Impact of college-level courses on self-control: Comparison between a self-management course and a physical exercise course

Abstract

We examined the relative effectiveness of a college-level self-management (SM) course and a physical exercise (PE) course on improving self-control. In Study 1, Barratt Impulsiveness Scale (BIS-11) and general regulatory behavior questionnaire were administered before and after the courses to students from an SM course (experimental group 1; n = 87), a PE course (experimental group 2; n = 22), and a liberal arts course (control group; n = 28). There was a significant decrease in impulsivity and improvement in daily self-control behaviors in the SM group only. In Study 2, the same tests were administered before, after, and 3 months after the courses to the SM (n = 47) and PE groups (n = 20). Impulsivity and daily self-control behaviors were improved only in the SM group and maintained after 3 months. Thus, self-control can be improved and stabilized by teaching and directing self-control behaviors among college students.

Keywords:

• self-management
• physical exercise
• self-control
• impulsivity
• college-level course

Introduction

Self-control refers to the ability to restrain impulsive desires and regulate thoughts and behaviors to pursue individuals’ goals.^1,2 Evidence supporting the positive associations between self-control and adaptive outcomes across various domains, including academic achievement, work performance, and psychological well-being, are replete.³ Previous studies consistently indicate that a high level of self-control is imperative for healthy lifestyles. People with high self-control engage in less binge eating and overdrinking,⁴ choose healthier diets,⁵ and exhibit a low risk of problematic smartphone use than do people with low self-control.⁶ In contrast, impulsivity, which is conceptualized as the inverse of self-control,⁷ is associated with problematic behaviors, such as smoking and habitual drinking or excessive use of smartphones.^8,9

One of the best-studied models of self-control, the strength model,¹ suggests that improvement of self-control can be achieved via repetitive practice. This model is built on two major assumptions. First, self-control relies on a universal but limited resource to control emotion, behaviors, thoughts, and impulses. Thus, if the resource is depleted from supporting self-control in one task, a temporary decline in self-control in other tasks is expected in the short term. Second, however, repeated exercises of self-control can increase capacity in the long term, like muscle training. The strength model has contributed to conceptualizing self-control as a malleable skill, with room for improvement, rather than as an inherent trait. Indeed, several studies have reported findings consistent with the strength model. For example, participants persisted longer at solving difficult puzzles,^7,10 cut down on smoking,¹¹ increased healthy eating and studying habits,^12,13 and reduced physical and language abuse toward others,^14,15 after the training. Furthermore, the effects of training are generalized to other domains of behavior. A meta-analysis of 18 studies examining the effect of self-control training showed that practicing a self-control behavior in one domain, such as controlling emotions or food consumption, led to improvements in other domains, such as decision making and Stroop task performance.¹⁶

One of the most widely studied strategies for improving self-control is self-management (SM) training, which refers to the promotion of changes in lifestyle and self-designated behaviors to enhance individuals’ physical and psychosocial health.¹⁷ Target behaviors in SM training include, but are not limited to, increasing physical activity and health-promoting behaviors, decreasing excessive use of drugs/alcohol, and improving personal habits,¹⁸ which all require self-control. SM training has been successfully applied in various settings since the 1960s, especially as college-level courses, and its effectiveness in behavioral changes has been well-established.¹⁹ For example, Menges and Dobroski²⁰ reviewed 20 articles investigating the effectiveness of SM courses using both objective data (e.g., frequency of target behaviors) and self-report scales (e.g., a goal-attainment scale) and concluded that SM courses are effective in improving various target behaviors. After a lapse of almost 20 years, a few more studies have been published. For example, an intensive SM course was found to be more effective in modifying student behaviors than low-intensity SM and non-SM courses.²¹ Moreover, Yang and Chung²² reported the relative effectiveness of an SM course in reducing nervous habits compared to modifying sleep and eating habits and physical activity. These studies consistently reported that SM courses can help approximately 50% to 70% of students to achieve significant progress in their target behaviors.

However, studies directly exploring the effects of SM courses on self-control are scarce. To our knowledge, only one study has examined whether SM courses improve self-control using both subjective and objective measures of impulsivity,²³ including the Barratt Impulsiveness Scale-11²⁴ and a delay discounting task.²⁴ The researchers also investigated whether the effects of SM training are generalizable to other daily behaviors that require self-control (e.g., drinking, exercising, and studying) using the general regulatory behavior questionnaire.^12,13 Their results showed a significant decrease in impulsivity as well as improvements in daily self-control behaviors among the participants who successfully achieved their goals. In particular, those who targeted increased exercise reported higher improvements in daily behaviors than those who targeted other behaviors (i.e., studying, improving habits), suggesting that exercise may be effective in improving self-control.

