See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/221804276

Is Loneliness the Same as Being Alone?

Article  in  The Journal of Psychology Interdisciplinary and Applied · January 2012

DOI: 10.1080/00223980.2011.589414 · Source: PubMed

CITATIONS READS

100 3,598

4 authors, including:

Daniel Wayne Russell Cynthia McRae

Iowa State University University of Denver
190 PUBLICATIONS   29,273 CITATIONS    43 PUBLICATIONS   880 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Big Brothers Big Sisters View project

family and community health View project

All content following this page was uploaded by Daniel Wayne Russell on 28 January 2015.

The user has requested enhancement of the downloaded file.

 The Journal of Psychology, 2012, 146(1–2), 7–22
Copyright

C 2012 Taylor & Francis Group, LLC

Is Loneliness the Same as Being Alone?

DANIEL W. RUSSELL
CAROLYN E. CUTRONA
Iowa State University

CYNTHIA MCRAE
University of Denver

MARY GOMEZ
Cherry Creek School District

ABSTRACT. The cognitive discrepancy model predicts that loneliness occurs when indi-
viduals perceive a difference between their desired and actual levels of social involvement.
Using data from a sample of high school sophomore students, the present investigation
was designed to go beyond previous research that has tested this model by examining
the predicted nonlinear relationships between desired and actual social contact and feel-
ings of loneliness. Analyses indicated that support for the cognitive discrepancy model of
loneliness was found only for measures of close friendships. Specifically, the discrepancy
between the students’ ideal number and actual number of close friends was found to be
related in a nonlinear fashion to feelings of satisfaction with close friendships and loneliness
after control for the number of close friends. Implications of these findings for theoretical
models of loneliness are discussed.
Keywords: loneliness, social activity, cognitive discrepancy, comparison level

LONELINESS RESEARCHERS commonly distinguish between the experience

of loneliness and being alone (Cutrona, 1982; Paloutzian & Janigan, 1987; Perlman
& Peplau, 1982; Russell, 1982). This reflects the recognition that some individuals
may be alone or socially isolated and be quite happy with that situation. In contrast,
other individuals may be involved in a large number of interpersonal relationships
yet be dissatisfied with important aspects of their relationships (e.g., the quality
of the relationships; lack of a particular type of relationship, such as a romantic
relationship) and experience feelings of loneliness.

Address correspondence to Daniel W. Russell, Department of Human Development and

Family Studies, 4380 Palmer Building, Iowa State University, Ames, IA 50011-4380, USA;
[email protected] (e-mail).

7
8 The Journal of Psychology

To understand the distinction between social isolation and loneliness, the cog-
nitive discrepancy model of loneliness was developed (Peplau & Perlman, 1982).
According to this model, individuals develop an internal standard or expectation
against which they judge their interpersonal relationships. If their relationships
with others exceed this standard, then individuals are satisfied with their relation-
ships and do not experience feelings of loneliness. If their current relationships
with others are below this standard, then individuals are dissatisfied with their
relationships and experience feelings of loneliness.
The cognitive discrepancy model of loneliness represents an extension of
earlier theoretical ideas developed by Thibaut and Kelley (1959). They present
an analysis of satisfaction and attraction in dyadic relationships based on the
individual’s Comparison Level (CL). If the outcomes experienced by the individual
from a given relationship are above the CL, then the individual is satisfied and
attracted to the relationship. If the outcomes experienced by the individual for that
relationship are below the CL, then the individual is dissatisfied and not attracted
to the relationship. The cognitive discrepancy model of loneliness proposes that
the individual develops a CL for his or her entire network of social relationships.
This CL can be thought of as representing the quantity or quality of social contact
the person desires and is used by the individual to evaluate the adequacy of his or
her current social network. Thus, the cognitive discrepancy model hypothesizes
that satisfaction with social relationships and feelings of loneliness are jointly
determined by the person’s current social relationships and his or her CL for
social relationships.
One factor hypothesized by Thibaut and Kelley (1959) to determine the
person’s CL is social comparison. That is, the individual formulates expectations
concerning the number and type of social relationships he or she should have in
part based on the social relationships of similar others. So, for example, students
should develop expectations regarding their interpersonal relationships based in
part on the relationships of other students who are perceived as being similar to
themselves.
An interesting issue concerns how discrepancies between the person’s actual
interpersonal relationships and desired or expected relationships are related to
satisfaction with relationships and loneliness. Kelley and Thibaut (1978) suggest
that this relationship may not be linear. That is, near the person’s CL a unit increase
or decrease in the quantity or quality of relationships may be especially important
to the person in determining satisfaction and loneliness. Thus, there may not be
a simple linear relationship between changes in the discrepancy between current
relationships and the CL and feelings of satisfaction and loneliness.
To understand this prediction, imagine that a student desires four close friends.
If this student has four close friends and loses one, that loss may affect greatly his
or her satisfaction with friendships and loneliness. What if the same student has
eight close friends and loses one, or has two close friends and loses one? Clearly,
in these latter cases the person is so far above or below the CL that a loss of one
 Russell, Cutrona, McRae, & Gomez 9

