The Journal of Genetic Psychology, 2002, 163(3), 296–304
Worry and Sleep Length Revisited: Worry,
Sleep Length, and Sleep Disturbance
Ascribed to Worry
WILLIAM E. KELLY
Department of Counseling
University of Nevada, Las Vegas
ABSTRACT. The author administered university students (N = 222; 152 women, 70 men)
the Worry Domains Questionnaire (F. Tallis, G. C. L. Davey, & A. Bond, 1994) and a newly
constructed scale (the Sleep Disturbance Ascribed to Worry Scale) to measure sleep dis-
turbance attributed to worry. To revisit previous studies (i.e., E. Hartmann, F. Baekeland,
& G. R. Zwilling, 1972; S. J. H. McCann & L. L. Stewin, 1988) that suggested that sleep
length was positively related to worry, the author also asked the students a question about
habitual sleep length. The results indicated that worry and sleep disturbance attributed to
worry were negatively related to sleep length. A regression analysis revealed that worry
was significantly negatively related to habitual sleep length irrespective of sleep distur-
bance ascribed to worry.
Key words: sleep disturbance, sleep length, worry
RECENT STUDIES of individual differences in sleep length have indicated that
individuals who habitually sleep 6 hr or fewer every 24 hr (short sleepers), as
compared with those who sleep 9 hr or more every 24 hr (long sleepers), are more
likely to be neurotic, demonstrate less creativity, report more hallucinations,
endorse more eating disorders, and perform more poorly academically (Hicks,
Guista, Schretlen, & Pellegrini, 1980; Hicks & Rozette, 1986; Kelly, Kelly, &
Clanton, 2001; Kumar & Vaidya, 1982; Soper, Kelly, & Von Bergen, 1997).
Research has also been conducted on mood states associated with sleep length.
Kumar and Vaidya (1984) administered an anxiety scale to 25 long sleepers and
25 short sleepers randomly selected from a sample of 500 undergraduate students.
The results indicated that short sleepers reported significantly more anxiety than
did long sleepers.
Address correspondence to William E. Kelly, who is now at the Graduate Department of
Counseling, George Fox University, Portland Center, 12753 S.W. 68th Avenue, Portland,
OR 97223; email@example.com (e-mail).
It has been posited that the lower levels of cortical arousal of short sleep-
ers may influence their psychological makeup and symptomatology (Skinner,
1983; Soper et al., 1997). Specifically, short sleepers may compensate for lower
levels of cortical arousal with increased levels of cognitive activity. This sug-
gestion has been partially supported by other research. Coursey, Buchsbaum,
and Frankel (1975) found that short sleepers demonstrated lower levels of sen-
sory response and performed worse on a perceptual-motor task, both of which
suggested lower levels of cortical arousability (Banich, 1997). The brain
appears to combat decreases in arousability by using compensatory strategies
and plasticity, or flexibility (Banich; Howard, 1994; Salthouse, 1991). This
point illustrates the resiliency of the human brain and its attempts to activate
itself in instances of damage and lower levels of activity. It seems likely, then,
that short sleepers would manifest cognitive symptoms to compensate for low
cortical arousal. Soper et al. (1997) argued that their findings of increased hal-
lucinations among short sleepers supported this hypothesis.
In addition to hallucinations, another potential cognitive symptom of lower
cortical activity is worrying. It has been posited that worry serves as a stimu-
lant during periods of decreased activity, such as boredom (Kelly & Markos,
2001). Moreover, worry has been associated with increased levels of cortical
activity (Borkovec, Shadick, & Hopkins, 1991). Thus, it is possible that short
sleepers, who have decreased levels of cortical arousal, would show an increase
in worry as their brain attempts to increase cortical activity. Interestingly, how-
ever, previous researchers have found just the opposite—that worry, loosely
defined, was reported more often by long sleepers.
Nearly 3 decades ago, Hartmann, Baekeland, and Zwilling (1972) con-
ducted a pioneer study examining individual differences among 28 short sleep-
ers and 24 long sleepers. One primary finding was that long sleepers were
described as “worriers” and were psychologically less healthy, whereas short
sleepers were psychologically healthy. The research reviewed previously using
larger, more diverse samples, however, indicated the opposite: that short sleep-
ers are psychologically less healthy than long sleepers. Only one subsequent
study has been published examining sleep length and worry. McCann and
Stewin (1988) administered to 101 undergraduate students a sleep-length pref-
erence question and a Likert-type single-item assessment of individuals’ status
as worriers. The results indicated a modest, but significant, positive correlation
between worrier status and sleep preference.