Apart from SM training, a handful of studies have empirically examined whether physical exercise (PE) can be utilized as self-control training. Their rationale comes from the observation that maintaining an exercise routine require a certain level of self-control.^25,26 Oaten and Cheng¹³ investigated changes in self-control using a laboratory task and self-reporting of daily behaviors among 24 college students who were assigned to a 2-month PE program. The results showed a significant improvement in self-control, as observed in impulsivity reduction after the program and its generalization to daily behaviors. Similarly, Kim et al.²⁷ used a delay discounting task to compare the impulsivity of 200 students enrolled in an intensive PE course to enhance physical strength and 150 students enrolled in a general physical education course focusing on a sports activity. The findings revealed a significant decrease in impulsivity in the intensive PE course group only.

Taken together, research on the effects of SM and PE courses suggests that self-control can be improved via these two types of college course. However, there has been a dearth of research on the relative effectiveness of these two courses or an examination of the extent to which the course effect is generalizable to other daily behaviors and retainable, despite this being a prominent issue surrounding the practicality of self-control training.²⁸

The purposes of this study are to investigate whether college-level SM and PE courses can improve self-control in reducing impulsivity and whether the effects can be generalized to daily self-control behaviors and maintained over time. Therefore, in Study 1, the effects of SM and PE courses in reducing impulsivity and their generalization to daily self-control behaviors were examined. In Study 2, the effects of SM and PE courses on impulsivity and their generalization to daily self-control behaviors were reexamined, and their maintenance 3 months after the course termination were explored.

Study 1

In Study 1, the effects of an SM course and a PE course were compared on impulsivity and the generalization of course effect on daily self-control behaviors. Students recruited from SM and the PE courses were assigned to the experimental groups, and a group of students from a liberal arts course was assigned the control group. Using a behavioral task and self-report questionnaires, changes in impulsivity and daily self-control behaviors among the three groups were compared before and after the course.

Methods

Participants

The participants were college students enrolled in an SM course (“Impulsivity and Self-Management”), a PE course (“Strength Enhancement”), and a liberal arts (LA) course (“Introduction to Psychology”) at a university located in Seoul, South Korea. Of the 159 students who consented to participate in the study, five students who enrolled in more than one of these courses and three students who submitted unreliable responses in the behavioral task were excluded from the data analyses. In addition, 14 students failed to complete the course, leaving a total of 137 students (87 in the SM group, 22 in the PE group, and 28 in the LA group; see Table 1). The participants were all ethnic Koreans. This study was approved by the institutional review board (IRB: 7001988-201810-HR-262-04).

Table 1. Characteristics of participants (Study 1).

	SM group (n = 87)	PE group (n = 22)	LA (control) group (n = 28)
Gender
Female	51	6	16
Male	36	16	12
Age
Average (SD)	21.85 (1.97)	23.05 (2.19)	21.68 (2.04)
Range (years)	19–26	19–28	19–27

Note. SM = self-management; PE = physical exercise; LA = liberal arts.

Measures

Delay discounting task

The study used the computerized delay discounting task developed by Choi and Chung²⁹ with algorithms adapted from Richards et al.²⁴ to assess impulsivity. The delay discounting task repeatedly assesses preferences between an immediate but small reward and a delayed but bigger reward and calculates the relative value of the delayed reward in comparison to the equally preferred immediate reward. The rewards were presented at $5 incremental differences, ranging from $5 to $100 per trial, and the delay duration was set to “now,” “2 days later,” “1 month later,” “6 months later,” or “1 year later.”^23,30 The area under the curve (AUC) method was used to calculate the discount rate,³¹ with smaller AUC values reflecting greater delay discounting, that is, higher impulsivity.

Barratt impulsiveness scale-11

The Korean version of the Barratt Impulsiveness Scale-11 (BIS-11) was administered to measure impulsivity.³² The BIS-11 consists of 23 items assessing non-planning impulsiveness, motor impulsiveness, and cognitive impulsiveness on a 4-point scale (1 = Not at all to 4 = Always). The internal consistency (Cronbach’s alpha) for the total BIS score ranged from .79 to .83 in the original study,³² and that in the current study was .83 for pretest and .84 for post-test.

General regulatory behavior questionnaire

To measure various daily self-control behaviors, five items were used to assess frequency levels in five self-control domains on a 7-point scale (0 = Never to 6 = More than once a day),²³ adapted from the general regulatory behavior questionnaire (GRBQ).^12,13 The five domains include alcohol consumption, exercise, junk food intake, studying, and time management in the preceding week. The scores from alcohol consumption and junk food intake were reverse coded, so that the higher total score would indicate higher self-control level across domains.

