Evaluating voluntary aloneness in childhood: Initial validation of the Children's Solitude Scale

Abstract

This work intends to psychometrically evaluate the newly developed Children's Solitude Scale (CSS), a measure of voluntary aloneness in childhood. The scale assesses individual differences on what children prefer to do when they want to be alone, a rather neglected, although important developmental issue. Participants were 833 fourth and sixth graders from Athens, Greece. Confirmatory factor analysis indicated four factors, although with a less-than-optimal fit: Self-Reflection, Autonomy/Privacy, Activities and Concentration. The scale showed adequate internal consistency and test-retest reliability as well as convergent and divergent validity (i.e., positive association with positive attitude towards aloneness and ability to be alone, no association with negative attitude towards aloneness, and low positive correlation with loneliness and social dissatisfaction). The results indicated the suitability of the CSS to assess reasons or motives for solitude during middle and late childhood and the necessity to further examine the structure of this type of solitude experiences in this age period.

Full text

Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=pedp20
Download by: [T&F Internal Users], [Rozanna Harrison] Date: 11 November 2015, At: 02:19
European Journal of Developmental Psychology
ISSN: 1740-5629 (Print) 1740-5610 (Online) Journal homepage: http://www.tandfonline.com/loi/pedp20
Evaluating voluntary aloneness in childhood:
Initial validation of the Children's Solitude Scale
Evangelia P. Galanaki, Kostas Mylonas & Panagiota S. Vogiatzoglou
To cite this article: Evangelia P. Galanaki, Kostas Mylonas & Panagiota S. Vogiatzoglou (2015):
Evaluating voluntary aloneness in childhood: Initial validation of the Children's Solitude Scale,
European Journal of Developmental Psychology, DOI: 10.1080/17405629.2015.1071253
To link to this article: http://dx.doi.org/10.1080/17405629.2015.1071253
Published online: 10 Nov 2015.
Submit your article to this journal
View related articles
View Crossmark data

Evaluating voluntary aloneness in childhood: Initial
validation of the Children’s Solitude Scale
Evangelia P. Galanaki
1
, Kostas Mylonas
2
, and
Panagiota S. Vogiatzoglou
1
1
Faculty of Primary Education, School of Education, National and Kapodistrian
University of Athens, Navarinou 13A, 10680 Athens, Greece
2
Faculty of Psychology, School of Philosophy, National and Kapodistrian
University of Athens, Panepistimiopolis, 15784 Athens, Greece
This work intends to psychometrically evaluate the newly developed Children’s
Solitude Scale (CSS), a measure of voluntary aloneness in childhood. The scale assesses
individual differences on what children prefer to do when they want to be alone, a rather
neglected, although important developmental issue. Participants were 833 fourth and
sixth graders from Athens, Greece. Confirmatory factor analysis indicated four factors,
although with a less-than-optimal fit: Self-Reflection, Autonomy/Privacy, Activities
and Concentration. The scale showed adequate internal consistency and test-retest
reliability as well as convergent and divergent validity (i.e., positive association with
positive attitude towards aloneness and ability to be alone, no association with negative
attitude towards aloneness, and low positive correlation with loneliness and social
dissatisfaction). The results indicated the suitability of the CSS to assess reasons or
motives for solitude during middle and late childhood and the necessity to further
examine the structure of this type of solitude experiences in this age period.
Keywords: Solitude; Aloneness; Loneliness; Social Dissatisfaction; Attitude
towards Aloneness; Children’s Solitude Scale; Middle Childhood; Late Childhood,
Confirmatory Factor Analysis.
Solitude is defined as either voluntary or involuntary aloneness, with either
positive or negative functions, occurring usually in the absence of others
(Coplan & Bowker, 2014; Long & Averill, 2003; Storr, 1988). Actively
striving for time spent on one’s own and making constructive use of such time
q 2015 Taylor & Francis
Email: egalanaki@primedu.uoa.gr
The authors are grateful to Professor Robert Coplan for his valuable help in forming the English
version of the Children’s Solitude Scale.
No potential conflict of interest was reported by the authors.
EUROPEAN JOURNAL OF DEVELOPMENTAL PSYCHOLOGY, 2015
http://dx.doi.org/10.1080/17405629.2015.1071253
Developmetrics
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