Although there is some consistency between the Hartmann et al. (1972)
and McCann and Stewin (1988) studies, there are methodological inconsis-
tencies and problems with these studies. For instance, whereas Hartmann et al.
found that long sleepers scored higher on a measure of nervousness and anx-
iety, no definition or standardized measure of worry was used. Instead, the
authors merely described nervousness and anxiety as worry. More recent find-
ings, however, indicate that worry and anxiety are separate constructs (Davey,
298 The Journal of Genetic Psychology
1992; Davey, Hampton, Farrell, & Davidson, 1991). Furthermore, the partici-
pants appear to have been limited to men recruited from newspaper advertise-
ments, nearly half of whom were eliminated by the researchers prior to data
analyses because of sleep irregularities, because they had been identified as
average sleepers (sleeping between 6 and 9 hr per night), or because they had
scored high on measures of psychopathology. Because of the quantity of par-
ticipant deletions, however, it is difficult to understand what population the
McCann and Stewin (1988) assessed worry using only a single-item
assessment and reported data of a sleep-preference assessment as opposed to
a habitual-sleep-length assessment. Hence, the samples and sleep-length
assessments of these studies were not consistent. In addition, the samples of
both studies were small. One potential methodological problem with both of
these studies is the lack of a reliable, valid assessment of worry and a measure
of sleep disturbance related to worry. Previous studies have implicated cogni-
tive activity as a source of sleep disturbance, especially insomnia, and sleep
reduction (Nicassio, Mendlowitz, Fussell, & Petras, 1985; Watts, Coyle, &
An important element in the discussion of sleep length and worry is the
lack of a cause–effect relationship regarding whether worriers are at risk of
developing sleep disturbances or having specific habitual sleep lengths, or
whether individuals with sleep disturbances or certain habitual sleep lengths are
more likely to worry. Thus, it could be argued that worriers report less sleep
because they are often unable to sleep because of worrying, rather than that
short sleepers tend to compensate for suboptimal cortical arousal by worrying.
My purpose in the present study, therefore, was to clarify the relationship
between worry and sleep length in a college-student population by investigat-
ing further the relationship between a reliable, valid measure of worry and
habitual sleep length and the possible effects on this relationship by sleep dis-
turbance ascribed to worry. For the latter question, it was necessary to devise a
scale to measure sleep disturbance attributed to worry. Therefore, I attempted
to develop a brief scale to measure sleep disturbance ascribed to worry and to
investigate the relationships among habitual sleep length, the Sleep Disturbance
Ascribed to Worry Scale (SAW), and a measure of worry.
Because of possible methodological problems with the Hartmann et al.
(1972) and McCann and Stewin (1988) studies, as well as the preponderance
of research findings indicating that short sleepers are more anxious and less
psychologically healthy (see previous review), I expected that high scores on a
reliable, valid measure of worry would be reported more by short sleepers than
by long sleepers. Considering the notion that short sleepers compensate for sub-
optimal cortical arousal through cognitive activities, I expected that individu-
als who worry more would endorse less sleep regardless of sleep difficulties
related to worry.
Participants included 222 students (152 women, 70 men) enrolled in under-
graduate and graduate human services courses at a midsized U.S. university.
The mean age of the sample was 24.4 years (SD = 8.0), with ages ranging from
18 to 65 years. There was no significant difference between the genders on age,
t(220) = .61, p = .54. The majority of participants identified themselves as Cau-
casian (66%). Other ethnicities included African American (10%), Latino
(10%), and Asian American (11%). Five (2%) identified their ethnicities as
“other.” Two respondents (1%) did not identify their ethnicities.
The Sleep Disturbance Ascribed to Worry (SAW) Scale. Because no scales specif-
ically measuring sleep disturbance related to worry could be located at the time
of this study, I developed the SAW for the present study as a brief instrument to
measure sleep disturbances ascribed to worry. Previous literature indicated that
the predominant sleep disturbance associated with worry was insomnia (e.g.,
Nicassio et al., 1985; Watts et al., 1994). Hence, insomnia was the primary sleep
disturbance evaluated and used in developing this scale. I used the following pro-
cedure to develop the SAW: After reviews of the criteria for insomnia in the
Diagnostic and Statistical Manual of Mental Disorders (fourth edition; DSM–IV;
American Psychiatric Association, 1994), the pertinent literature on sleep distur-
bance and worry, and Coren’s (1988) Insomnia Scale, I developed 5 items assess-
ing the attribution of sleep disturbance to worry, effects of worry on sleep, and
frequency of sleep disturbance ascribed to worry. The result was a 5-item ques-
tionnaire that I administered by asking participants to respond to each item using
an 11-point anchored-response scale (0 = never, 10 = very often). Higher scores
on the SAW indicate more sleep disturbance ascribed to worry.