Procedure

Students from the three courses (SM, PE, & LA) were recruited on the first day of those classes. Students were informed that participation in the study would not affect their grades and that lecturers would have no access to participants’ information. Those who were interested in participating in the study were asked to fill out the participant information sheet, and those who signed the informed consent form completed the pretest self-report scales and a computerized behavioral task. Both the scales and the task were administered via email or through the online research management system of the university. In the last week of class, an online link containing the post-test task and scales were distributed to all participants. Participants who completed both pre-and post-tests were compensated with research credits toward a course requirement or a mobile voucher worth $5.

Course description

The SM course was designed to help students change their target behaviors during the semester on the basis of principles and strategies acquired from the course. The SM course was a 45-hour three-credit course that took place two times a week (divided into a 1-hour and a 2-hour class). The course was taught by the corresponding author, a clinical psychologist who has been teaching the SM course for several years. The textbook used was the Korean version of Watson and Tharp¹⁸’s Self-Directed Behavior (10th edition). Table 2 shows the details of the course syllabus. The PE course was designed to improve physical fitness and develop exercise habits. This course consisted of resistance training and cardio exercises. The course had five subdivisions and was taught for a total of 2 hours a week along with a weekly assignment of at least 1 hour of exercise. All subdivisions had the same syllabus but were taught by different instructors who were either a PhD candidate or a PhD in PE education. The LA course was a three-credit introductory course in psychology, team-taught by 13 professors who led a 3-hour class each week.

Table 2. Syllabus of the self-management course.

Week	Theoretical classes	Behavior modification project
1	Course introduction Ch. 3. Self-Knowledge: Observation and Recording	Target establishment
2	Ch. 2. Forethought: Planning for Success	Pre-assessment
3	Ch. 5. Antecedents	Baseline measurement
4	Ch. 5. Antecedents	Baseline measurement
5	Ch. 8. Developing a Successful Plan	Application of behavioral strategies
6	Ch. 6. Behaviors: Actions, Thoughts, and Feelings	Application of behavioral strategies
7	Ch. 6. Behaviors: Actions, Thoughts, and Feelings	Application of behavioral strategies
8	(Mid-term exam)
9	Ch. 4. The Principles of Self-Direction: Theory and Practice	Application of behavioral strategies
10	Ch. 7. Consequences	Application of behavioral strategies
11	Cognitive Skills of Self-Management	Application of behavioral strategies
12	Behavioral Economics and Applied Behavior Analysis	Application of behavioral strategies
13	Ch. 9. Problem Solving and Relapse Prevention	Application of behavioral strategies
14	Ch. 10. Staying in Control	Application of behavioral strategies
15	Presentation on behavior modification project	Submission of final report
16	(Final exam)

Analysis

An a priori power analysis was conducted using the G*Power program to determine sample size with an α error probability of 0.05 (two-tailed), power of 0.90, and three groups. A small to medium effect size, according to the conventions of Cohen,³³ was estimated on the basis of a recent meta-analysis of self-control training effects.³⁴ This power analysis indicated that at least 54 participants were required in total (18 per group) when the effect size was medium (f = .50).

We used SPSS Version 24.0 and ARTool for data analysis.³⁵ The dependent variables were the two impulsivity measures (AUC value and BIS-11 total score) and GRBQ score. The normality of all dependent variables was tested using the Shapiro-Wilk test. BIS-11 and GRBQ scores satisfied the normality assumption, whereas AUC values did not (p < .001). Hence, non-parametric methods were used to analyze AUC values. Next, the homogeneity of the three groups was tested for age, gender, and the pretest scores of all dependent variables. A one-way analysis of variance (ANOVA) was used for age and a Chi-square analysis was used for gender. For BIS-11 and GRBQ scores that satisfied the normality assumption, a one-way ANOVA was conducted. For AUC values in which the normality assumption was not satisfied, the Kruskal-Wallis H test was used. The groups differed significantly in age, F(2,134) = 3.54, p = .032, and gender, χ² = 7.14, df = 2, p = .028; hence, they were controlled in the pre-post comparison analysis. The results of the ANOVA and Kruskal-Wallis H test indicated that there were no group differences in pretest scores (AUC: Z = 5.254, p = .072; BIS-11: F(2,134) = 2.606, p = .078; GRBQ: F(2,134) = 1.153, p = .319).