has been referred to as the capacity to be alone. This capacity is often regarded
as an important developmental achievement and, indeed, a necessary
precondition for the development of a mature personality (Larson, 1997;
Winnicott, 1965). Voluntary aloneness, thus defined, should be clearly
distinguished from loneliness, which refers to the painful experience that stems
from perceived deficits in intimacy and/or belonging (Peplau & Perlman,
1982). The two concepts—voluntary aloneness and loneliness— are different
from one another, because people who regularly want to be alone may or may
not feel lonely.
Despite the potential beneficial role of solitude in development, there is a lack
of research on what children prefer to do (i.e., what kind of activities they engage
in) when they want to be alone. Extant research, using measures such as the
Loneliness and Aloneness Scale for Children and Adoles cents (Marcoen &
Goossens, 1993) or the Ability to Be Alone Questionnaire (Youngblade, Berlin,
& Belsky, 1999), assesses the negative and positive attitude towards aloneness.
These instruments measure affinity for aloneness (e.g., “I enjoy being on my
own”) or aversion to aloneness (e.g., “When I am alone, I feel bored”), but the
items are vague about the proposed functions of time spent alone by oneself or
the activities displayed when being alone. Furthermore, whereas recent research
(e.g., Goossens et al., 2009; Maes, Klimstra, Van den Noortgate, & Goossens,
2015; Teppers, Luyckx, Vanhalst, Klimstra, & Goossens, 2014 ) focuses on
aloneness experiences in adolescence, much less research has been conducted on
children’s solitude.
Based on the presumed positive functions of time spent alone in adults
(Long & Averill, 2003), we can expect to find four such functions in children
that can each be associated with a specific type of activity. More specifica lly,
it is hypothesized that children want to be alone in order (a) to engage in self-
reflection, (b) to protect their privacy, (c) to indulge in their favourite solitary
pastimes (i.e., hobbies), and (d) to concentrate on important tasks in their lives
(e.g., homework). Therefore, the aim of this research was to develop a self-
report measure, the Children’s Solitude Scale (CSS), which assesses
individual difference s in the uses of voluntary solitude, and to evaluate its
reliability, factorial structure and convergent and divergent validity.
We expected to find four factors that refer to self-reflection, privacy, hobbies
and increased concentration, respective ly. The four subscales that inquire into
the frequency of the activities associated with these functions were expected
to show a sufficient degree of reliability. Regarding convergent and divergent
validity, we expected the four subscales to show significant positive
associations with measures that tap into the positive attitude to being alone,
significant negative associations with measures that tap into a negative attitude
to being alone, and weak or null associations with a measure of loneliness.
2
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

METHOD
Participants and procedure
Participants were 833 fourth and sixth graders (51.2% boys; 50.3% fourth
graders). Children were 9 –10 and 11 –12 years old (we were not granted access
to children’s birth dates). They came from 43 classes of 13 randomly selected
public primary schools located in Athens, Gr eece. The areas represented middle
and lower-middle socioeconomic status. About 15% of students were of non-
Greek descent, but born in Greece. According to teachers’ reports, all students
could comprehend, read and write Greek adequately. All fourth and sixth grade
classes of these schools participated. Written parental consent was obtained for
all students. The initial sample consisted of 857 children, but 24 of them (i.e., 16
fourth graders and 8 sixth graders) returned incomplete data and were excluded
from the final sample.
Data were collected by the third author in two-hour group sessions. Items
were read aloud to children. To reduce order effects, the instruments were
randomized across portions of the sample. Six weeks after the first data
collection, the CSS was administered afresh to a randomly selected subsample
of 113 children—62 fourth graders and 51 sixth graders. The average test-retest
reliability coefficient was quite satisfactory: r ¼ .76. For the overall sample of
833 children, missing values (, 0.5%) were replaced by the estimates
computed throug h maximum likelihoo d criteria and the e xpectation
maximization (EM) algorithm. Employing the PMM algorithm instead would
fit the metric level better (Asendorpf, van de Schoot, Denissen, & Hutteman,
2014), but in this case and with so few missing values, any biasing EM effect
on the data seems trivial.
Measures
Children’s Solitude Scale. The scale was constructed on the basis of (a) two
interview studies (total N ¼ 449) aiming at satisfying content validity criteria
(Galanaki, 2004); and (b) a pilot study with the initial version of the CSS aiming
at eliminating psychometrically redundant or inappropriate items.
The initial version of the CSS consisted of 60 items describing the following
uses of solitude: self-discovery, daydreaming, concentration, problem-so lving,
relaxation, peacefulness, restoration, intimacy (with others in fantasy),
spirituality, autonomy, privacy, secrecy, self-control, maste ry, freedom from
criticism, and indoor and outdoor activities (e.g., solitary play). The rating scale
ranged from 1 (that’s not at all true about me)to5(that’s always true about me).
The scale was originally constructed in Greek. Then it was translated into English
and back translated into Greek by two bilingual translators and minor
adjustments were made. Finally, a native-English speaking expert on social
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 3
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