friend has very little meaning in terms of changes in relationship satisfaction or

loneliness. Thus, the units of satisfaction or loneliness may be “stretched” in the
area surrounding the CL.
The cognitive discrepancy model of loneliness was tested by Russell, Steffen,
and Salih (1981). Using data from a randomly selected sample of over 500 college
students (including nearly equal numbers of the four undergraduate classes and
graduate students), measures were taken of students’ current social relationships,
CL for social relationships, typical social relationships of college students, and
satisfaction with social relationships in three relationship domains: friendships,
romantic/dating relationships, and family relationships. In addition, participant’s
level of loneliness was measured using the UCLA Loneliness Scale (Russell,
Peplau, & Cutrona, 1980). Analyses were conducted examining the relationship
between a discrepancy score, formed by subtracting the person’s current relation-
ships from his or her CL, and satisfaction with each type of relationship.
As predicted by Kelley and Thibaut (1978), nonlinear associations were found
between this discrepancy score and relationship satisfaction and loneliness for
all three types of relationships. For example, Russell, Steffen, and Salih (1981)
found that the relationship between increases in the number of close friendships
and satisfaction with friendships and decreases in loneliness was linear until the
point where the person’s current friendships and his or her CL for friendships
were equal. Above that point, there was no further increase in satisfaction or
decrease in loneliness with further increases in the number of close friendships
above the CL. These results were also found for measures of the quality of
relationships. For example, the association between increases in the intimacy of
romantic relationships and satisfaction with romantic relationships and decreases
in loneliness was linear until the point where the person’s current relationship
intimacy and his or her CL for intimacy were equal; there were no further increase
in satisfaction or decrease in loneliness with further increases in intimacy. These
results indicate that, as suggested by Kelley and Thibaut (1978), individuals are
differentially sensitive to increases or decreases in the quantity or quality of
their interpersonal relationships, depending upon how their current relationships
compare to the CL. Below the CL, satisfaction and loneliness are determined by
the number of close friends and the level of romantic relationship intimacy. Above
the CL, further increases in the number of close friendships or the intimacy of the
romantic relationship had no effect on relationship satisfaction or loneliness.
Archibald, Bartholomew, and Marx (1995) attempted to test the cognitive
discrepancy model of loneliness in a sample of high school sophomores. They
assessed the current social relationships of these students, their ideal social rela-
tionships (i.e., the CL), and their perceptions of the relationships of the typical
high school student. Feelings of satisfaction with social relationships and lone-
liness were also assessed. After controlling for the current social relationships
of these students, Archibald et al. (1995) found that the discrepancy between
their ideal and actual social relationships was strongly related to relationship
10 The Journal of Psychology

satisfaction (
R2 = .14) but was unrelated to feelings of loneliness. By contrast,

