Addicted to Running?

Controversy exists regarding the psychological effects of distance running. While running has been used as an adjunctive treatment for depression and anxiety, it has also been characterized as a negative addiction. In the present study, anxiety and depression traits were measured for a heterogenous group of runners. These runners also provided information regarding their motivations for running and their training patterns. Those runners classified as Highly Committed had lower anxiety traits and depression traits than those classified as Recreational Runners. Regardless of their level of commitment, health and fitness concerns were the strongest motivators for training. These data provide evidence that strong dedication to distance running is associated with positive traits rather than with negative aspects of addiction.

Distance running, initially popularized as a means of achieving physical health (Fixx, 1977), has been promoted as an aid to maintaining or achieving mental health as well (Byrne & Byrne, 1993). According to Glasser (1976), running is the "hardest but surest way to positive addiction", assisting people in finding strength and power, and leading a fuller life (p. 100). The characterization of running as a positive addiction helped popularize the first "running boom" in the 1970's, spurring Carmack and Martens (1979) to develop a scale to quantify the concept of positive addiction.

The positive effects of aerobic exercise, such as running, have continued to be an important topic in the intervening years. Martinsen and Morgan (1997) conclude that, in spite of relatively few controlled studies, there is sufficient evidence to support exercise as an antidepressant for those persons who are clinically depressed or who have elevated depression scores. Running has been associated with decreased levels of depression, comparable to that seen with counseling (Greist, Klein, Eischens, Gurmam, & Morgan, 1979). "Running therapy" has even been utilized to promote cognitive processing, and non-verbal communication between client and therapist (Hays, 1994).

Similarly, Raglin (1997) concludes that acute vigorous exercise can reduce transient levels of anxiety, while chronic exercise programs reduce trait anxiety levels, or general disposition for anxiety. This reduction in anxiety, like that reported for depression, is most apparent in clinical populations or for those subjects with elevated anxiety levels prior to exercise. The specific effects of distance running on mood states in non-clinical populations have also been studied. Morris and Salmon (1994) determined that distance runners experience a decreased negative mood and an increased positive mood just after a run.

In contrast, chronic exercise regimes have been suggested to have a negative impact on mental health. Running in particular has been viewed as resulting in a "negative addiction", in that runners have been reported to use their running to help them cope with daily stresses, just as some people turn to drugs or alcohol (Conboy, 1994; Hailey & Bailey, 1982; Rudy & Estok, 1989;1990).

Scales of negative addiction have been developed to measure the degree to which running dominates a person's life. Runners with higher scores on these scales have histories of running for more years and running longer distances than more moderate runners (Hailey & Bailey, 1982). Also, runners with higher negative addiction scores appear to have lower levels of self-esteem and higher levels of anxiety, according to Rudy and Estok (1989).

As further evidence of running addiction, it has been argued that regular runners may experience withdrawal when they are unable to run (Morgan, 1979). Conboy (1994) reports that, while runners' scores on the Profile of Mood States (POMS) are quite positive overall, their mood states are more negative on non-running days than on running days. Longer-term withdrawal effects, following two weeks of enforced non-running, have also been reported. Somatic complaints increased within the first non-running week, and symptoms of anxiety and depression were evident during the second week of enforced lay off (Morris, Steinberg, Sykes & Salmon, 1990).

In the face of these conflicting conclusions regarding the qualities of aerobic exercise, particularly distance running, the present study was designed to further explore the relationships between commitment to distance running and mental traits. By examining multiple variables and utilizing a heterogeneous sample of runners, as well as a comparison group of non-runners, specific information regarding these relationships could be determined.

Method

Participants and Procedures

All participants in the present study were recruited at a sports and fitness exposition and pre-race dinner held in conjunction with a running event in the Midwest. The event included races of 5 kilometers (5K), 10 kilometers (10K), a marathon relay (each participant ran approximately 5 miles), half marathon (13.1 miles) and full marathon (26.2 miles). Runners and their friends and family members attending the exposition were asked to participate in the project. All participants completed the questionnaire prior to leaving the exposition.