withdrawal and solitude made further adjustments to the English version. The
written instructions to participants are give n in the Appendix.
Three other instruments were employed in order to examine the conver gent
and divergent validity of the CSS.
The Ability to Be Alone Questionnaire (Youngblade et al., 1999) consists of
25 items assessing children’s perceptions of their ability to be alone (i.e., attitude
towards aloneness). The 25 items were split into two subscales: Aversion to Being
Alone (12 items) and Ability to Be Alone (13 items).
The Aversion to Aloneness and the Affinity for Aloneness subscales are part of
the Loneliness and Alonene ss Scale for Children and Adolescents (Marcoen &
Goossens, 1993). Each subscale consists of 12 items assessing children’s
negative and positive attitude towards alon eness.
The Children’s Loneliness and Social Dissatisfaction Sca le (Asher
& Wheeler, 1985) is a 16-item scale measuring children’s loneliness and social
dissatisfaction at school.
Internal consistency estimates for each of the previous scales as computed for
the present data are presented in Table 3. These scales were also translated into
English and back translated into Greek by two bilingual tran slators, the same
ones who translated the CSS.
RESULTS
Confirmatory factor analysis and CSS dimensions
The initial 60-item CSS version was psychometrically examined for extreme
skewness, multicollinearity, large measurement errors, extensive confidence
limits and poor reliability potential. This examination in combination with
exploratory factor analyses (through SPSS 18) led to the exclusion of 15 items.
Next, confirmatory factor analysis (Jo
¨
reskog & So
¨
rbom, 1996) was conducted
using LISREL 8.3 and MS-Excel on the final 45-item version of the CSS.
We defined our CFA models based on previous preliminary data. More
specifically, in a Belgian sample, Goossens (2014) found a three-factor solution
with one single factor including self-reflection and concentration and two
separate factors for privacy (freedom from criticism) and activities. In a Greek
sample, Galanaki, Mylonas, and Vogiatzoglou (2008) found a four-factor
solution with four separate factors, that is, self-reflec tion, freedom from criticism/
independence/privacy, activities and concentration. Thus, we tested for the
independence model and three consecutive models: a unifactorial model, a four-
factor model and a modifi ed four-factor model (Figure 1).
The independence and unifactorial structure models were easily rejected
(Table 1). The four-factor model was not an acceptable solution: RMSEA was
adequate; the x
2
/df ratio dropped dramatically showing vast model improvement;
the Tucker-Lewis, BIC and AIC indices showed acceptable improvement across
4
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