the discrepancy between the typical student’s social relations and their own so-
cial relationships was only weakly predictive of either relationship satisfaction or
loneliness (
R2 = .02).
Ward and Rana-Deuba (2000) also examined the effects of actual social
contact versus desired social contact on satisfaction with social relationships and
loneliness for a sample of foreigners living in Nepal (M Age = 39.6 years).
They assessed the amount of actual social contact and desired social contact in
12 areas (e.g., going to social functions, discussing social and political issues,
celebrating holidays) for relationships with Nepalese and individuals from their
own country. Their results indicated that the discrepancy between their actual and
desired social contacts across these 12 areas was related to satisfaction with the
quality and quantity of relationships with Nepalese and conationals (r = .35–.48).
The measures of relationship satisfaction were also significantly related to feelings
of loneliness. However, as Archibald et al. (1995) found, the discrepancy measures
were unrelated to loneliness.
Unfortunately, these investigations did not test for the nonlinear relationships
between their discrepancy measures and feelings of satisfaction and loneliness
predicted by Kelley and Thibaut (1978) and found by Russell et al. (1981). As
a consequence, it is unclear whether or not their results are consistent with the
predictions of the cognitive discrepancy model of loneliness. The present study
collected data from a sample of high school sophomores using the measures of
social relationships employed by Archibald et al. (1995). In addition, data were
collected using the measures of close friendships employed by Russell et al.
(1981). Using these data, analyses were conducted to test for both linear and
nonlinear relationships between the cognitive discrepancy measures and feelings
of satisfaction with relationships and loneliness.

Method

Participants
A sample of 188 sophomore students from a suburban high school participated
in the study. In recruiting participants, teachers were first contacted to determine
if they would permit students in their class to participate during class time. A total
of 195 students, representing approximately 45% of the 435 sophomores in the
school, were allowed to participate in the study. The final sample represented 43%
of the sophomore class, or 96% of those who were permitted by their teachers to
participate.
Participants were 85 males and 95 females. The remaining eight students did
not indicate their sex. The students were 15–16 years of age. A majority of the
sample (57%) was white, 14% were African American, 7% were Hispanic, and
6% were Asian. Remaining participants were either of mixed racial background
(9%) or in another racial group (7%).
 Russell, Cutrona, McRae, & Gomez 11

List-wise deletion of cases with missing data on the measures was used to
ensure that the results were consistent for the different sets of analyses that were
conducted. As a consequence, the number of cases used in the analyses was
169. Individuals with missing data on one or more of the variables (n = 19)
were compared to cases with complete data (n = 169) in terms of demographic
characteristics and the measures of interpersonal relationships and loneliness.
Perhaps not surprisingly, male students (12.9%) were more likely to have missing
data on these measures than female students (2.1%), χ (1, N = 180) = 6.33, p <
.05. None of the differences between these two groups on the remaining measures
were statistically significant (p < .05).

Measures
Social Life Questionnaire. This measure was developed by Archibald et al.
(1995) to assess various aspects of the social lives of high school students. It
consists of 78 items that ask students about their actual and ideal levels of social
activity as well as their perceptions of the social activities engaged in by the
typical high school student. The measure covers activities with social friends,
close friends, romantic relationships, and specific social events. For example, one
item asks participants, “How often before classes start: (a) do you get together
with friends? (b) would you like to get together with friends? and (c) does the
typical student get together with friends?” Students responded to these items
using a 9-point Likert scale that ranged from “every school day” to “almost
never.”
Scores were computed following the procedures used by Archibald et al.
(1995). A measure of actual social activity was derived by standardizing and
summing together the 26 questions that dealt with student involvement in social
activities. This measure was found to be highly reliable (α = .95). Ideal-actual
discrepancy was derived by subtracting students’ ratings of their actual social
activities from their ideal social activities, standardizing these scores, and then
summing them. This measure was also found to be reliable (α = .94). An identical
procedure was used to derive scores for the typical-actual discrepancy measure,
which was also found to be reliable (α = .95).
A final set of seven questions asked the students to rate how satisfied they
were with various aspects of their interpersonal relationships on a 7-point Likert
scale that ranged from “completely satisfied” to “completely dissatisfied.” So, for
example, students were asked, “How satisfied are you with the amount of weekday
social activity you engage in?” Responses to these items were standardized and
summed together to form a measure of social satisfaction (α = .90).
Close friendships. A series of questions identical to those used by Russell
et al. (1981) was asked concerning close friendships. First, students were asked to
indicate their current number of close friends on a 7-point scale that ranged from
“none” to “11 or more.” They were also asked to indicate their ideal number of
close friends and how many close friends they thought the typical sophomore at
12 The Journal of Psychology

their high school has, using the same 7-point scale. Finally, they were asked to
indicate how satisfied they were with their current number of close friends on a
9-point scale that ranged from “not at all” to “very satisfied.”
To be consistent with the measures computed from the Social Life Question-
naire, measures of ideal-actual discrepancy and typical-actual discrepancy were
also calculated from the measures of close friendships.
Loneliness. Loneliness was assessed using Version 3 of the UCLA Loneliness
Scale (Russell, 1996). This latest version of the Loneliness Scale was designed to
improve on earlier versions of the scale (e.g., Russell et al., 1980) by simplifying
both the wording of the questions and the response format. Data presented by
Russell (1996) provide support for the reliability and validity of the loneliness
scale for a variety of populations. The measure was found to be reliable in this
sample (α = .90).