For pre-post comparisons, a repeated-measures analysis of covariance (ANCOVA) was conducted for all dependent variables with time (pre/post) as a within-group variable, the group (SM/PE/control) as a between-group variable, and age and gender as covariates. In addition, an intent-to-treat (ITT) analysis was performed to compare the results from completers and ITT samples using last observation carried forward imputation for missing data. The AUC values that did not satisfy the normality assumption were transformed for analysis following the aligned and rank transform procedure suggested by Wobbrock et al.³⁵ For variables in which a significant interaction was found, a post hoc analysis of a paired sample t-test or a Wilcoxon sign rank test was performed.

Results

Pre-post comparisons

The pre- and post-test scores of each variable by group are presented in Table 3.

Table 3. Pre- and post-test scores for each variable by group (Study 1).

	SM group (n = 87)		PE group (n = 22)		LA (control) group (n = 28)
	Pre M (SD)	Post M (SD)	Pre M (SD)	Post M (SD)	Pre M (SD)	Post M (SD)
AUC	.68 (.23)	.75 (.20)	.77 (.23)	.79 (.20)	.74 (.23)	.70 (.25)
BIS-11	52.66 (8.57)	49.59 (8.92)	49.23 (8.52)	51.18 (8.07)	49.14 (8.64)	49.82 (8.22)
GRBQ	15.94 (3.84)	18.22 (3.77)	17.00 (4.39)	17.05 (3.70)	17.00 (3.80)	17.25 (3.23)

Note. SM = self-management; PE = physical exercise; LA = liberal arts; AUC = area under the curve; BIS-11 = Barratt Impulsiveness Scale-11; GRBQ = the general regulatory behavior questionnaire.

Delay discounting task and BIS-11

To test the group differences in impulsivity change, a repeated-measures ANCOVA was conducted on the AUC values derived from the delay discounting task and the BIS-11 total scores. A significant time (2) × group (3) interaction effect was reported for the AUC value, F(2,132) = 4.165, p = .018, η² = .06, and BIS-11 score, F(2,132) = 8.974, p < .001, η² = .12. Post hoc analyses revealed that the AUC values increased significantly in the post-test compared to the pretest, Z = −3.211, p = .001, and that the BIS-11 total score significantly decreased, t(86) = 4.762, p < .001, in the SM group, both indicating a significant decrease in impulsivity after the SM course. In contrast, no change in AUC value and the BIS-11 score was observed in the PE group (AUC: Z = −.915, p = .360; BIS-11: t(21) = −1.407, p = .174) or the control group (AUC: Z = −1.491, p = .136; BIS-11: t(27) = −1.063, p = .297). Main effects were not found for time or group in either AUC (time: F(1,132) = .035, p = .852, η² = .00; group: F(1,132) = 1.317, p = .272, η² = .02) or BIS-11 scores (time: F(1,132) = .436, p = .510, η² = .00; group: F(1,132) = .472, p = .625, η² = .00). The results were equivalent in the ITT analysis.

GRBQ

To test for the group differences in the pre-post comparisons of daily self-control behaviors, a repeated-measures ANCOVA with age and gender as covariates was conducted on the GRBQ score. A significant time (2) × group (3) interaction effect was found, F(2, 132) = 4.997, p = .008, η² = .07. The post hoc analysis showed that the GRBQ score significantly increased in the post-test compared to the pretest in the SM group (t(86) = −5.762, p < .001), whereas the PE group (t(21) = −.081, p = .936) and the control group (t(27) = −.434, p = .668) did not see any change. Main effects for time, F(1,132) = .075, p = .784, η² = .00, and group, F(1,132) = .013, p = .987, η² = .00, were not significant. The results were equivalent in the ITT analysis.

Discussion

In Study 1, the effects of the SM and PE courses on impulsivity and the generalization of course effect on daily self-control behaviors were compared to those of the LA course (control group). The results showed that the SM course is effective in reducing impulsivity, whereas the PE and LA courses are not. This study is the first study to show directly that impulsivity can be reduced through repeated training in self-control of various behaviors. It has also shown that such training is possible in a classroom-based course. The practicality of a one-on-one training approach, which is the most common form of behavioral intervention, is undermined by the costs associated with it, but the didactic teaching method is highly accessible and can still have an individualized component through personalized goals for each student. Considering that the effectiveness of the SM course has been established, repeated validation of reduction in impulsivity via an SM course would provide stronger support for disseminating the use of SM as an intervention strategy.

This study also showed that the effects of the SM course on self-control were generalized to daily behaviors, which can be attributed to the expansion of self-control skills to everyday life, learned through training in behavioral strategies. For example, one of the strategies taught to students was the self-monitoring and reinforcement of desirable behaviors, which the students may have been able to use in their daily self-control repertoire.³⁶ Thus, the repeated practice of behavioral strategies to deal with a target behavior can improve self-control for the targeted behavior and beyond, with the effects being extended to other non-targeted behaviors. In fact, this could be another reason to support the active use of SM courses in university settings.

