regulatory behavior questionnaire are measuring by correlating Session I (baseline) scores with Session 2 something in common, which we believe to be changes in (follow-up) scores from the control phase. Retest reliabilities self-control. (Table 5) were generally high, with all but one at .90 or better. We also tested whether the predepletion VTT perfor- mance predicted the degree of change in regulatory behavior across the intervention phase (study program). There were no Relation Between VTT significant correlations. This suggests that the observed posi- and Behavioral Self-Reports tive changes in regulatory behavior are not related to task per- We tested whether the degree of change in VTT performance formance. Thus, VTT performance before any manipulations across the intervention phase (study program) predicted: (a) did not predict the improvements in behavior. EFTA01113840 12 OATEN AND CHENG TABLE 7 Emotional Response Means and Standard Errors Baseline: Exams: Baseline: Exams: No Intervention No Intervention Intervention Intervention Measure SE Al SE SE SE Perceived Stress Scale 19.47 0.6 29.41 0.5 19.12 0.5 19.29 0.8 Emotional Distress (GHQ) 18.76 0.4 30.82 0.3 18.88 0.5 18.65 0.3 Self-Efficacy (GSES) 19.41 0.6 19.53 1.0 19.49 0.5 19.27 0.4 Note. GHQ = General Health Questionnaire: GSES = General Self-Efficacy Scale. Psychosocial Self-Reports emotional distress) were not significant. This suggests that the positive improvements observed in regulatory behavior Table 7 shows the reported changes in self-efficacy, per- are due to the effects of the study program. ceived stress, and emotional distress across the no-inter- vention phase (waiting-list control) and the intervention phase (study program). Cohort 2 was the only cohort to par- DISCUSSION ticipate in both phases (no-intervention and intervention) and was therefore the only cohort we include in the following The results are consistent with the predictions of a limited analyses. strength model of self-control. We found that students who were dealing with academic examination stress reported Self-efficacy. Reports of self-efficacy remained stable breakdowns in regulatory behavior that were not matched across sessions. The repeated-measures analysis for the during the control phase, replicating our previous findings GSES showed no effect of session across the intervention, (Oaten & Cheng, 2005a). The main new finding to emerge waiting-list control, or control phases. was that students who participated in a study program over a 2-month period reported no increased stress at exam time and Perceived stress. Inspection of Table 7 suggests that significant improvements in a wide range of regulatory be- perceived stress increased in anticipation of examinations for haviors. We found improvements in both a laboratory task those participants in the waiting-list control (no-intervention (VTT) and on all self-reported regulatory behaviors. The par- phase). The repeated measures analysis of the PSS showed a ticipants not only studied more and improved study habits significant effect of session, F(I, 16) = 349.370, p< .001, in- but also improved their behavior in many ways outside the dicating that perceived stress increased in anticipation of ex- context of academic habits. We observed a decrease in to- aminations for those not in the study program. Reports of bacco, alcohol, and caffeine consumption and an increase in perceived stress, however, remained stable across sessions healthy dietary habits, emotional control, maintenance of for those in the study program (intervention phase). The re- household chores and self-care habits, attendance to commit- peated measures analysis for the PSS showed no effect of ments, and monitoring of spending. The laboratory measure session across the intervention phase. and the self-reported behaviors bore no direct resemblance to the study program other than that they all involved self-regu- Emotional distress. The emotional distress data in Ta- lation. Participation in the study program improved perfor- ble 7 show a similar pattern to PSS ratings. The repeated mance on all the behaviors of interest. measures analysis of the GHQ also showed a significant ef- As already mentioned, the VTT has been administered fect of session, F(1, 16) = 694.63, p < .001, indicating that following both regulatory and nonregulatory tasks (Oaten et emotional distress also increased for those participants not in al., 2005; Oaten & Cheng, 2005b), and performance on this the study program during the lead up to examination. Reports task is sensitive to depletion manipulations but not to of emotional distress also remained