s1
0.75
s2
0.76
s3
0.70
s4
0.67
s5
0.84
s6
0.72
s7
0.69
s8
0.74
s9
0.84
s10
0.72
s11
0.74
s12
0.82
s13
0.71
s14
0.68
s15
0.81
s16
0.80
s17
0.64
s18
0.81
s19
0.77
s20
0.80
s21
0.67
s22
0.67
s23
0.66
s24
0.73
s25
0.72
s26
0.75
s27
0.63
s28
0.61
s29
0.66
s30
0.78
s31
0.58
s32
0.82
s33
0.59
s34
0.65
s35
0.74
s36
0.57
s37
0.82
s38
0.78
s39
0.81
s40
0.68
s41
0.67
s42
0.79
s43
0.72
s44
0.71
s45
0.82
f1 1.00
f2
1.00
f3
1.00
f4
1.00
0.50
0.48
0.55
0.58
0.40
0.53
0.55
0.51
0.40
0.53
0.51
0.42
0.54
0.56
0.44
0.45
0.60
0.43
0.48
0.44
0.57
0.58
0.58
0.52
0.52
0.50
0.61
0.62
0.58
0.46
0.65
0.42
0.64
0.59
0.51
0.66
0.43
0.47
0.44
0.56
0.57
0.45
0.53
0.54
0.42
0.61
0.44
0.62
0.72
0.34
0.35
0.12
0.19
0.20
0.16
0.16
0.12
0.22
0.15
0.14
0.14
0.15
0.15
0.15
Chi-Square=2611.37, df=926, P-value=0.00000, RMSEA=0.047
Figure 1. Confirmatory factor analysis outcomes: Four-factor 45-item solution (modified, including
error covariances strictly within factors).
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 5
CFA models, and GFI and AGFI increased. However, root mean square residual
(RMR) was still high and CFI wa slow. The model improved when we included
error covariances (modified four-factor model), allowing for 13 such error
covariances to differ from zero, strictly within factors (for more details,
see Appendix).
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

TABLE 1
Confirmatory factor analysis: Independence, unifactorial and four-factor solutions
Model
x
2
df p
x
2
/df RMSEA [90% CI] GFI RMR CFI TLI D
x
2
Ddf AIC BIC
a 10,894.91 990 , .00001 11.01 – – – – – – – 10,984.91 11,197.54
b 7,817.95 945 , .00001 8.27 .093 [.092, .095] .71 .079 .60 .27
a-b
2,076.96** 45 7,797.95 8,423.20
c 3,196.95 939 , .00001 3.40 .054 [.055, .059] .85 .067 .80 .88
a-c
.67
b-c
7,697.96**
4,421.00**
51
6
3,388.95 3,842.52
d 2,611.37 926 , .00001 2.82 .047 [.045, .049] .88 .057 .84 .91
a-d
.75
b–d
8,283.54**
5,206.58**
64
19
2,829.37 3,344.40
Notes: a, Independence model; b, Unifactorial solution (Single-factor model); c, Four-factor model, no error covariances estimated; d, Four-factor model solution
including error covariance estimates strictly within factors.
RMSEA, Root Mean Square of Approximation; CI, Confidence Interval; GFI, Goodness-of-Fit Index; RMR, Root Mean Square Residual; CFI, Comparative Fit
Index; TLI, Tucker-Lewis Index; AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion.
** p , .001.
6
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

We recalculated the model accepting some estimable collinearity error within
each of the four factors. As expected due to the large sample size, chi-square
remained statistically significant; however, RMSEA, RMR and x
2
/df decreased,
whereas GFI, AGFI and CFI increased. All these fit indices const itute a better—
but not perfect— solution overall, with BIC and AIC also further improving.
Finally, TLI further improved; with respect to the reduced four-factor model—
without any error covariance parameters tested—TLI reached .24, also a non-
zero value. We do need to focus on the less than perfect CFI index for the four-
factor modified model (d) which is indeed rather low (.84). However, this is a
scale tested for the first time and under a stringent rationale, therefore some
irregularities seem inevitable. Reasons may be the self-report nature of the scale,
the age of the participants and the collinearity problems modelled, albeit not
avoided; these and other limitations may have added statistical noise. We could
have dropped some items to enhance factor consistency and reduce error levels,
thus stabilizing the solution, but this would have negative effects on the factors’
theoretically expected validity, leading to a theoretically non-defensible solution.
The four factors were named as following: Self-Reflection (18 items),
Autonomy/Privacy (13 items), Activities (7 items) and Concentration (7 items)
(Table 2).
Table 3 includes basic psychometric properties for the four factors and
descriptive statistics for all variabl es, with all factors reaching satisfactory
internal consistency. With respect to reliability, the commonly employed
Cronbach’s alpha and the respective confidence intervals were first computed and
are reported in Table 3. However, as Cronbach alpha is considered invalid in case
of latent variables with multiple indicators and this invalidity is more profound
when correlated errors exist, we also computed McDonald’s v indices (Gignac,
2009) for each factor and for the overall 45-item scale. These v indices reached
.85, .85, .74 and .70 for the four factors, respectively. For the full 45-item scale v
reached .94.
Convergent-divergent validity
To gain insight in the convergent and divergent validity of the CSS, we correlated
the four solitude aggregate scores with other available measures with which
we might theoretically expect certain levels of association. The four solitude
aggregate scores were significantly and positively associated (Pearson r
correlations) with Ability to Be Alone and Affinity for Aloneness (see Table 4).
The associations of the four solitude dimensions were stronger for Affinity for
Aloneness than for Ability to Be Alone. In contrast, the correlations between the
negative attitude towards aloneness (i.e., Aversion to Being Alone and Aversion
to Aloneness) and the four solitude dimensions were weak. Self-Reflection was
the only solitude factor that had a significant but low positive correlation with
Loneliness and Social Dissa tisfaction.
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 7
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