In contrast, the PE course did not change the level of impulsivity, a result inconsistent with those of previous studies.^13,27 The mixed results may be due to differences in the exercise intensity across studies and groups. For example, in Kim et al,²⁷ the authors found that only the group with a high intensity level of exercise showed decreased impulsivity compared to the low-intensity group. Their study included participants who worked out more than 2 days a week and at least 1 hour each day in the data analyses, supposing that intensive exercise requires self-control. However, this inconsistency might be related to the fact that the PE course in this study was less intense than the courses in the previous studies. Further examination of the effectiveness of PE, controlling the intensity of the course, is necessary.

Study 2

Study 2, the effects of SM and PE courses on impulsivity and their generalization to daily self-control behaviors were replicated, and the maintenance of these effects were assessed through a 3-month follow-up (FU) test. In Study 2, only SM and PE courses were compared since significant differences between SM and control group were clearly observed in Study 1. The measures and procedures were similar to those of Study 1.

Methods

Participants

The participants were college students enrolled in an SM course (“Impulsivity and Self-Management”) and a PE course (“Strength Enhancement”) recruited 1 year after Study 1. Recruitment procedures and inclusion criteria were the same as those for Study 1. Of the 97 students recruited in this study, three students who enrolled in both courses and five students who submitted unreliable responses in the behavioral task or questionnaires were excluded from the data analyses. Of the remaining 89 students, 67 students completed both the pre- and post-tests (Table 4). All participants were ethnic Koreans. Thirty of the students participated in a 3-month FU after the end of the course. This study was approved by the institutional review board (IRB: 7001988-201809-HR-441-04).

Table 4. Characteristics of participants (Study 2).

	SM group		PE group
	Pre-post completers (n = 47)	FU participants (n = 18)	Pre-post completers (n = 20)	FU participants (n = 12)
Gender
Female	30	11	1	1
Male	17	7	19	11
Age
Average (SD)	21.02 (1.67)	21.72 (1.67)	23.35 (2.43)	23.67 (2.31)
Range (years)	19–25	19–25	19–28	20–28

Note. SM = self-management; PE = physical exercise; FU = follow-up.

Measures

Delay discounting task

The delay discounting task was used to measure impulsivity as in Study 1. The maximum compensation amount and the maximum delay duration were also the same, which were $100 and 1 year, respectively.

BIS-11

The BIS-11 was used to measure impulsivity, as in Study 1. Cronbach’s alphas for the total scores were .79, .76, and .85 for the pretest, post-test, and FU, respectively.

GRBQ

In Study 2, three items measuring behavioral frequency in additional domains (impulse buying, emotion regulation, and excessive use of the Internet/smartphone) were added to the five items of the GRBQ used in Study 1 to reflect changes in broader domains of self-control.¹² A total of eight items measured daily self-control behaviors on a 7-point scale (0 = Never to 6 = More than once a day), as in Study 1.

Items on exerted self-control and effort

To investigate whether the SM and PE courses differed in self-control training intensity, two items measuring levels of self-control (“How much self-control was required for your behavioral change on the course?”) and effort exerted to meet the course expectations (“How much effort did you make to change your behavior?”) were adopted from a previous study.¹¹ The items were administered in the post-test using a 5-point scale (1 = Almost none to 5 = Very high).

Procedure

The procedure was identical to that of Study 1 except for the 3-month FU. Prior to beginning the study, participants were informed that the FU was scheduled; the FU was conducted individually using an online link. Participants who completed both the pre- and post-tests were compensated with additional course credits or a mobile voucher worth $10. Those who completed the FU were compensated with an additional mobile voucher worth $10. The course curriculum for each course was identical to that of Study 1.

Analysis

An a priori power analysis was conducted using the G*Power program to determine sample size with an α error probability of 0.05 (two-tailed), power of 0.90, and two groups. The effect size was set to medium (f = 50). The power analysis indicated that at least 46 participants were required in total (23 per group). The data analysis was conducted using SPSS Version 24.0 and ARTool.³⁵ The normality of the dependent variables and homogeneity between groups at pretest were tested first. An independent samples t-test for age and Chi-square analysis for gender were conducted to test homogeneity between the two groups, and a significant difference was found for age, t(65) = −4.525, p < .001, and gender, χ² = 19.531, df = 1, p < .001. Consequently, gender and age were controlled in the following analyses. In addition, an independent samples t-test or Mann Whitney U test was performed to examine whether the two groups were identical at pretest. No group differences were found in any dependent variables (AUC: Z = 1.191, p = .234; BIS-11: t(65) = 1.254, p = .214; GRBQ: t(65) = −1.866, p = .067). Finally, the differences in age, gender, and dependent variables between the participants who completed the FU and those who did not were tested using an independent samples t-test, Chi-square analysis, and Mann Whitney U test. No difference was reported for gender, age, or any dependent variable in the pre- and post-tests between the participants who completed the FU and those who did not, or within each group.