stable across sessions for nondepleting tasks. On this laboratory task, we found evi- those in the study program. The repeated measures analysis dence of depletion as indicated by a deterioration in perfor- for the GHQ showed no effect of session across the interven- mance following the thought suppression task in all experi- tion phase. mental sessions. This effect of depletion, however, was We also entered V11' performance (see data in Figure 3) attenuated significantly after the study program was intro- and self-reported regulatory behavior (see data in Figures duced. The study program helped to reduce but not eliminate 4-10) into a Session (Baseline, Exams) x Time (before the effects of depletion. Although the etiology of the self-re- thought suppression vs. after thought suppression) repeated ported improvements is unclear, the laboratory data suggest measures analyses of covariance, with self-efficacy, per- that the adoption of the study program made students less ceived stress, and emotional distress as covariates. The ef- susceptible to the debilitating effects of regulatory depletion, fects of these covariates (self-efficacy, perceived stress, or an improvement in stamina. EFTA01113841 IMPROVED SELF-CONTROL 13 We see the improvement in regulatory behaviors at exam The design of this study ensured that the manipulations time for the intervention condition as having two causes de- and measures were as dissimilar as possible. There was no rived from the program of regular studying. One is that the obvious reason why the study program should alter the effect study program effectively wiped out exam stress as indicated of trying to suppress thoughts of a white bear on VTT perfor- by similar scores on the PSS and the GHQ during exam time mance. Nor was there any apparent explanation why the and during mid-semester. Prevention of stress escalation by study program might benefit, for example, attendance to itself, however, should have only brought self-regulation commitments. The link between all of these behaviors was back to baseline at a comparable level to mid-semester. This that they all required self-regulation. It was important to results from eliminating the need to expend regulatory re- demonstrate depletion and improvement in circumstances as sources to deal with stress. To explain improved self-regula- diverse as possible to rule out the possibility that the results tion, however, requires another ingredient. In addition, we ar- are artifacts of a particular method. gue that the study program improved regulatory stamina. Appropriate controls show that the study program was This, coupled with a lack of increase in stress. led to im- necessary for the improvement. People on the waiting list for proved self-regulation. the study program reported an increase in regulatory failures These findings converge with those of Muraven et al. during the examination period. Such regulatory impairment, (1999), Oaten et al. (2003) and Oaten and Cheng (2005b) to however, was reversed following the adoption of the study show that the repeated practice of self-control can improve program. Therefore, the students had to actually adhere to the the strength or capacity for self-regulation. All our proce- study program for a period of time before any improvements dures and measures were different from what the Muraven et were observed. It did not matter what semester the program al. (1999), Oaten et al. (2003), and Oaten and Cheng (2005b) was administered in; what was important was that the stu- studies used, which therefore increases confidence in the dents actually undertook the program. The fact that regula- generality of the pattern. The initial studies of Muraven et al. tory behavior was stable during the control phase also indi- (1999) and Oaten et al. (2003) provide preliminary evidence cates that any observed improvement was not an outcome of that regulatory strength may be improved, but the findings multiple testing sessions or practice at the laboratory task but are limited. The regulatory exercises in these studies were ar- instead was produced by the adoption of the study program. tificial and short (2 weeks) in total duration. Oaten and Furthermore, improvements in both self-reported behaviors Cheng (2005b) improved on this design by measuring more and objective behavioral measures obtained in the laboratory than just a single laboratory task and by having participants give converging evidence. literally exercising, that is, begin and maintain a program of Longitudinal studies lose some of the control that cardiovascular exercise. This is a form of self-regulatory