Worry Domains Questionnaire (WDQ). The original WDQ (Tallis, Eysenck, &
Mathews, 1992) included 30 items assessing six domains of worry. However, one
domain (social concerns) was found to be highly related to social desirability.
Therefore, the 25-item WDQ presented by Tallis, Davey, and Bond (1994) was
used for this study. The 25-item WDQ measures a tendency to worry across five
general worry domains: (a) relationships, (b) lack of confidence, (c) aimless
future, (d) work-related, and (e) financial. WDQ items were presented with a 5-
point Likert-type scale on which individuals described the extent (not at all to
extremely) to which their experience was in agreement with individual items. I
summed responses from the appropriate items to create a total WDQ score.
Higher scores denote more worry. Test–retest reliability (r = .79 after 4 weeks),
300 The Journal of Genetic Psychology
internal consistency (α = .92), and validity of the WDQ have been found to be
satisfactory (Tallis et al., 1994).
Sleep length. I provided sleep length estimates as a continuous variable by using
the method of Kumar and Vaidya (1984), whereby participants are asked to write
the amount of time, in hours and minutes, that they habitually sleep, on average,
in every 24-hr period.
After the participants gave informed consent, they were administered the
WDQ, SAW, and sleep length question in random order. As a validity check for the
SAW, 46 of the participants were also asked to respond to the following question:
“Use the following scale to describe the general quality of your sleep.” The partic-
ipants responded to this question on an 11-point Likert-type scale ranging from
awful to great. A demographics survey that solicited information about the partic-
ipants’ age, gender, and ethnicity was also completed. The participants completed
the questionnaires in a group setting. Specific information regarding the nature of
the study was not disclosed until the participants had returned the questionnaires.
Properties of the SAW
To explore the factor structure of the 5 SAW items, I calculated a principal
components factor analysis. This analysis resulted in a single factor accounting
for 62.5% (eigenvalue = 3.12) of the systematic variance. SAW items, their
respective factor loadings, and means and standard deviations are presented in
Table 1. The internal consistency of the SAW was also explored. The coefficient
alpha of the 5 items was .85. Age was not correlated with SAW scores, r = –.01,
p = .79. There was a significant difference between the genders for SAW scores,
t(220) = 3.03, p < .003, with women (M = 19.3) endorsing higher SAW scores
than men (M = 14.8). A one-way analysis of variance (ANOVA) revealed no sig-
nificant difference in SAW scores among ethnicities, F(5, 214) = 1.23, p = .29.
The correlation between the quality of sleep question and SAW scores was sig-
nificant, r = –.37, p < .01, indicating that sleep disturbance resulting from worry
negatively affects quality of sleep.
WDQ, Sleep Length, and SAW Analyses
Pearson correlations among WDQ scores, SAW scores, and sleep length
were calculated. Sleep length was significantly negatively related to both WDQ
(r = –.24, p < .01) and SAW scores (r = –.20, p < .01). SAW scores and WDQ
scores were also significantly related (r = .42, p < .01). Next, gender differences
in WDQ scores and sleep length were explored. No significant gender differences
were found for WDQ scores, t(220) = 1.63, p = .12, or sleep length, t(220) = 1.29,
p = .20. Also, no significant correlations were observed between age and sleep
length (r = –.05, p = .42) or age and WDQ scores (r = –.09, p = .18).
Because of the significant gender differences found for SAW scores, it was
necessary to identify any significant Gender × SAW interactions on WDQ scores
and sleep length prior to further analyses involving these variables. Hence, I used
a median split to create a high and low group on the basis of SAW scores. I cal-
culated two separate 2 (gender) × 2 (SAW group: high or low) ANOVAs to iden-
tify significant interactions between gender and SAW scores on sleep length and
WDQ scores. The first ANOVA revealed no significant interaction between SAW
scores and gender for sleep length, F(1, 218) = .82, p = .37. The second ANOVA
also did not reveal a significant Gender × SAW score interaction for WDQ scores,
F(1, 218) = 1.05, p = .31. Thus, subsequent analyses involving SAW scores, WDQ
scores, and sleep length were analyzed irrespective of gender.