Questionnaire

The questionnaire contained statements, requiring a 5-point Likert-type response, designed to measure the participants' overall disposition for anxiety and for depression. These statements were designed specifically for this study of runners, and were based on the Diagnostic and Statistical Manual-IV, Fourth Edition (DSM-IV) criteria (APA, 1994) for generalized anxiety disorder or panic attack for anxiety traits, and for major depressive disorder for depression traits. These scales were not meant to provide differential diagnosis, nor to measure the current symptom state of the individuals. Rather, they were designed to provide an overall indication of the level of anxiety and depression traits experienced by the respondents.

Statements for both of these scales related to subjective components: "I often feel nervous and anxious" (anxiety) and "I am generally a happy person" (depression); cognitive components: "I generally believe things will turn out all right" (anxiety) and "I find it difficult to make decisions" (depression); and for physical components: "I am bothered by headaches, neck or back pains" (anxiety) and "I usually get a good night's sleep" (depression). Finally, all participants were asked to indicate if they had ever been diagnosed with an anxiety disorder or with depression.

Runners were asked to supply information regarding their history of running and racing. In addition, they were asked to respond to ten statements adapted from the running as a negative addiction scale (Rudy & Estok, 1989), and 19 statements adapted from the personal incentives for exercise scale (Duda & Tappe, 1989). A 5-point Likert scale was utilized for both sets of statements. The content of these statements is presented in Tables 2 and 4.

Results

Participant Characteristics

A total of 276 people completed the questionnaires, including 37 non-runners and 239 runners. The majority of the non-runners were women (62.2%), while the majority of the runners were men (56.1%). The mean age was 40.5 years for the non-runners and 37.9 years for the runners. There were no differences in the ages based on running status or gender.

Using the respondents' self-reports, 16.2% of the non-runners and 4.6% of the runners indicated that they had been diagnosed with an anxiety disorder or prescribed an anxiolytic medication at some point in their life. Likewise, 27.0% of the non-runners and 11.8% of the runners reported that they had been diagnosed with depression or prescribed an antidepressant medication at some point in their lives. No differences in the reported rate of diagnosis of either depression or anxiety were found based on gender.

Composite scores were computed for the depression and anxiety traits for all participants. The Chronbach's alpha for the depression trait scale was .89, with split half correlations of.87. For the anxiety trait scale, Chronbach's alpha was .82, with split half correlations of.78. Content validity of these trait measures was built in by including items which relate to the three aspects of anxiety and/or depression detailed in the DSM-IV, i.e., physical, cognitive and emotional components (APA, 1994). As a measure of criterion validity, a one-way analysis of variance (ANOVA) was run on the trait levels, comparing those people who had received a relevant diagnosis to those who had not. Those participants who reported having been diagnosed with an anxiety disorder or having been prescribed anxiolytic medications (n = 17) had significantly higher anxiety trait scores than those without such a diagnosis (50.71 2.39 and 40.15 t.60, respectively), F(1, 274) = 18.87, p < .0001. Likewise, those participants who reported having been diagnosed with depression or having been prescribed antidepressants (n = 38) had significantly higher measures of depression traits (59.87 2.48, and 49.43 .79), F (1, 274) = 22.46,p <.0001.

Characteristics of Runners. The runners averaged approximately 11.5 years of running, with no difference between the men and the women. However, proportionately more men than women were registered to run the marathon, while more women were running the shorter races (10K or less), )2 (2, N=236) = 22.0,p <.001. Two-way ANOVAs, using runner gender and race distance as the main effects and participant age as the covariate, were computed for various runner characteristics. Those runners entered in the longer races (half marathon or longer) reported higher training mileage, F (2,224) = 26.52, p <.001, and more long races completed, F (2,228) = 5.03,p <.01. While men ran slightly more miles per week and had completed slightly more long races than women, both of these differences failed to reach significance when race distance was taken into account, p < . 10. These results are summarized in Table 1.

Two-way ANOVAs were also computed for the anxiety and depression trait scores. No differences were found for either main effects of gender or race length, and there was no interaction between these two variables for either trait.