The main analysis was performed in two steps and in reference to previous studies that examined the maintenance of intervention effects,^37,38 given that only 45% of participants completed the FU. This issue is addressed in the general discussion as a limitation to this study. First, the pre-post comparisons between two groups were performed with the repeated-measures ANCOVA. In addition, an ITT analysis was performed as in Study 1. The AUC values that did not satisfy the normality assumption were transformed using the aligned and rank transform procedure before analysis.³⁵ Second, to test whether the effects of courses on self-control persisted to FU, a paired sample t-test or Wilcoxon signed-rank test was conducted within groups for significant differences between pretest and FU and post-test and FU for those who completed all three tests. In addition, an independent samples t-test was conducted to determine whether there was a difference in the self-control and effort required for behavioral changes between two groups.

Results

Pre-post comparisons

The pre- and post-test scores of each variable are presented by group in Table 5.

Table 5. Pre- and post-test scores for each variable by group (Study 2).

	SM group (n = 47)		PE group (n = 20)
	Pre M (SD)	Post M (SD)	Pre M (SD)	Post M (SD)
AUC	.64 (.28)	.79 (.21)	.74 (.27)	.72 (.25)
BIS-11	49.77 (8.28)	47.79 (7.16)	47.10 (7.14)	46.65 (6.32)
GRBQ	29.55 (4.49)	33.60 (4.22)	31.88 (5.06)	32.50 (4.85)

Note. SM = self-management; PE = physical exercise; AUC = area under the curve; BIS-11 = Barratt Impulsiveness Scale-11; GRBQ = the general regulatory behavior questionnaire.

Delay discounting task and BIS-11

The results of a repeated-measures ANCOVA found a significant interaction effect between time and group in AUC, F(1,63) = 4.061, p = .048, η² = .06. The post hoc analysis showed that the AUC value increased significantly in the post-test compared to the pretest in the SM group, Z = 3.893, p < .001, while no change was observed in the PE group, Z = −.207, p = .836. The main effects of time, F(1,63) = .000, p = .994, η² = .00, and group, F(1,63) = 1.427, p = .237, η² = .02, were not significant. As for the BIS-11 total score, there were no significant effects of group × time interaction (F(1,63) = .018, p = .893, η² = .00), time (F(1, 63) = .015, p = .903, η² = .00), or group (F(1,63) = .356, p = .553, η² = .00). The results were equivalent in the ITT analysis.

GRBQ

A significant interaction between time and group was found, F(1,63) = 4.723, p = .034, η² = .07. A paired sample t-test for post hoc analysis showed that the GRBQ score significantly increased in the SM group, t(46) = −6.266, p < .001, but not in the PE group, t(19) = −.511, p = .615. The results were equivalent in the ITT analysis.

Maintenance of changes

The pre- and post-test and FU scores of each variable from those who completed the FU (SM: n = 18, PE: n = 12) are shown in Table 6. The maintenance of changes was examined in the within-group analyses between the pretest and FU and the post-test and FU.

Table 6. Pre-, post-, and follow-up test scores for each variable by group from follow-up completers (Study 2).

	SM group (n = 18)			PE group (n = 12)
	Pre M (SD)	Post M (SD)	FU M (SD)	Pre M (SD)	Post M (SD)	FU M (SD)
AUC	.63 (.26)	.78 (.21)	.77 (.22)	.76 (.28)	.75 (.29)	.76 (.27)
BIS-11	49.11 (10.32)	46.83 (8.61)	46.61 (8.10)	47.50 (8.30)	47.33 (7.17)	49.00 (9.14)
GRBQ	30.39 (4.53)	34.61 (3.53)	33.17 (4.48)	31.46 (4.87)	33.17 (5.02)	30.67 (4.27)

Note. SM = self-management; PE = physical exercise; FU = follow-up; AUC = area under the curve; BIS-11 = Barratt Impulsiveness Scale-11; GRBQ = the general regulatory behavior questionnaire.

