Preliminary Models of Risk and Protective Factors for Childhood Homesickness: Review and Empirical Synthesis

by Christopher A. Thurber, Marian D. Sigman
Citation
Title:
Preliminary Models of Risk and Protective Factors for Childhood Homesickness: Review and Empirical Synthesis
Author:
Christopher A. Thurber, Marian D. Sigman
Year: 
1998
Publication: 
Child Development
Volume: 
69
Issue: 
4
Start Page: 
903
End Page: 
934
Publisher: 
Language: 
English
URL: 
Select license: 
Select License
DOI: 
PMID: 
ISSN: 
Abstract:

Child Development, August 1998, Volume 69, Number 4, Pages 903-934

Preliminary Models of Risk and Protective Factors for Childhood

Homesickness: Review and Empirical Synthesis

Christopher A. Thurber and Marian D. Sigman

Empirical research and conventional wisdom have suggested numerous risk and protective factors for the development of homesickness. Yet no study has integrated predictors and sequelae of homesickness into a testable statistical model. As a first step in developing a pathogenic model of homesickness in children, this study measured, factor analyzed, and modeled 14 predictors and 8 sequelae of homesickness. Using a sample of 293 boys, ages 8-16, spending 2 weeks at an overnight summer camp, this study tested 2 alternate models, focusing on the roles of boys' interpersonal attitudes, perceived control, and separation expectations in the subsequent development of homesickness. Results indicated that interpersonal attitudes and perceived control may predict boys' preseparation beliefs about whether they will become homesick. This "homesick disposi- tion" combines with little prior separation experience to account for 69%of the variance in self-reported home- sickness. Homesickness was not a powerful predictor of negative emotion, whereas interpersonal attitudes and perceived control predicted 70% of the variance in negative emotion. Results are discussed in the context

of contemporary theories of homesickness.

INTRODUCTION

Homesickness is the distress or impairment caused
by an actual or anticipated separation from home. It is characterized by acute longing and preoccupying thoughts of home and attachment objects. Typically, this distressing preoccupation is triggered by actual separation from home, but even an anticipated sepa- ration can incite homesickness (Thurber, 1998; van Tilburg, Vingerhoets, van Heck, & Kirschbaum, 1996). Moderately and severely homesick children and adolescents usually report depressed and anx- ious emotions and exhibit both internalizing and ex- ternalizing behavior problems (Brewin, Furnham, & Howes, 1989; Eurelings-Bontekoe, Vingergoets, & Fontijn, 1994; Fisher & Hood, 1987,1988; Fisher, Mur- ray, & Frazer, 1985; Hojat & Herman, 1985; Thurber, 1995, 1998; Thurber, Sigman, Weisz, & Schmidt,

1998). The defining feature of homesickness is recur- rent cognitions focused on home (e.g., house, loved ones, homeland, home cooking, returning home).

Since the time of the Greek physician Hippocrates (ca. 460-ca. 377 B.c.), who believed that homesick- ness was due to a surfeit of black bile in the blood (Zwingmann, 1959), scholars have invoked a variety of theories to explain homesickness. The neurological explanation of Swiss physician Johannes Hofer (1688) held that homesickness resulted from exposure to foreign environments. This exposure caused "vital spirits [to] constantly surge back and forth through the nerve fibres in which the impressions of the na- tive land are stored." AS a result, "vital spirits . . . do not flow in adequate quantity or potency to other parts of the brain to serve the natural functions" (paraphrased in Rosen, 1975, p. 342). Hofer observed that the homesick person would then become weak,
anxious, and febrile, and sometimes die. Other early theorists discounted stressful environmental factors, arguing that "homesickness is a disease whose cause will be sought for in vain in altered conditions of cli- mate and air, since it is clearly the result of an emo- tional disturbance" (Ziickert, 1768, paraphrased in Rather, 1965, p. 177). Each of the modern theories and ideas about homesickness etiology has its own em- phasis, but most are inherently transactional, in- tegrating characteristics of the person, the environment, and the circumstances surrounding the separation. For example, Vernberg and Field (1990) hypothesized that homesickness results from the di- athesis of insecure attachment to caregivers and the stress of social and other skill deficits in the novel separation environment. Table 1 summarizes key theories and ideas that contemporary researchers have constructed themselves or cited from related work to explain homesickness (see McCann, 1941; Woulff, 1975; and especially Zwingmann, 1959, for historical reviews).