Running Commitment Scores

A principal components factor analysis, using quartermax rotation, was conducted for the 10 statements that were adapted from Rudy and Estok's (1989) Running Addiction Scale. Two factors emerged, accounting for 46.4% of the variance. The first, labeled Stress Relief, was composed of statements that focus on running to help the person cope with life stress. The second factor, labeled Adherence to Training, was composed of statements that reflect consistent running, even under adverse conditions. The statements defining each factor, and their factor loadings, are presented in Table 2.

Composite factor scores were calculated by summing the participants' scores for each of the statements contained in that factor. Two-way ANOVAs, with runner gender and race distance as main effects and age as a covariate, were computed for the Stress Relief Factor. As shown in Figure 1, the women's Stress Relief scores were significantly higher than the men's, F(l,229) =20.51 ,p <.001. Scores on this factor also varied across race length, F(2, 229) = 6.47, p <.005, with subsequent Scheffe's analysis indicating that the runners entered in the 5K - 10K races had lower Stress Relief scores than those running the half or full marathon. No interaction between gender and race length was found.

Table 1

Table 2

Women had higher scores than the men on the Adherence to Training factor as well, F(1, 229) = 6.27, p <.05. Again, there was a difference across race distances, F (2,229) = 12.47,p < .01, with the 5K - 10K racers having lower Adherence to Training scores than either the Half or Full Marathoners. A significant interaction between runner gender and race distance was found for this factor, F (2, 229) = 5.32, p < .01. As shown in Figure 1, men's Adherence to Training scores were fairly constant regardless of race distance, while women in the longer races had considerably higher Adherence to Training scores than those women running the shorter races.

Correlations were computed between these two factor scores and various runner characteristics. Stress Relief scores were negatively correlated with age, and both Stress Relief and Adherence to Training scores were positively correlated with weekly mileage. However, there were important differences between the relationships of these two components and the measures of negative mental traits. Adherence to Training was negatively correlated with depression scores, and Stress Relief was positively correlated with the anxiety scores. Correlations among these variables are presented in Table 3. It should be noted that these correlations, while significant, account for only a small portion of the total variance.

Figure 1.

Committed and Recreational Runners

Runners were classified as either Committed or Recreational Runners based on their weekly mileage, Adherence to Training score, and number of long races completed. Initially, decile ranks were determined for each of these three variables, for men and women separately. These ranks were then summed to provide a Commitment score. The top and bottom 20%, regardless of gender, were then labeled Committed and Recreational Runners, respectively.

A one-way ANOVA, using age as a covariate, was used to compare anxiety and depression trait scores across the Committed Runner (n = 31), Recreational Runner (n = 46), and NonRunner groups (n = 39). As shown in Figure 2, a significant difference for the depression scores was found across these groups, F (2, 113) = 8.00, p < .001. Subsequent Scheffe's analyses found no difference between the non-runners and either of the runners groups; however, the Committed Runners had significantly lower depression scores than the Recreational Runners. Similar results were found for anxiety scores, F(2,113) = 5.73,p <.01. Again, Scheffe's tests indicated that the Committed Runners had lower anxiety scores than the Recreational Runners, while there was no difference between either runner group and the Non-- Runners.

Motivations for Running

Using a principal components factor analysis, the 19 statements that dealt with the runners' motivations for running were reduced to three factors, accounting for 49.1% of the variance. The first factor was characterized as Health /Fitness, and was composed of statements related to running for health, fitness, and weight control. The second factor was characterized as Challenge and included statements related to competing and running for stimulation. Factor 3 was characterized as Time Alone, and included statements related to running for solitude rather than social contact. These factor statements and their loadings are presented in Table 4.

Table 3

Figure 2.