TABLE 2
Confirmatory factor analysis loadings: Four-factor solution for the CSS
“I like to be alone ...” Loading Factor
36 ...in order to imagine nice things 0.66 SR
33 ...to daydream 0.64 SR
23 ...in order to think about my problems and how to solve them 0.58 SR
29 ...in order to think how I’m going to solve future problems 0.58 SR
41 ...at night to think about the events that happened during the day 0.57 SR
14 ...so that my mind can wander 0.56 SR
44 ...to think about a mistake I’ve made and how to correct it 0.54 SR
10 ...because it is peaceful and quiet 0.53 SR
8 ...to get to know myself better, what I like and don’t like about me 0.51 SR
11 ...to make plans for the future 0.51 SR
35 ...to think about what I can do to get over my loneliness 0.51 SR
19 ...in order to imagine that I’ve got someone for company 0.48 SR
30 ...to get over my sorrow when others have hurt me 0.46 SR
16 ...in order to imagine that I’m very close to the ones I love, even though they are not there 0.45 SR
15 ...in order to think what a nice time I have when I’m with my loved ones 0.44 SR
20 ...to think of why I’m alone 0.44 SR
18 ...to write in my diary 0.43 SR
32 ...in order to hide my sadness and cry in secret 0.42 SR
31 ...in order to be able to do something without anybody watching me 0.65 AP
28 ...because I can have all the space (e.g., the room) for myself 0.62 AP
27 ...to do what I want without others telling me what to do 0.61 AP
17 ...to do something in secret, without others seeing me 0.60 AP
21 ...because I can misbehave without anyone telling me not to 0.57 AP
40 ...so that I can control things 0.56 AP
43 ...in order to eat whatever I want 0.53 AP
26 ...to mess about, wander around, or simply do nothing 0.50 AP
38 ...to do things I can’t do with other children 0.47 AP
39 ...to use the computer 0.44 AP
37 ...to lie down and sleep 0.43 AP
12 ...so that others don’t make me do something I don’t want to 0.42 AP
45 ...so that nobody knows me as if I were a stranger and unknown 0.42 AP
34 ...to play with my toys 0.59 A
22
...to go for a walk (e.g., to a playground or a park) 0.58 A
7 ...to exercise or get involved in a sport (e.g., ball, bicycle) 0.55 A
3 ...to play video or computer games 0.55 A
25 ...to do crafts and puzzles 0.52 A
1 ...in order to come up with new ideas of how to make several things 0.50 A
9 ...in order to draw 0.40 A
4 ...because it is easier to concentrate better 0.58 C
13 ...because it helps me to better understand what I read 0.54 C
6 ...because I do some things better when nobody bothers me 0.53 C
24 ...to get close to God and pray 0.52 C
2 ...in order to do my homework 0.48 C
42 ...to read a book (other than for school) 0.45 C
5 ...in order to do chores around the house or the garden 0.40 C
Note: SR, Self-Reflection; AP, Autonomy/Privacy; A, Activities; C, Concentration.
8
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