Delay discounting task and BIS-11

The Wilcoxon signed-rank test was conducted on the AUC values to evaluate whether significant changes in impulsivity were maintained 3 months after the course in each group. The AUC value in the FU increased significantly from the pretest, Z = 1.988, p = .047, but not from the post-test, Z = −.398, p = .691, in the SM group. Differences between the pretest and FU, Z = 1.245, p = .213, and the post-test and FU, Z = .701, p = .483, were not significant in the PE group.

A paired sample t-test revealed that the BIS-11 score in the FU did not significantly differ between the pretest, t(17) = 1.567, p = .135, and the post-test, t(17) = .167, p = .874, in the SM group. The same pattern emerged for the PE group. Neither the changes from the pretest to FU, t(11) = −.799, p = .441, nor those between the post-test and FU, t(11) = −.758, p = .465, were significant.

GRBQ

A paired sample t-test revealed that the GRBQ score at FU was significantly higher than at the pretest for the SM group, t(17) = −2.184, p = .043, but did not differ from the post-test, t(17) = 1.309, p = .208. For the PE group, the GRBQ score at FU did not differ from the pretest, t(11) = .557, p = .589, or from the post-test, t(11) = 1.778, p = .103.

Exerted self-control and effort

An independent samples t-test was conducted to investigate whether the SM and PE courses were homogenous for the levels of exerted self-control and effort. There was no significant group difference in the levels of self-control and effort exerted in each course (Table 7), indicating that the two courses are not different in terms of self-control training intensity.

Table 7. Group differences in self-control and effort required for behavioral change.

	M (SD)		df	t
	SM group	PE group
Exerted self-control	3.96 (.81)	3.75 (.79)	65	.971
Effort toward behavioral change	3.91 (.69)	3.50 (.95)	27.885	1.773

Note. SM = self-management; PE = physical exercise.

Discussion

In Study 2, the effects of the SM course in reducing impulsivity and their generalization to various daily self-control behaviors were replicated. Moreover, positive changes observed in the SM course were maintained even 3 months after the course termination. Considering the scant research on the long-term effects of self-control training,^10,39 the maintenance effects found in the 3-month FU are encouraging. Even though the current study did not examine the mechanisms underlying the generalization or maintenance of effects, the observed effects could be related to behavior strategies that the SM course incorporated to induce the change in target behavior. This is inferred from recent meta-analyses on strategies to promote self-control,^40,41 which revealed that cognitive and behavioral training focused on enhancing specific skills was effective and long-lasting.

As in Study 1, a reduction in impulsivity and a maintenance of effects were not found in the PE group, which is inconsistent with previous studies. As pointed out in Study 1, although the differences in the intensity of PE programs across studies may account for such inconsistency, replicated findings from this study suggest that PE courses may not be as effective as SM courses. The effects of PE on self-control require further examination.

In contrast with Study 1, group differences in changes of impulsivity level measured via the BIS-11 in Study 2 were not significant. It is plausible that the differences in sample size between the SM and PE groups and the measurement sensitivity of the tool used may explain the different results. In Study 1, the BIS-11 score of the SM group decreased at post-test, while the BIS-11 post-test score of the PE group increased. In other words, the impulsivity level decreased after the course in the SM group only. However, in Study 2, the BIS-11 scores decreased at post-test for both the SM and PE courses. Relatedly, one study showed that the delay discounting task was more sensitive to detecting behavioral changes related to self-control than the BIS-11.²⁹ In another study, students who successfully accomplished their goals after the SM course showed reduced impulsivity only in the delay discounting task, not in the BIS-11.²³ Therefore, the results of the current study support that a behavioral task is a more sensitive tool for measuring actual behavioral changes than a self-report method.

General discussion

The aims of this study were to investigate the effects of college-level SM and PE courses on improving self-control, assessed in terms of reducing impulsivity, the generalization of improved self-control to daily behaviors, and the maintenance of the effects over time. Study 1 showed a significant decrease in impulsivity and a generalization of self-control after the course only in the SM group. In Study 2, in addition to the replication of the findings from Study 1, the effects were maintained for 3 months after the course’s termination. In contrast, significant changes in impulsivity and self-control behaviors were not observed in the PE group.

The current study has several implications. First, it is encouraging that the SM course can be used as a training tool for improving self-control, given its effects in reducing impulsivity, generalizability, and maintenance effects. Impulsivity is a major factor in health-threatening behaviors, such as substance abuse, drinking, and smoking, and a large number of college students engage in these impulsive behaviors.^42–47 Thus, it is vital for college students to acquire effective self-control skills. Despite their proven effectiveness,¹⁸ SM courses have rarely been offered by public and private universities in Korea and the US. The finding that impulsivity reduced after the SM course could provide additional support for disseminating such courses in college settings. The classroom-based delivery of SM programs is cost-effective and easily accessible in that it can reach a large number of people simultaneously and does not require continuous intervention efforts or external assistance. Such programs also have a broader impact as school- and community-based interventions to improve the quality of life of community members. In addition, in contrast to laboratory tasks with low social validity (e.g., practice using non-dominant hands or prohibiting using certain words) adopted in previous studies, SM programs have a high social validity as they target behavioral changes in daily life by teaching various strategies that learners can flexibly utilize according to their individual motivations and interests.