ex- nonlongitudinal studies offer. Several aspects of this re- ercise that many aspire to do in everyday life. It is worth not- search, however, increase confidence in the findings. First, ing that a similar pattern of results was found: an improve- this research was able to sustain a 100% rate of participa- ment in both the laboratory VTT and on self-reported, tion in contrast to the high dropout rate found in many lon- day-to-day regulatory behaviors. In the VT!', it was again gitudinal designs. This is in part owing to the relatively stamina that improved. short duration of the study and to the distribution of study Students on the program waiting list, however, did not fare diaries that helped to keep students focused on the study so well during the examination period. They reported an in- program. The experimental design also required that partic- crease in smoking, caffeine consumption, and alcohol con- ipants report regularly to us (experimental sessions, study sumption; a decrease in healthy dietary habits, emotional register, study diary, false deadlines, and study schedule). control, frequency and duration of physical activity, mainte- Such interpersonal monitoring gave a good manipulation nance of household chores, attendance to commitments, and check and might have also influenced participant retention. monitoring of spending; as well as deterioration of study Second, participants in the no-intervention condition were habits. As we noted previously, we found evidence of deple- willing to report significant degradations in behavior. This tion as indicated by impaired visual tracking performance gives some confidence that participants in the intervention following the thought suppression task in all testing sessions. condition were not merely presenting a good image of This effect of depletion, however, increased significantly at a themselves. Third, the reported behaviors all had high reli- stressful time. During the exam period, performance on the ability (Table 5). Fourth, improvements in the self-reported predepletion VTT was equivalent to baseline performance. behaviors correlated with the improvement on the labora- The postdepletion VTT performance, however, worsened tory task (Table 6). This suggests that the reported improve- more than it did at a less stressful time, showing that these ments in behavior shared something in common with im- students became more depleted after engaging in equivalent provements in an objective laboratory task, something that regulatory effort as during a less stressful time. This greater we interpret to be improvement in self-control. In this way, effect of depletion supports the claim that students on the both types of measures gain some validity, although further program waiting list had less regulatory stamina during the psychometric testing is required before firm inferences can exam period. be drawn regarding validity. EFTA01113842 14 OATEN AND CHENG Alternative Explanations plained by such self-report bias. Second, as we already men- tioned, the extent of improvement on the laboratory task pre- The reported breakdowns in self-control at exam time in the dicted the extent of improvement on all self-report measures no-intervention condition could be considered a function of (Table 6), lending some validity to the self-report measures. stress. For example, managing stress might cause an increase Past research has found links between success in aca- in the desire to smoke, eat, or drink. In fact, this is consistent demic programs of study and improved self-efficacy (Archer with these findings. Students on the program waiting list & Lamnin, 1985; Britton & Tesser, 1991). We, however, showed an impaired ability to regulate these very behaviors found that students who engaged in the study program did at reported periods of perceived high stress (exam time). not differ in perceived self-efficacy from participants in the These same students, however, also showed an impaired abil- waiting-list control (Table 5). In fact, self-efficacy was stable ity to control behaviors that do not necessarily serve a across all phases. This suggests that the "active ingredient" in stress-relieving function (emotional control, maintenance of the observed regulatory improvement is something specific household chores, attendance to commitments, monitoring to the ongoing participation in the program and cannot be ex- of spending and study habits) during the exam period. This plained by a heightened sense of self-efficacy. suggests that the negative regulatory effects observed at Improved mood may have generated the positive results. exam time were not solely in