TABLE 3
Descriptive statistics for all variables and basic psychometric properties for the four factors of the CSS
Measures
Xs
ˆ
e
s
ˆ
Skewness Kurtosis Min-Max Cronbach’s
a
95% CI
iii
Self-Reflection
a
3.17 .27
i
.76 2 .22 2 .63 1.18– 4.83 .87 [.85, .88]
Autonomy/Privacy
a
3.11 .34
i
.84 2 .13 2 .65 1.08– 5.00 .84 [.82, .85]
Activities
a
2.90 .45
i
.88 .20 2 .60 1.00– 5.00 .74 [.71, .76]
Concentration
a
3.37 .45
i
.83 2 .20 2 .52 1.00– 5.00 .71 [.68, .74]
Ability to Be Alone
b
.89 .013
ii
.37 .15 2 .34 0–1.99 .75 [.72, .77]
Aversion to Being Alone
b
1.15 .015
ii
.44 2 .44 2 .41 0–2.00 .82 [.80, .84]
Affinity for Aloneness
c
2.70 .019
ii
.55 2 .21 2 .20 1.00– 4.00 .78 [.76, .80]
Aversion to Aloneness
c
2.84 .020
ii
.58 2 .40 2 .16 1.08– 4.00 .81 [.79, .83]
Loneliness and Social Dissatisfaction
d
1.77 .023
ii
.67 1.38 2.16 1.00–4.77 .85 [.84, .87]
Notes: a,d, Expected range of scores: 1–5; b, Expected range of scores: 0 – 2; c, Expected range of scores: 1 – 4.
i, Standard errors
^
s
e
were calculated using Cronbach’s a; ii, Standard errors were calculated using the central limit theorem; iii, Feldt confidence intervals for
Cronbach’s a coefficients.
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 9
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

DISCUSSION
This paper describes the initial validation of a 45-item self-report scale—the
Children’s Solitude Scale—that measures uses of solitude among children. Self-
Reflection, Autonomy/Privacy, Activities and Concentration emerged as
meaningful uses or functions of children’s solitude and are in line with the few
existing conceptualizations of the content of these solitude experiences (e.g.,
Averill & Sundararajan, 2014).
Adequate internal consistency and test-retest reliability were found. As to
convergent validity, the CSS is associated, as expected, with a positive attitude
towards aloneness and the ability to be alone. As to its divergent validity,
contrary to expectations, the CSS is rather unrelated to the negative attitude
towards aloneness. Children seem to seek solitude for all sorts of purposes,
independent of the extent to which they negatively evaluate aloneness. Young
people’s ambivalence to being alone, which implies that they can both
appreciate some of the benefit s of aloneness on subsequent behaviour but
report mostly negative feelings when they actually are on their own (Larson,
1999), may cause them to feel positive about being alone at some points in
time and negative at other points. This so-called paradox of solit ude could be
another explanation for our low correlations between the CSS subscales
and measures of negative attitudes to being alone. Also, the CSS has a
significant but low correlation with loneliness and social dissatisfaction, which
supports the distinctiveness of attitude towards aloneness and loneliness during
late childhood as found in earlier factor analytic work (Goossens & Beyers,
2002).
Overall, the CSS can be considered a reliable and valid measure to assess
the uses of voluntary aloneness during childhood. However, given the sub-
optimal fit, further research is needed with regard to the structure of this type of
solitude experiences in the Greek and other cultural contexts. For example, an
TABLE 4
Pearson r correlations between measures
Measures 1 2 3 4 5 6 7 8 9
1. Self-Reflection –
2. Autonomy/Privacy .50* –
3. Activities .34* .49* –
4. Concentration .55* .22* .25* –
5. Ability to Be Alone .26* .46* .41* .15* –
6. Aversion to Being Alone .12* – .09 –.15* .10 –.50* –
7. Affinity for Aloneness .53* .46* .31* .33* .50* –.15* –
8. Aversion to Aloneness .06 –.04 –.12* .02 –.42* .66* – .09 –
9. Loneliness and Social Dissatisfaction .12* .08 .11 .06 .14* .03 .14* – .07 –
Note:*p , .001.
10
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