Results

Descriptive Statistics
The average loneliness score for this sample was 40.57 (SD = 9.16). This
is comparable to the average scores on Version 3 of the UCLA Loneliness Scale
for college student samples (Russell, 1996). Male students (M = 41.86) reported
higher levels of loneliness than female students (M = 39.91), although the differ-
ence was not statistically significant, t(168) = 1.38. It should be noted that most
studies have not found sex differences in loneliness using the UCLA Loneliness
Scale (see review by Borys & Perlman, 1985).
Students in the present study reported a higher level of ideal social activity
than their actual level of social activity, t(168) = 8.90, p < .001. Similarly, students
also reported a higher number of ideal close friendships than their actual number
of close friends, t(168) = 5.44, p < .001. They rated the typical student’s level of
social activity as higher than their own, t(168) = 6.63, p < .001. However, their
rating of the typical student’s number of close friends did not differ from their own
actual number of close friends, t(168) < 1. There was also no difference in their
rating of the ideal and typical level of social activity, t(168) < 1. Finally, students
rated their ideal number of close friends as higher than the number of close friends
of the typical high school sophomore, t(168) = 3.86, p < .001.
Correlations among the measures of loneliness, satisfaction, and social activity
are presented in Table 1, along with the parallel measures for close friendships.
The measures derived from the Social Life Questionnaire (Archibald et al., 1995)
were all very highly correlated. By contrast, the measures of close friendships
were less strongly related to one another. The two sets of measures were nearly
equally related to scores on the loneliness measure. Finally, the measures of social
activity and the two discrepancy scales were all correlated above .40 in magnitude
with the satisfaction measure derived from the Social Life Questionnaire.
TABLE 1. Correlations Among the Measures of Loneliness, Satisfaction, Social Activity, and Close Friendships

Social Ideal-Actual Typ.-Actual # of Ideal-Actual Typ.-Actual Satisfaction

Variable Activity Discrepancy Discrepancy Satisfaction Friends Friends Friends w/ Friends

Social Activity 1.00

Ideal-Actual Discrepancy –.71∗∗∗ 1.00
Typ.-Actual Discrepancy –.85∗∗∗ .81∗∗∗ 1.00
Satisfaction .46∗∗∗ –.44∗∗∗ –.44∗∗∗ 1.00
# of Friends .31∗∗∗ –.26∗∗ –.31∗∗∗ .22∗∗ 1.00
Ideal-Actual Friends –.23∗∗ .36∗∗∗ .25∗∗ –.22∗∗ –.26∗∗ 1.00
Typ.-Actual Friends –.33∗∗∗ .29∗∗∗ .41∗∗∗ –.19∗ –.61∗∗∗ .28∗∗∗ 1.00
Satisfaction w/ Friends .34∗∗∗ –.17∗ –.28∗∗∗ .40∗∗∗ .25∗∗ –.13 –.29∗∗∗ 1.00
Loneliness –.35∗∗∗ .23∗∗ .35∗∗∗ –.44∗∗∗ –.31∗∗∗ .23∗∗ .43∗∗∗ –.49∗∗∗

Note. N = 169. Typ. = Typical.

∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.
Russell, Cutrona, McRae, & Gomez
13
14 The Journal of Psychology