the service of stress reduction. Participants reported an increase in regular physical activity We have argued that attempts to regulate stress during the during the intervention phase. Regular physical activity has exam period decreased self-control stamina for students on been linked to improved mood (Lawlor & Hopker, 2001). the program waiting list. Another possible explanation for Being well prepared in studies might also improve mood. the observed regulatory failures involves time management The pattern of performance on the laboratory task, however, strategies. It may be that the breakdowns in regulatory behav- suggests an improvement specifically related to self-regula- ior for those students in the program waiting list reflect a tion following depletion. If, for example, improved mood time-saving tactic adopted to accommodate the higher order were producing these effects, why would we only observe goal of study in the period preceding examinations. Physical any improvement in task performance following thought activity, the preparation of healthy meals, and household suppression? It seems doubtful then that the reported chores seem like obvious sacrifices during periods of high changes in regulatory behaviors would be produced solely by academic demand. Students, however, reported increases in improved mood. Nevertheless, it will take a number of stud- unchecked spending, loss of emotional control, procrastina- ies to fully unconfound the effects of various possible factors. tion, nonattendance to planned schedules, and spending time Considering together all studies on the exercise of self-con- with friends and watching television instead of studying. trol (Muraven et al., 1999; Oaten & Cheng, 2005a; Oaten & These behaviors are costly to the individual and seem coun- Cheng, 2005b; Oaten et al., 2003), the converging evidence terproductive in the lead up to examinations. Moreover, stu- points to the repeated practice of self-control as the key in- dents participating in the study program were not only able to gredient in improving self-control. We therefore consider the manage all regulatory behaviors, but we also saw increases in resource model as offering the best fit for these findings. the frequency of regulatory behavior, this in conjunction with increased hours of studying. It seems unlikely, then, that the reported changes in regulatory behavior are produced solely Implications by time-management strategies. The results suggest that when life events impose extra de- Perhaps demand characteristics and socially desirable re- mands on self-regulatory resources (such as during final ex- sponding produced the reported improvements in general aminations), self-regulation may begin to fail in other regulatory behavior. Conway and Ross (1984) randomly se- spheres (e.g., dieting) where control has normally been suc- lected students for a study skills program and a waiting list cessful. Stress management strategies, however, should ben- for the program. At the initial interview, participants and con- efit people to the extent that by coping with life pressures ef- trols did not differ significantly on any measure of skill or fectively, people may store their regulatory strength for time spent studying. Both groups performed equally well and actions of self-change that truly demand the effort the program itself was not found to improve study skills. The (Baumeister & Exline, 2000). program participants, however, reported an improvement in a Even better news is that people may do more than just direction consistent with their beliefs regarding program out- manage stress. Our results show that individuals who studied comes (improved study skills). It is possible that participants regularly and systematically enjoyed an increased resistance in this study also exaggerated their reported improvements to to the debilitating effects of the examination period reported fit with expected program outcomes. We are inclined to be- no increase in stress. Stress prevention may work even better lieve, however, that exaggeration and self-report bias cannot than stress management. The regular practice of self-control explain all the reported improvements. First, the improve- needed to prevent stress has the added benefit of improving ment obtained in the laboratory task (VTT) cannot be ex- self-regulatory stamina (see also Oaten & Cheng, 2005b). In EFTA01113843 IMPROVED SELF-CONTROL 15 fact, the strength model posits that individuals should im- Britton. B. K.. & Tosser. A.(1991). Effects of time-management practices on prove in self-control ability even after failing a self-regula- college grades. Journal ofEducational Psychology. 