This study also provides supporting evidence for the strength model, according to which self-control can be enhanced through recurrent practice. On the basis of the strength model, the current study hypothesized that both SM and PE courses are effective in reducing impulsivity. Interestingly, however, only the SM course was found to be effective. The SM and PE courses were comparable in that they both incorporated recurring self-control training into the curriculum. However, only the SM course taught students about the behavioral theory and strategies. The relative effectiveness of SM in the current study suggests that learning about specific behavioral strategies and skills is more important than the recurrent practice of self-control behaviors or, at least, that they should accompany one another for self-control training to be effective. As discussed earlier, in light of studies that have demonstrated the facilitation of sustained and extensive self-control via cognitive-behavioral skill acquisition and practice, further research is required to understand the underlying mechanism of how SM courses improve self-control and how the effects are generalized and maintained. For example, recent studies have examined the associations between the strength model and executive functions (EFs).^48–50 These studies have explored the link between EFs and the strength model since EFs may be depleted in the short term when used repeatedly but can be reinforced with long-term training,⁵¹ and behavioral inhibition, one of the core components of EFs, is an important factor in self-control. In addition, there have been recent attempts to complement the strength model on the basis that cognitive factors such as motivation and attention can offset the effects of self-depletion.⁵² More integrated alternative approaches may therefore provide a better explanation of the mechanism of self-control improvement above and beyond the traditional strength model.

Meanwhile, careful interpretation is needed regarding the ineffectiveness of the PE course on self-control. Unlike previous studies that reported the effectiveness of PE in improving self-control,^13,27 we did not find any significant reduction in impulsivity after the PE course. This null finding could be attributable to the intensity level of the course, as noted in the Discussion sections above and in the results of Kim et al’s study,²⁷ which showed that a reduction in impulsivity was only observed after an intensive PE course. This suggests that a PE course needs to be highly intensive to be effective at reducing impulsivity; this needs to be tested via experimental studies. Considering the greater accessibility and efficiency in terms of required resources and preparation for a PE program than for an SM program, future studies should examine whether different intensity levels of PE courses have an impact on self-control improvements to determine the functionality of PE as a self-control improvement strategy.

The limitations of the study and suggestions for future studies are as follows. First, the sample sizes and gender distribution were not equal among the groups. While these factors were statistically controlled in the analysis, it did not entirely resolve the issues regarding validity and reliability of the results. However, this study, as a type of translational research, focused on examining the effects of interventions in a real-world setting. In this study, participants were recruited from real college courses; thus the data for each group reflected the class size and gender distribution of each course. Given the samples are representative of the population in question in translational research,⁵³ the present study still remains as a meaningful attempt to bridge the gap between research and practice in the field of college health promotion, despite having some methodological limitations. Future studies recruiting equal number of participants across groups or randomized controlled trials are desirable for the generalization of the results from this study. Second, the attrition rate of participants in the FU was high, at around 50%, in Study 2. Although participation in the FU was additionally rewarded to lower the attrition rate, there were limitations to encouraging participation via email after the course ended. A high dropout rate above 50% is often expected in behavioral SM studies^54,55 and in Internet-based studies that communicate with participants solely by email.⁵⁶ Future studies should consider recruiting a larger number of participants to compensate for the high attrition rate or to implement a motivation system to encourage higher participation in the FU study. Finally, this study examined a relatively short-term period of maintenance of the effects of the SM program, 3 months after the end of course. Future studies can examine longer-term changes in effect by using a 6-month or 1-year FU, for instance. Despite the theoretical and empirical emphasis on the maintenance of the self-control training effect, the related research is very limited. Investigating the underlying mechanisms and factors that influence the maintenance of the effects is also recommended.

Conflict of interest disclosure

The authors have no conflicts of interest to report. The authors confirm that the research presented in this article met the ethical guidelines, including adherence to the legal requirements, of Korea and received approval from the Institutional Review Board of Yonsei University.

Additional information

Funding

This work was supported by the Bio Industrial Technology Development Program (20009392, Development of mental and physical health assessment algorithm based on digital phenotyping data and intervention infrastructure) funded By the Ministry of Trade, Industry and Energy (MOTIE, Korea).