Prediction of Satisfaction
Analyses were conducted to evaluate the ability of the cognitive discrepancy
measures to predict the two measures of satisfaction (i.e., satisfaction with social
activities and satisfaction with close friendships) over and above the influence
of actual social relationships. A series of hierarchical regression analyses was
conducted. In these analyses, sex of student was included as a predictor in Step 1 of
the regression analyses. The measure of actual relationships was entered in Step 2,
followed by the linear discrepancy measure (e.g., ideal-actual relationships) in
Step 3 and the square of the discrepancy measure in Step 4. Inclusion of this
latter term permitted us to test for the nonlinear relationships of the discrepancy
measures predicted by Kelley and Thibaut (1979) and found by Russell et al.
(1981).
Table 2 presents the results of the regression analysis predicting satisfaction
with social activities. Sex of the student was not related to satisfaction, whereas
the level of actual social activity was a highly significant predictor (R2 = .24).
The ideal-actual discrepancy score was also a statistically significant predictor,
accounting for an additional 3% of the variance in the satisfaction measure. The
typical-actual discrepancy score accounted for only 1% of the variance in satisfac-
tion. Finally, in contrast to our predictions, there was no evidence of a nonlinear
relationship between the discrepancy measures and satisfaction, with neither of
the squared discrepancy measures found to be statistically significant.
Hierarchical regression results for the measure of satisfaction with close
friendships are presented in Table 3. Sex of student was not significantly related
to satisfaction with close friendships, whereas the number of close friends was
found to be a statistically significant predictor (R2 = .06). After control for these
two predictor variables, the ideal-actual discrepancy measure was not related to
satisfaction. However, as predicted, the squared discrepancy measure was found

TABLE 2. Hierarchical Regression Analyses Predicting Satisfaction with Social

Activities

Step
R2 F

1. Sex .01 2.13

2. Social Activity .24 53.51∗∗∗
3. Ideal-Actual Discrepancy .03 5.57∗
4. Ideal-Actual Discrepancy Squared .01 1.50
3. Typical-Actual Discrepancy .01 2.42
4. Typical-Actual Discrepancy Squared .00 <1

Note. N = 169.
∗∗∗ p < .001. ∗∗ p < .01. ∗ p < .05.
 Russell, Cutrona, McRae, & Gomez 15

TABLE 3. Hierarchical Regression Analyses Predicting Satisfaction with Close

Friendships

Step
R2 F

1. Sex .00 <1

2. # of Close Friends .06 10.89∗∗
3. Ideal-Actual Discrepancy .01 1.06
4. Ideal-Actual Discrepancy Squared .03 6.20∗
3. Typical-Actual Discrepancy .03 6.22∗
4. Typical-Actual Discrepancy Squared .00 <1

Note. N = 169.
∗∗∗ p < .001. ∗∗ p < .01. ∗ p < .05.

to be a statistically significant predictor, accounting for an additional 3% of the

variance in satisfaction with close friendships. For the typical-actual discrepancy
measure, only the linear term was found to be a statistically significant predictor
after control for sex and the number of close friends, accounting for an additional
3% of the variance in satisfaction.
These latter results therefore appear to be consistent with the findings of Rus-
sell et al. (1981) concerning the nonlinear relationship between the ideal-actual
discrepancy measure and satisfaction with close friendships. Figure 1 presents a
plot of the relationship between the ideal-actual discrepancy score and a standard-
ized residual satisfaction score, wherein the effects of sex and the number of close
friends have been removed statistically. Satisfaction with close friends increased
as the actual number of close friends became closer to the student’s ideal number
of close friends. However, in contrast to predictions, the level of satisfaction was
found to decrease as the actual number of close friends became greater than the
ideal number of close friends. Satisfaction with close friends was found to be the
greatest when there was a match between the ideal and actual numbers of close
friends.

Prediction of Loneliness
An identical set of regression analyses was conducted to predict loneliness
from the measures of social activity and close friendships. Table 4 presents the
results of the analyses predicting levels of loneliness from the measures of social
activity. After control for sex of the student, actual level of social activity ac-
counted for 11% of the variance in loneliness scores. The ideal-actual discrepancy
measure was not a statistically significant predictor, whereas the typical-actual
discrepancy measure was marginally significant, accounting for 2% of the vari-
ance in loneliness. In contrast to predictions, neither of the squared discrepancy
terms was a statistically significant predictor of loneliness.
16 The Journal of Psychology

FIGURE 1. Ideal–actual close friends and standardized level of residual satis-

faction with close friends.