I calculated a simple regression using sleep length as the criterion and WDQ
scores as the predictor variable to test the relationship between worry and habitual
sleep length. WDQ scores accounted for 6% of the variance in sleep length, which
was significant, F(1, 220) = 13.50, p < .0001. An analysis of the beta weights indi-
cated a negative relationship between worry and sleep length (β = –.24).
I calculated a multiple regression to explore the influences of SAW scores
on the relationship between worry and sleep length, using sleep length as the
Means, Standard Deviations, and Factor Loadings for SAW Items
1. How often do you awaken early from your normal
sleeping time and are completely unable to return to
sleep because of worry? 3.39 2.47 .79
2. How often are you unable to stop worrying at bedtime? 4.09 2.75 .71
3. How often can you not get to sleep for several
minutes because of worry? 4.25 2.94 .74
4. How often do you awake from sleep worrying? 2.81 2.59 .87
5. During your usual sleep time, how often do you
awaken and are unable to return to sleep for several
minutes because of worry? 3.30 2.58 .83
Total scale 17.84 10.50
Note. N = 222. SAW = Sleep Disturbance Ascribed to Worry Scale.
302 The Journal of Genetic Psychology
criterion and SAW and WDQ scores as predictor variables. To account for SAW
variance in the relationship between sleep length and worry, I forced SAW
scores to enter first in the equation. SAW scores accounted for 4% of the vari-
ance in sleep length, which was significant, F(1, 220) = 9.02, p < .003. The beta
weight indicated a negative relationship between sleep disturbance ascribed to
worry and sleep length (β = –.20). On Step 2, WDQ scores were entered and
accounted for an additional 3% of the variance, which was also significant, F(1,
219) = 7.02, p < .009, and remained negative (β = –.19).
The results of the present study indicate that worry is negatively related to
habitual sleep length. In other words, individuals who sleep less tend to worry
more often. This finding is not consistent with that of previous researchers (i.e.,
Hartmann et al., 1972; McCann & Stewin, 1988) who found that worry was pos-
itively related to sleep length. However, the results of the present study are con-
sistent with those of other researchers that indicated that individuals who habit-
ually sleep less are less psychologically healthy and more anxious (Hicks &
Rozette, 1986; Kumar & Vaidya, 1982, 1984; Soper et al., 1997).
It also appears that the negative relationship between sleep length and worry
exists irrespective of sleep disturbances attributed to worry. Thus, the lower sleep
lengths of worriers do not appear to be the result of only sleep disturbance due to
worrying. As postulated by previous researchers (i.e., Skinner, 1983; Soper et al.,
1997), it may be that the less active neural systems of short sleepers tend to stim-
ulate themselves by creating additional sensory and cognitive activity through wor-
rying, but not necessarily during times of sleep. Further research is needed to sub-
stantiate this claim. Measures of excitement seeking and need for cognition as well
as neuropsychological measures might help researchers explore the cortical
arousal explanation of worry among short sleepers. Furthermore, the relatively low
percentage of variance accounted for by SAW and WDQ scores suggests that addi-
tional variables may be involved in determining habitual sleep length.
The significant relationships found between the SAW and the WDQ, sleep
length, and the sleep quality question provide some initial evidence for the valid-
ity of the SAW and, thus, provide some credence to the findings of this study
involving the SAW. Nevertheless, future researchers who use the SAW to exam-
ine sleep disturbance attributed to worry should include additional measures to
validate the SAW. Furthermore, a distinction has been drawn between transient
and chronic sleep disturbance (Giesecke, 1987). Hence, future researchers should
collect test–retest data on the SAW to determine whether it measures transient or
chronic sleep disturbance attributed to worry.
The discrepancy between the results of the present study and the results of
McCann and Stewin (1988) suggest that habitual sleep length and sleep length
preference are separate constructs. Researchers should, therefore, investigate both
of these constructs in relation to worry. In addition, future research should also
include additional measures of worry. The WDQ is coined as a measure of non-
pathological worry (Tallis et al., 1992). Hence, it might be useful for researchers
to include a measure of worry such as the Penn State Worry Questionnaire
(Meyer, Miller, Metzger, & Borkovec, 1990), designed to assess pathological
worrying. There may be differences in sleep length among nonpathological and
pathological worriers. Also, sleep diaries should be considered as a measure of
sleep length in addition to single-item measures of habitual sleep length.