alternative rating scale, such as how often they engage or how much they like
engaging in these solitary pursuits, could be used, which might be better
understood by school-age children. The instrument could also help in
examining age and gender differences as well as disentangling the associations
between children’s reasons or motives to search solitude and types (healthy or
maladaptive ones) of social withdrawal (e.g., Coplan & Bowker, 2014) as well
as ways of coping with loneliness (e.g., Besevegis & Galanaki, 2010). In all,
this novel measure could advance current understanding of time spent alone in
childhood.
Limitations
A confirmatory factor analysis model containing estimated error covariances
does include some estimable collinearity effects despite the fact that these remain
within factors. Although they do not indicate cross-loadings, they can still pose
threat to the solution as they may indicate metric interference, at considerable
levels for future research. Possible reasons for the less-than-perfect fit observed
in our data were suggested in the Method section. This kind of fit implies that the
scale’s metric and statistical strength requires some further testing and replication
in future research.
REFERENCES
Asendorpf, J. P., van de Schoot, R., Denissen, J. J. A., & Hutteman, R. (2014). Reducing bias due to
systematic attrition in longitudinal studies: The benefits of multiple imputation. International
Journal of Behavioral Development, 38, 453– 460. doi:10.1177/0165025414542713
Asher, S. R., & Wheeler, V. A. (1985). Children’s loneliness: A comparison of rejected and neglected
peer status. Journal of Consulting and Clinical Psychology, 53, 500– 505. doi:10.1037/0022-
006X.53.4.500
Averill, J. R., & Sundararajan, L. (2014). Experiences of solitude: Issues of assessment, theory, and
culture. In R. J. Coplan & J. C. Bowker (Eds.), The handbook of solitude: Psychological
perspectives on social isolation, social withdrawal, and being alone (pp. 90 – 108). Malden, MA:
Wiley Blackwell.
Besevegis, E., & Galanaki, E. P. (2010). Coping with loneliness in childhood. European Journal of
Developmental Psychology, 7, 653 – 673. doi:10.1080/17405620903113306
Coplan, R. J., & Bowker, J. C. (2014). All alone: Multiple perspectives on the study of solitude.
In R. J. Coplan & J. C. Bowker (Eds.), The handbook of solitude: Psychological perspectives
on social isolation, social withdrawal, and being alone (pp. 3 – 13). Malden, MA: Wiley
Blackwell.
Galanaki, E. P. (2004). Are children able to distinguish among the concepts of aloneness, loneliness,
and solitude? International Journal of Behavioral Development, 28, 435– 443. doi:10.1080/
01650250444000153
Galanaki, E. P., Mylonas, K., & Vogiatzoglou, P. (2008, July). Children’s Solitude Scale: A new
instrument for measuring beneficial aloneness in childhood. Paper presented at the 20th Biennial
Meeting of the International Society for the Study of Behavioural Development. University of
Wu
¨
rzburg, Wu
¨
rzburg, Germany.
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 11
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