TABLE 4. Hierarchical Regression Analyses Predicting Loneliness from Mea-

sures of Social Activity

Step
R2 F

1. Sex .01 1.91

2. Social Activity .11 19.47∗∗∗
3. Ideal-Actual Discrepancy .00 <1
4. Ideal-Actual Discrepancy Squared .01 1.21
3. Typical-Actual Discrepancy .02 3.52
4. Typical-Actual Discrepancy Squared .00 <1

Note. N = 169.
∗∗∗ p < .001. ∗∗ p < .01. ∗ p < .05.
 Russell, Cutrona, McRae, & Gomez 17

TABLE 5. Hierarchical Regression Analyses Predicting Loneliness from Mea-

sures of Close Friendships

Step
R2 F

1. Sex .01 1.91

2. # of Close Friends .09 16.99∗∗∗
3. Ideal-Actual Discrepancy .03 5.75∗
4. Ideal-Actual Discrepancy Squared .07 13.69∗∗∗
3. Typical-Actual Discrepancy .08 15.83∗∗∗
4. Typical-Actual Discrepancy Squared .01 1.16

Note. N = 169.
∗∗∗ p < .001. ∗∗ p < .01. ∗ p < .05.

Table 5 presents parallel results using the measures of close friendships to

predict loneliness scores. The actual number of close friends was found to predict
9% of the variance in loneliness scores. After control for the number of close
friends, the ideal-actual discrepancy measure for close friends was found to be
statistically significant, accounting for 3% of the variance. Consistent with predic-
tions, the squared discrepancy term was also statistically significant, accounting
for an additional 7% of the variance in loneliness scores. Finally, as was found for
the measure of satisfaction with friendships, only the linear term for the typical-
actual discrepancy measure was statistically significant, accounting for 8% of the
variance in loneliness scores.
The relation between the ideal-actual discrepancy score and the standardized
residual measure of loneliness (with control for the effects of sex and number of
close friends) is presented in Figure 2. As expected, the results are essentially a
mirror image of the findings for satisfaction with close friendships. Loneliness
decreased as the actual number of close friends approached the ideal number of
close friends reported by the students. However, loneliness increased as the actual
number of close friends exceeded the ideal number of close friends. Once again,
the lowest levels of loneliness were reported by students for whom there was a
match between their ideal and actual numbers of close friends.

Discussion

The current study tested the nonlinear association between the ideal-actual
discrepancy and relationship outcomes such as loneliness. According to Kelley and
Thibaut (1978), discrepancies around the ideal level or CL are most important to
the individual, having the greatest effect on relationship satisfaction and loneliness.
Discrepancies that are farther away from the ideal level or CL are hypothesized to
18 The Journal of Psychology

FIGURE 2. Ideal–actual close friends and standardized level of residual

loneliness.

have less effect on satisfaction and loneliness, leading to the hypothesized nonlin-
ear relationship. Recent studies testing for the relationship between the difference
between ideal and actual relationships and satisfaction and loneliness have not
evaluated this nonlinear relationship. The present study therefore builds on the
analyses presented by Archibald et al. (1995) and Ward and Rana-Deuba (2000)
by adding a test for nonlinear association between the ideal-actual relationship
difference and relationship satisfaction and loneliness.
The results of this investigation provide mixed support for this cognitive
discrepancy model of loneliness. There was no evidence of the predicted nonlinear
relationship between the ideal-actual discrepancy measure for social activities and
either satisfaction with social activities or loneliness. By contrast, a statistically
significant nonlinear relationship was found between the ideal-actual discrepancy
measure for close friendships and both satisfaction with close friendships and
loneliness. This latter relationship was found after control for the number of close
friendships reported by the students.
What might account for these differences in results for social activities versus
close friendships? One possible explanation concerns the nature of the measures
 Russell, Cutrona, McRae, & Gomez 19