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC: Author.
Banich, M. T. (1997). Neuropsychology: The neural bases of mental function. New York:
Borkovec, T. D., Shadick, R. N., & Hopkins, M. (1991). The nature of normal and patho-
logical worry. In R. M. Rapee & D. H. Barlow (Eds.), Chronic anxiety: Generalized
anxiety disorder and mixed anxiety-depression (pp. 29–51). New York: Guilford Press.
Coren, S. (1988). Prediction of insomnia from arousability predisposition scores: Scale
development and cross-validation. Behaviour Research and Therapy, 26(5), 415–420.
Coursey, R. D., Buchsbaum, M., & Frankel, B. L. (1975). Personality measures and evoked
response in chronic insomniacs. Journal of Abnormal Psychology, 84(3), 239–249.
Davey, G. C. L. (1992). A comparison of three worry questionnaires. Behaviour Research
and Therapy, 31(1), 51–56.
Davey, G. C. L., Hampton, J., Farrell, J., & Davidson, S. (1991). Some characteristics of
worrying: Evidence for worrying and anxiety as separate constructs. Personality and
Individual Differences, 13(2), 133–147.
Giesecke, M. E. (1987). The symptom of insomnia in university students. Journal of Amer-
ican College Health, 35(5), 215–221.
Hartmann, E., Baekeland, F., & Zwilling, G. R. (1972). Psychological differences between
long and short sleepers. Archives of General Psychiatry, 26, 463–468.
Hicks, R. A., Guista M., Schretlen, D., & Pellegrini, R. J. (1980). Habitual duration of
sleep and divergent thinking. Psychological Reports, 46, 426.
Hicks, R. A., & Rozette, E. (1986). Habitual sleep duration and eating disorders in college
students. Perceptual and Motor Skills, 62, 209–210.
Howard, P. J. (1994). The owner’s manual for the brain: Everyday applications from mind-
brain research. Austin, TX: Bard Press.
Kelly, W. E., Kelly, K. E., & Clanton, R. C. (2001). The relationship between sleep length
and grade-point average among college students. College Student Journal, 35(1),84–86.
Kelly, W. E., & Markos, P. A. (2001). The role of boredom in worry: An empirical inves-
tigation with implications for counsellors. Guidance and Counselling, 16(3), 81–85.
Kumar, A., & Vaidya, A. K. (1982). Neuroticism in short and long sleepers. Perceptual
and Motor Skills, 54, 962.
Kumar, A., & Vaidya, A. K. (1984). Anxiety as a personality dimension of short and long
sleepers. Journal of Clinical Psychology, 40, 197–198.
McCann, S. J. H., & Stewin, L. L. (1988). Worry, anxiety, and preferred sleep length. The
Journal of Genetic Psychology, 149(3), 413–418.
Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and
validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy,
304 The Journal of Genetic Psychology
Nicassio, P. M., Mendlowitz, D. R., Fussell, J. J., & Petras, L. (1985). The phenomenolo-
gy of the pre-sleep state: The development of the pre-sleep arousal scale. Behaviour
Research and Therapy, 23(3), 263–271.
Salthouse, T. A. (1991). Expertise as the circumvention of human processing limitations.
In K. A. Ericsson & J. Smith (Eds.), Toward a general theory of expertise: Prospects
and limits (pp. 286–300). New York: Cambridge University Press.
Skinner, N. F. (1983). Neuroticism, extraversion, and sex differences in short and long
sleepers. Psychological Reports, 53, 669–670.
Soper, B., Kelly, W. E., & Von Bergen, C. W. (1997). A preliminary study of sleep length
and hallucinations in a college student population. College Student Journal, 31(2),
Tallis, F., Davey, G. C. L., & Bond, A. (1994). The Worry Domains Questionnaire. In G.
Davey & F. Tallis (Eds.), Worrying: Perspectives on theory, assessment and treatment
(pp. 286–297). New York: Wiley.
Tallis, F., Eysenck, M., & Mathews, A. (1992). A questionnaire for the measurement of
nonpathological worry. Personality and Individual Differences, 13, 161–168.
Watts, F. N., Coyle, K., & East, M. P. (1994). The contribution of worry to insomnia.
British Journal of Clinical Psychology, 33(2), 211–220.
Received July 5, 2001