Composite scores for each factor were calculated for all participants, and adjusted to a 5-- point scale. A two-way repeated measures ANOVA, using age as a covariate, was computed for the main effects of gender and race distance. Overall, the strongest motivator was Health/ Fitness, F(2,229) = 135.3,p <.001, for both men and women, and for all three race distances. Women had higher scores than men when collapsing across the three motivation factors, F (I, 229) = 7.00, p < .01. Further, there was a significant interaction between gender and the motivation factor scores, F(2,229)= 8.17,p<.01. As shown in Figure 3, this interaction was due to women's higher scores on Health /Fitness and Time Alone factors,,and men's higher scores on the Challenge factor.

Comparisons of the motivation factor scores were also made between the Committed and Recreational Runners. The Committed Runners had significantly higher scores across the motives, compared to the Recreational Runners, F(l, 156) = 7.00, p < .01. Again, the most strongly endorsed motivation factor was Health/Fitness, F(2, 156) = 39.13,p < .001. There was no interaction between runner group and motivation factor.

Discussion

The primary purpose of the present study was to examine the competing ideas that distance running can represent either a positive behavior that promotes mental as well as physical health, or a negative behavior analogous to substance addiction. By analyzing the relationships between running characteristics and motivations, and negative mood traits in a heterogeneous group of runners, a more complete picture is provided. The present study confirmed the relationship between anxiety traits and running for stress relief, one aspect of the scale of running as a negative addiction (Rudy & Estok, 1989; 1990). However, even the most committed runners in the present study do not experience an elevation in negative mood traits. Rather, these runners actually showed lower levels of depression and anxiety traits. Considering these results, it is now necessary to carefully examine the concept of negative addiction to running.

Table 4

Figure 3.

The question of whether commitment to running represents an addiction might be most appropriately analyzed in accordance with current mental disorder classification systems. Both the International Classification of Diseases, Tenth Edition (ICD- 10, World Health Organization, 1992) and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV, APA, 1994) refer to substance "dependence", rather than "addiction", which was dropped by the World Health Organization in 1964. As a general description of dependence, the DSM-IV states, "The essential feature of Substance Dependence is a cluster of cognitive, behavioral, and physiological symptoms indicating that the individual continues use of the substance despite significant substance-related problems" (APA, 1994, p. 178).

For running to be accurately characterized as a "negative addiction", it must be shown that running results in impairments, and that the running continues despite these impairments. While distance running is associated with several kinds of musculoskeletal injuries, these are generally associated with a recent change in training, and can often be treated with modifications of the training program (Fredericson, 1996). Evidence that such injuries are related to "addiction" is quite limited. While Estok and Rudy (1987) have found that runners with the highest level of commitment to running were more likely to experience torn ligaments or hematuria, there were no differences found in the rates of several other types of running related injuries. In the present study, the most highly committed runners showed greater agreement with the statement that they continue to run when injured or ill than the least committed runners (means 3.67 and 2.24, respectively, with 3 indicating neither agreement nor disagreement). However, this statement was only moderately endorsed even by this group.

Further evidence of impairment resulting from a strong commitment to running is even more limited. Rudy and Estok (1990) found a negative correlation between non-runner's ratings of their spouse's level of addiction and their marital adjustment scores. It is important to note, however, that both the runners' and the spouses' levels of marital adjustment compared favorably to the normative data, thus failing to show an overall impairment in marital satisfaction for couples that include a distance runner.

Evidence of negative addiction to long distance running has also relied on the presence of negative mood states when runners are deprived of running (Conboy, 1994; Hailey & Bailey, 1982; Morgan, 1979; Morris et al., 1990; Rudy & Estok, 1989,1990). There is no clear evidence, though, that this dysphoria is actually an indication of "withdrawal". As stated by Conboy (1994), "The reason for the dysphoric mood change (or withdrawal) is still unclear. It may be that not running produces negative change or that running produces negative change or that running produces positive change" (p. 198). Likewise, Morris, et al., (1990) used a 2-week deprivation period and found increases in somatic and emotional symptoms during this "withdrawal" period. In their discussion, they state that, "An alternative explanation for the 'withdrawal' syndrome is that subjects have reverted to their 'natural', pre-running, state. That is, a beneficial effect of exercise to reduce such symptoms has been lost relatively rapidly once regular exercise has stopped." Regardless of the mechanism, the response to enforced non-running can be highly uncomfortable to committed runners.