83.405-410. Butler. D. L. & Winne. P. H. (1995). Feedback and self-regulated learn- tory task because the exertion of self-control is more impor- ing: A theoretical synthesis. Review of Educational Research. 65. tant than the outcome (Muraven & Baumeister, 2000). 245-281. Therefore, setting oneself small but frequent challenges for Cartwright. M.. Wardle. J.. Steggles. N.. Simon. A. E.. Croker. H.. & Janis. self-improvement may be useful for building up a good ca- M. J. (2003). Stress and dietary practices in adolescents. Health Psychol- pacity for self-discipline. ogy. 22. 362-369. Cohen. S.. ICamarck. T.. & Mennelstein. R. (1983). A global measure of per- Our findings are both theoretically and practically impor- ceived stress. Journal ofHealth and Social Behavior. 24. 385-396. tant. Theoretically, we demonstrated in the study that the re- Conway. M.. & Ross. M. (1984). Getting what you want by revising what source of self-control is not fixed and may be augmented by you had. Journal ofPersonality and Social Psychology: 47. 738-748. suitable behaviors. Although the range of routes to improve- Cotton. S. J.. Dollard. M. F.. & de Jonge. J. (2002). Stress and student job de- ment remains unclear, the data we report clearly demonstrate sign: Satisfaction. well-being, and performance in university students. In- ternational Journal ofStress Management. 9. 147-162. one route that has wide-ranging consequences. Our study on Gardner. G. T. (1978). Effects of federal human subjects' regulation on data the effects of regular physical exercise (Oaten & Cheng, obtained in environmental stressor research. Journal of Personality and 2005b) demonstrated another route with broad benefits for Social Psychology, 36.628-634. self-control. Therefore, practicing self-regulation in any do- Glass. D. C.. & Singer. J. E. (1972). Urban stress: Experiments on noise and main may have beneficial long-teen consequences. The social stressors. New York: Academic. Glass. D. C.. Singer. J. E.. & Friedman. L. N. (1969). Psychic cost of adapta- practical importance of building self-control strength can be tion to an environmental stressor. Journal ofPersonality and Social Psv- more fully appreciated when considering that many societal chology, 12, 200-210. problems involve breakdowns in self-regulation. For exam- Goldberg. D. (1972). Manual of the General Health Questionnaire. Wind- ple, addiction is marked by the inability to control cravings sor. Berkshire: NFER-Nelson. for alcohol or drugs. Additionally, many people may incur fi- Hockey. R.(1984). Varieties of attentional state: The effects of environment. In R. Parasuraman & D. R. Davies (Eds.). Varieties of attention (pp. nancial debt as a result of the inability to limit consumer 449483). Orlando. FL: Academic. spending. By improving regulatory strength, people may in Jerusalem. M.. & Schwager. R. (1992). Self-efficacy as a resource factor in turn diminish their vulnerability to such breakdowns and live stress appraisal processes. In R. Schwager (Ed.). Self-efficacy: Thought longer and healthier lives. control ofaction (pp. 195-213). Washington. DC: Hemisphere. !Cohen. J. P.. & Frazer. G. H. (1986). An academic stress scale: Identification and rated importance of academic stressors. Psychological Reports. 59. 415-426. ACKNOWLEDGMENTS Lawlor. D. A.. & Hopker. S. W. (2001). The effectiveness of exercise as an intervention in the management of depression: Systematic review and This research was conducted to meet in part the requirements meta-regression analysis of randomised controlled trials. British Medical Journal. 322. 763-766. of the Doctor of Philosophy (Psychology) by M. Oaten under Mace. F. C.. Belfiore. J. P.. & Shea. C. NI. (1989). Operant theory and re- the direction of K. Cheng. The Department of Psychology search on self-regulation. In B. J. Zimmerman & D. H. Schunk (Eds.). Postgraduate Research Grant, Macquarie University, and an Self-regulated learning and academic achievements (pp. 27-50). New Australian Postgraduate Award supported M. Oaten's effort. York: Springer. During the preparation of this article, K. Cheng was a Fellow Mumford. G. K.. Neill. D. B.. & Holtzman. S. G. (1988). Caffeine elevates reinforcement threshold for electrical brain stimulation: Tolerance and at the Berlin Institute for Advanced Study whose support is withdrawal changes. Brain Research. 459. 163-167. greatly appreciated. We thank two anonymous reviewers for Muraven. M.. & Baumeister. R. (2000). Self-regulation and depletion of lim- helpful suggestions. ited resources: Does self-control resemble a muscle? Psychological Bul- letin. 