Gignac, G. C. (2009). Psychometrics and the measurement of emotional intelligence. In C. Stough,
D. Saklofske, & J. D. A. Parker (Eds.), Assessing emotional intelligence: Theory, research, and
applications (pp. 9 – 40). New York: Springer.
Goossens, L. (2014). Affinity for aloneness in adolescence and preference for solitude in childhood:
Linking two research traditions. In R. J. Coplan & J. C. Bowker (Eds.), The handbook of solitude:
Psychological perspectives on social isolation, social withdrawal, and being alone (pp. 150– 166).
Malden, MA: Wiley Blackwell.
Goossens, L., & Beyers, W. (2002). Comparing measures of childhood loneliness: Internal
consistency and confirmatory factor analysis. Journal of Clinical Child Psychology, 31, 252–262.
doi:10.1207/S15374424JCCP3102_10
Goossens, L., Lasgaard, M., Luyckx, K., Vanhalst, J., Mathias, S., & Masy, E. (2009). Loneliness and
solitude in adolescence: A confirmatory factor analysis of alternative models. Personality and
Individual Differences, 47, 890 – 894. doi:10.1016/j.paid.2009.07.011
Jo
¨
reskog, K. G., & So
¨
rbom, D. (1996). LISREL 8 user’s reference guide. Uppsala, Sweden: Scientific
Software International.
Larson, R. W. (1997). The emergence of solitude as a constructive domain of experience in early
adolescence. Child Development, 68, 80–93. doi:10.2307/1131927
Larson, R. W. (1999). The uses of loneliness in adolescence. In K. J. Rotenberg & S. Hymel (Eds.),
Loneliness in childhood and adolescence (pp. 244 – 262). New York: Cambridge University Press.
Long, C. R., & Averill, J. R. (2003). Solitude: An exploration of benefits of being alone. Journal for
the Theory of Social Behaviour, 33, 21–44. doi:10.1111/1468-5914.00204
Maes, M., Klimstra, T., Van den Noortgate, W., & Goossens, L. (2015). Factor structure and
measurement invariance of a multidimensional loneliness scale: Comparisons across gender and
age. Journal of Child and Family Studies, 24, 1829– 1837.
Marcoen, A., & Goossens, L. (1993). Loneliness, attitude towards loneliness and solitude: Age
differences and developmental significance during adolescence. In S. Jackson & H. Rodriguez-
Tome
´
(Eds.), Adolescence and its social worlds (pp. 197– 227). Hillsdale, NJ: Erlbaum.
Peplau, L. A., & Perlman, D. (1982). Loneliness: A sourcebook of current theory, research, and
therapy. New York: Wiley.
Storr, A. (1988). Solitude: A return to the self. New York: Free Press.
Teppers, E., Luyckx, K., Vanhalst, J., Klimstra, T., & Goossens, L. (2014). Attitudes towards
aloneness during adolescence: A person-centred approach. Infant and Child Development, 23,
239– 248. doi:10.1002/icd.1856
Winnicott, D. W. (1965). The maturational processes and the facilitating environment: Studies in the
theory of emotional development. New York: International Universities Press.
Youngblade, L. M., Berlin, L. J., & Belsky, J. (1999). Connections among loneliness, the ability to
be alone, and peer relationships in young children. In K. J. Rotenberg & S. Hymel (Eds.),
Loneliness in childhood and adolescence (pp. 135–152). Cambridge, MA: Cambridge
University Press.
12
E. P. GALANAKI ET AL.
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015

APPENDIX
CSS: Instructions
Most people sometimes feel the need to be alone, without others around them, because it
does them some good, or helps them in some ways. We would like you to tell us the
reasons why you like to be alone. There are no right or wrong answers; we are interested in
what you believe and feel. For each question, circle one answer that shows how true each
statement is for you.
Confirmatory factor analysis models
In our analysis, we specified four confirmatory factor models: the independence model,
the unifactorial model and two theory-driven four-factor models. The last two differ only
with respect to fixed parameters and to constraints not imposed on the final model. For
this final and best-fitting model, we decided to allow for 13 error covariances to differ
from zero (with the remaining 767 possible ones still fixed to zero), as confirmatory
factor analysis can include error covariances designating that two measures covary due
to other than the shared factor’s influence, such as method effects. Although in a perfect-
fit model and according to the true score theory, error terms across observed variables
should not be correlated, especially if these appear across factors, this does not apply to
our study because (a) the final model is not optimal in terms of fit due to the lower than
.90 CFI index, and (b) we allowed for error covariances to be estimated strictly within
factors in our attempt to model indications for external sources of variance. We kept this
estimation to a minimum to avoid over-fitting and employed the generality rule of un-
constraining only those error covariances which would significantly reduce the model
chi-square in our final model solution. The transitivity rule (of u
1
-u
2
&u
2
-u
3
estimated,
and u
1
-u
3
estimated as well) was employed in one triad of correlated errors, as in this
case the chi-square drop was significant. As the estimated error covariances were
calculated strictly within already imposed factors, there was no need at this point to add
any additional factor to the model (as would be the case if at least three error terms
across factors were correlated). Through these procedures, we managed to describe the
best fitting four-factor solution and reach our conclusions about the scale. However, and
as method factors may always be lurking, we have to be cautious about possible
exogenous variables which may have some effect on the structure of the scale and the
scores computed.
EVALUATING VOLUNTARY ALONENESS IN CHILDHOOD 13
Downloaded by [T&F Internal Users], [Rozanna Harrison] at 02:19 11 November 2015