In the present study, the negative addiction scale was factored into two components: running as a means of coping with life stressors (Stress Relief), and maintaining training in spite of obstacles (Adherence to Training). While a positive correlation was found between Stress Relief scores and Anxiety Trait scores, similar to the finding of Rudy and Estok (1989), this association by itself cannot be used to infer that consistent running increases anxiety. Alternatively, it may be that those runners who experience high levels of anxiety are more inclined to report running as a way to deal with their environmental stress. Indeed, Adherence to Training was negatively correlated with depression, and not related to anxiety traits.

Further support for running as a positive coping mechanism comes from the results in the present study regarding the differences in anxiety and depression traits for the Recreational and Committed Runners. In contrast to the expectations based on running as a negative addiction, the Committed Runners, i.e., those who are most "addicted", had lower levels of depression and anxiety traits than did the Recreational Runners, or least "addicted" runners. Thus, in a non-clinical population, decreased levels of depression and anxiety traits were seen for the highly committed runners, compared to those who are less committed. Again, these results must be interpreted with caution. It is possible that the low levels of negative traits found for the Committed Runners are not due to running per se, but that only mentally healthy people can train at such a high intensity.

The present study also examined the motivations for running. Health/Fitness was the most strongly endorsed motivation, for both men and women, and for Committed as well as Recreational Runners. Positive correlations between the Health/Fitness scores and both Stress Relief and Adherence to Training scores were found. Thus, even when the runners are highly committed to their running, health and fitness are still their primary concerns. This contrasts with the conclusion of Thornton and Scott (1995) that runners interested primarily in Health/Fitness would be unlikely to score high on commitment scales, and presumably would not train more than the minimum needed to reach fitness levels.

Gender and Running

Women have made tremendous inroads into distance running in the past two decades. In the Boston Marathon, which allowed women to compete for the first time in 1972, the percentage of women runners has now increased to over 30% (Boston Athletic Association, 1999). In a major marathon held in San Diego in 1998, 55% of the 19,978 participants were women (Jenkins, 1998). Thus women have taken their place in the distance running community.

In the present study, only slight, non-significant differences were found between men and women in their training and racing levels. However, differences in motivation and commitment levels were found based on gender. The women had higher commitment scores, measured by both Stress Relief and Adherence to Training factor scores. Such differences have been reported by other researchers as well (Gill, Williams, Dowd, Beaudoin & Martin, 1996; Masters & Lambert, 1989; Ogles, et. al., 1995). This greater level of commitment by women may be associated with the ongoing pressure and time constraints that women experience. Ogles et al., (1995) and Yates (1991) have discussed such gender differences as resulting from the interaction of several psychosocial variables, including child-bearing and child-rearing, physiology, and time constraints. Women's strong commitment to running may be imperative to maintaining a training program geared toward competition or even physical and psychological health. Women who lack this level of commitment are more likely to drop out of a running program, due to their multiple competing roles.

Conclusion

While there is now substantial evidence that runners may become highly committed to their training and racing schedules (Conboy, 1994; Morris, et al., 1990; Rudy & Estok 1989; 1990), a more careful look at the research indicates that this commitment does not necessarily indicate a "negative addiction". To follow through with the analogy of distance running to substance dependence, there must be evidence that runners continue to run even though it results in impairments in physical or mental health, work, social or family life.

In contrast, Martinsen and Morgan conclude that "the relationship between regular physical activity and psychological well-being has been very well established" (1997, p. 231). What has not been so clearly established is a cause and effect relationship, especially in nonclinical populations. The present study, while not addressing this directly, does provide an indication that high levels of commitment to running need not be associated with dependency or addiction type problems. Instead, the most highly committed runners are strongly motivated by health and fitness benefits, and have more positive mental trait profiles than do the least committed runners. Whether long distance running actually serves a protective function, or if only the mentally healthiest runners can maintain such a commitment, has not yet been determined.