126. 247-259. Muraven. M.. Baumeister. R.. & Tice. D. M. (1999). Longitudinal improve- ment of self-regulation through practice: Building self-control strength REFERENCES through repeated exercise. The Journal of Social Psychology. 139. 446-456. Archer. J.. & Lamnin. A. (1985). An investigation of personal and academic Muraven. M.. Tice. D. M.. & Baumeister. R. (1998). Self-control as limited stressors in college campuses. Journal ofCollege Student Personnel. 26. resource: Regulatory depletion patterns. Journal of Personality and So- 210-215. cial Psychology. 74. 774-789. Ariely. D.. & Wenenbroch. K. (2002). Procrastination, deadlines and perfor- Murphy. E.. Tieken. R.. & Wachs. R. (Producers) & Cowers. B. (Director). mance: Self-control by precommittment. Psychological Science. 13. (1983). Eddie Murphy live on stage: Delirious! (Motion Picture]. United 219-224. States: Eddie Murphy Television. Inc. Bates. T.. & D'Oliviero. L (2000. June 25). PsyScript (Version 4) [Com- Nehlig. A. (1999). Are we dependent upon coffee and caffeine? A review on puter software]. Retrieved August 14. 2003. from http://www.maccs. human and animal data. Neuroscience and Biobehmioral Reviews. 23. mq.edu.auf-tirnipsyscripU 563-576. Baumeister. R.. & Exlinc. J. J. (2000). Self-control, morality and human Oaten. M.. Chau. D.. & Cheng. K. (2005). Sty-controlfailures and motiva- strength. Journal ofSocial and Clinical Ps>rhology, 19. 29-42. tion. Manuscript in preparation. Baumeister. R.. Heathenon, T. F.. & Tice. D. M. (1994). Losing control: Oaten. M.. & Cheng. K. (2005a). Academic examination stress impairs How and why people failat self:regulation. San Diego: Academic. self-control. Journal ofSocial and Clinical Psychology. 24. 254-279. EFTA01113844 16 OATEN AND CHENG Oaten. M.. & Cheng. K. (2005W. Longitudinal gains in self-controlfrom Tice. D. M.. & Baumeister. R. F. (1997). Longitudinal study of procrastina- regular physical exercise. Manuscript submitted for publication. tion. performance. stress and health: The costs and benefits of dawdling. Oaten. M.. Cheng. K.. & Baumeister. R. F. (2003). Strengthening the regula- Psychological Science. 8. 454-458. tory muscle: The longitudinal benefits ofexercising self-control. Unpub- Vohs. K. D.. & Heatherton. T. F. (2000). Self-regulatory failure: A resource lished manuscript. Macquarie University. Sydney. New South Wales. depletion approach. Psychological Science. II. 249-254. Australia. Vohs. K. D.. & Schmeichel. B. J. (2003). Self-regulation and the extended Pylyshyn. Z. W.. & Storm. R. (1988). Trackingmultiple independent targets: now: Controlling the self alters the subjective experience of time. Journal Evidence for a parallel tracking mechanism. Spatial lAsion. 3. 1-19. ofPersonality and Social Psychology. 83. 217-230. Sax. L. J. (1997). Health trends among college freshmen. Journal ofAmeri- Wegner. D. M.. & Pennebaker. J. W. (1993). Changing our minds: An intro- can College Health. 45. 252-262. duction to mental control. In D. M. Wegner & J. W. Pennebaker (Eds.). Scholl. B.J.. Pylyshyn. Z. W.. & Feldman. J. (2001). What is a visual object? Handbook ofmental control (pp. 1-12). Englewood Cliffs. NJ: Prentice Evidence from target merging in multiple object tracking. Cognition. 80. Hall. 159-177. Wegner. D. N.. Schneider. D. J.. Caner. S. R.. & White. T. L. (1987). Para- Schunk. D. H. (1995). Inherent details of self-regulated learning include stu- doxical effects of thought suppression. Journal ofPersonality and Social dent perception. Educational Psychologist 30. 213-216. Psychology. 53.5-13. Selye. H. (1956). The stress oflife. New York: McGraw-Hill. West. R.. & Lennox. S. (1992). Function of cigarette smoking in relation to Stepney. R. (19%). The concept of addiction: Its use and abuse in the media examinations. Psyrhophannacology. 108. 456-459. and science. Human Psychopharmacology. II. 15-20. Zillman. D. (1983). Transfer of excitation in emotional behavior. In J. T. Steptoe. A.. Wardle. J.. Pollard. T. M.. Canaan. L.. & Davies. G. J. (1996). Cacioppo & R. E. Petty (Eds.). Social psychophysiology: A sourcebook Stress, social support and health-related behavior A study of smoking. al- (pp. 215-240). New York: Guilford. cohol consumption and physical exercise. Journal ofPsychosomatic Re- Zimmerman. B. J. (1989). Models of self-regulated learning and academic search. 41. 171-180. achievements. In B. J. Zimmerman & D. H. Schunk (Eds.). Self-regulated Stroop. J. R. (1935). Studies of interference in serial verbal reactions. Jour- learning and academic arhiewinent (pp. 1-25). New York: Springer. nal ofExperimental Psychology: 18. 643-662. EFTA01113845