themselves. An assumption of the cognitive discrepancy model is that individuals

have developed a clear expectation concerning the quantity and/or quality of their
interpersonal relationships, which is in turn employed by individuals in evaluating
those relationships. Given the importance and salience of close friendships, it is
easy to believe that high school sophomores have a clear idea concerning the
number of close friendships they would ideally like to have. By contrast, these
students may not have a clear conception of how often they would like to participate
in various social activities (e.g., getting together with friends before classes start).
That is, these students may have a CL for close friendships but not for social
activities.
A second possible explanation involves colinearity between the measure of
actual social activity and the discrepancy measures derived from the Social Life
Questionnaire. As can be seen in Table 1, the two discrepancy measures were
strongly related to the measure of actual social activity; these correlations are
very similar to those reported by Archibald et al. (1995). Furthermore, scores on
the discrepancy measures were only weakly related to the measures of ideal or
typical social activities. So, for example, the correlation between the ideal-actual
discrepancy score and the measure of actual social activities was highly significant
(r = –.71); the fewer was the number of social activities reported, the greater was
the discrepancy. By contrast, the correlation of the ideal-actual discrepancy score
with the measure of ideal social activities was nonsignificant (r = .06). This
indicates that scores on the discrepancy measures were primarily a function of the
actual level of social activity reported by the students and in turn may explain why
the ideal-actual discrepancy measure was not predictive of either satisfaction with
social activity or loneliness after control for the level of actual social activity.
The parallel measures concerning close friends were less strongly related to
the number of close friends. For example, the correlation between the number of
close friends and the ideal-actual discrepancy score was –.26 (see Table 1). In
addition, the ideal number of close friends was much more strongly related to this
discrepancy score (r = .48), suggesting that the scores on the discrepancy measure
were not solely determined by the actual number of close friends reported by the
students.
Although we did find evidence supporting a nonlinear relationship between
the ideal-actual discrepancy measure for close friendships and both satisfaction
with close friendships and loneliness, the form of that relationship was clearly
different from that found by Russell et al. (1981) for college students. Among
high school sophomores, having both more and fewer close friends than desired
was associated with decreased satisfaction with close friendships and increased
loneliness. By contrast, Russell et al. (1981) found that decreased satisfaction and
increased loneliness occurred only if college students reported having fewer close
friends than they desired.
An important issue for future research concerns why these high school stu-
dents would find having more friends (relative to their CL) to be negative or
20 The Journal of Psychology

aversive. It is possible that having too many close friends is burdensome to these
students and overloads them with social obligations. Indeed, we may have caught
some of these students in the midst of a process wherein they are discovering
how many friends they desire or formulating their CL for close friendships. So,
for example, some students may have formed a large number of close friendships
early in their high school career, only to discover the negative aspects of being
involved in a large number of relationships. Indeed, these high school students
reported a greater number of close friends (M = 5.26) than did the college stu-
dents surveyed by Russell et al. (1981; M = 3.86). Over time, students may adjust
their CL downward and reduce the size of their social network. If true, we should
expect to see declines in the number of close friends over time among those stu-
dents who indicated that their actual number exceeded their ideal number of close
friends.
In contrast to the nonlinear relationship found for the ideal-actual discrepancy
measure, the typical-actual discrepancy measure was related to satisfaction and
loneliness in a linear fashion. High school students reported greater satisfaction
with friendships and less loneliness to the extent that their close friendships ex-
ceeded those of the typical high school student. According to Kelley and Thibaut
(1978), such social comparison information should serve as one determinant of
the person’s CL. Support for this prediction was found for the present sample.
The ideal number of close friends was correlated with the typical number of
close friends (r = .34). However, these two measures clearly operate differently
in relation to the actual number of close friends reported by students, given the
different pattern of relationships with satisfaction and loneliness. The discrepancy
between typical and actual close friendships remained a significant predictor of
both satisfaction and loneliness after control for the discrepancy between ideal
and actual close friendships. That is, in a regression model that included the ideal-
actual score as a predictor (and that controlled for both sex and actual number
of friendships), the typical-actual score was predictive of both satisfaction with
close friendships (β = –.23, p < .05) and loneliness (β = .33, p < .001). This
suggests that students may compare their current friendships to both their CL and
their perceptions of the relationships of other similar students in evaluating the
adequacy of those relationships.
An important limitation of the measure of social activities that was employed
in this study should be noted.1 The Social Life Questionnaire does not involve
contact with others using different forms of technology, such as the Internet or cell
phones, that are commonly employed by high school students. These new methods
of social contact may be particularly important for lonely students. Although
studies indicate that loneliness is not related to the formation of friendships over
the Internet (Russell, Flom, Gardner, Cutrona, & Hessling, 2003; Subrahmanyam
& Lin, 2007), loneliness among high school students has been found to be related to
compulsive use of the Internet (Van der Aa, Overbeek, Engels, Schotte, Meerkerk,
& Fejnden, 2009). Incorporating an assessment of social contact via the Internet
 Russell, Cutrona, McRae, & Gomez 21

and other forms of technology may have altered the relationship between loneliness
and the measure of social activity.
In summary, our results indicate that loneliness is not synonymous with being
alone, at least in the context of certain aspects of interpersonal relationships.
High school students who are lonely do not necessarily report having fewer close
friendships. Instead, it appears from our findings that in predicting loneliness,
we must take into consideration the number of close friendships desired by these
students. Individuals who report a match between their desired and actual numbers
of close friends are both the individuals who are most satisfied with their close
friendships and the individuals who are least lonely. In contrast, students who
report having more or fewer close friendships than they desire are the most likely
to be dissatisfied with their friendships and to be lonely. Future studies should
examine why these high school students found it aversive to have more friends
than they desired and should evaluate changes over time in both their CL for
friendships and their actual number of friendships. Measures of social contact
via the Internet and other technologies should also be incorporated into future
examinations of the relationship between loneliness and social contact.

NOTE
1. The authors thank an anonymous reviewer for noting this limitation of the measure.

AUTHOR NOTES
Daniel W. Russell is a Professor in the Department of Human Development and Family
Studies, Iowa State University. Carolyn E. Cutrona is a Professor in the Department
of Psychology, Iowa State University. Cynthia McRae is a Professor in the College of
Education, University of Denver. Mary Gomez is a Teacher in the Cherry Creek School
District, Englewood, Colorado.

REFERENCES
Archibald, F. S., Bartholomew, K., & Marx, R. (1995). Loneliness in early adolescence: A
test of the cognitive discrepancy model of loneliness. Personality and Social Psychology
Bulletin, 21, 296–301.
Borys, S., & Perlman, D. (1985). Gender differences in loneliness. Personality and Social
Psychology Bulletin, 11, 61–74.
Cutrona, C. E. (1982). Transition to college: Loneliness and the process of social adjustment.
In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theory,
research, and therapy (pp. 291–301). New York: Wiley Interscience.
Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory of interdependence.
New York: Wiley Interscience.
Paloutzian, R. F., & Janigan, A. S. (1987). Models and methods in loneliness research:
Their status and direction. Journal of Social Behavior and Personality, 2, 31–36.
Perlman, D., & Peplau, L. A. (1982). Theoretical approaches to loneliness. In L. A. Peplau
& D. Perlman (Eds.), Loneliness: A sourcebook of current theory, research, and therapy
(pp. 123–134). New York: Wiley Interscience.
 22 The Journal of Psychology

Russell, D. (1982). The measurement of loneliness. In L. A. Peplau & D. Perlman (Eds.),

Loneliness: A sourcebook of current theory, research, and therapy (pp. 81–104). New
York: Wiley Interscience.
Russell, D. (1996). The UCLA Loneliness Scale (Version 3): Reliability, validity, and factor
structure. Journal of Personality Assessment, 66, 20–40.
Russell, D. W., Flom, E. K., Gardner, K. A., Cutrona, C. E., & Hessling, R. S. (2003). Who
makes friends over the Internet? Loneliness and the “virtual” community. International
Scope Review, 5(10), 1–19.
Russell, D., Peplau, L.A., & Cutrona, C. E. (1980). The revised UCLA Loneliness Scale:
Concurrent and discriminant validity evidence. Journal of Personality and Social Psy-
chology, 39, 472–480.
Russell, D., Steffen, M., & Salih, F. A. (1981, August). Testing a cognitive model of
loneliness. Paper presented at the symposium New Directions in Loneliness Research
at the annual convention of the American Psychological Association in Los Angeles,
California.
Subrahmanyam, K., & Lin, G. Adolescents on the net: Internet use and well-being. Ado-
lescence, 42, 659–677.
Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: Wiley.
Van der Aa, N., Overbeek, G., Engles, R. C. M. E., Scholte, R. H. J., Meerkerk, G., & Van
den Eijnden, R. J. J. M. (2009). Daily and compulsive Internet use and well-being in
adolescence: A diathesis-stress model based on Big Five personality traits. Journal of
Youth and Adolescence, 38, 765–776.
Ward, C., & Rana-Duba, A. (2000). Home and host culture influences on sojourner adjust-
ment. International Journal of Intercultural Relations, 24, 291–306.

Original manuscript received February 28, 2011

Final version accepted May 10, 2011

View publication stats