In addition to empirical theories, there is consider- able conventional wisdom about what causes home- sickness, specifically in children. Most of these intu- itive theories come from the literature on residential summer camping. Camp directors, counselors, and pediatric nurses have alternately proposed that

01998 by the Society for Research in Child Development, Inc All rights reserved. 0009-3920/98/6904-0012$01.00

Table 1 Theories Used by Contemporary Researchers to Explain Homesickness

Theory or Idea and Selected References" Possible Explanation of Homesickness

Attachment theory (Bartholomew & Horowitz, 1991; Bowlby, 1969)

Attentional resources model (Fisher, Frazer, & Murray, 1986; Harris, 1989; Porritt & Taylor, 1981)

Competing demands model (Fisher, 1988, 1990b)

Control-demand discrepancy /job strain models (Fisher, 1984, 1989; Matheny & Cupp, 1983)

Contextual/environmental resource theories (Schwarzer, Hahn, & Jerusalem, 1993; Stokols, Schumaker, & Martinez, 1983)

Discontinuity /self-awareness/role change theories (Mechanic, 1983; Wapner, Kaplan, & Ciottone, 1981; Wickland, 1975)

Geographical attachment (Abt, 1983/ 1988)

Instinct theory (Hall, 1897; Seabury, 1924)

Interruption theory (Mandler, 1975)

Negative expectations (Brewin, Furnham, & Howes, 1989)

Neural habituation (Conklin, 1935)

Nostalgic reaction theory (Zwingmann, 1959)

Opponent process theory (Solomon, 1980)

Peer social support (Brewin et al., 1989; Taka- hashi & Majima, 1994)

Person-environment congruence model (French, Caplan, & van Harrison, 1982; Lerner, Baker, & Lerner, 1985)

Psychoanalytic theory (Freud, 1920/ 1966; Levy, 1952; Redl, 1947)

Social flexibility model (Rose, 1947, 1948)

Insecure attachment to primary caregivers (i.e., uncertainty about the reliability and/or positivity of caregivers' responses to displays of distress) results in anxi- ety and distress when people are separated from home and primary caregivers.

Preoccupying thoughts of home occupy a finite amount of attentional resources and preclude cognitive, behavioral, and emotional investment in the novel separa- tion environment, resulting in homesickness.

The novel environment fails to provide "sources of distracting material which com- pete with the demands made by distressing thoughts of home," and/or people fail to explore new opportunities in the novel environment, resulting in homesick- ness.

The perception of reduced control in a demanding or unpleasant environment cre- ates a discrepancy between people's desired goals and their perceived reality, promoting a homesick desire to return home and achieve control-demand congru- ence.

Preseparation contextual factors, such as frequent, forced relocation, combine with pathogenic environmental factors, such as unemployment and low social sup- port, to produce ill health, negative emotion, dissatisfaction, and homesickness.

Strange aspects of the novel environment, possibly including climate, customs, and the adaptation of a new role, enhance self-focus, promote an unstable self- concept, and encourage isolation, all of which contribute to anxiety and homesickness. Environmental attachment to the physical characteristics of the homeland, com- bined with a "failure to organize the world from a center of one's own," results in homesickness when people move from one environment to another.

Homesickness occurs when opposing instincts-inclinations toward home and away from home-struggle to dominate the individual. Alternately, homesick- ness occurs when the "love of home" instinct dominates the emotions. Interruption of a dominant/comfortable old routine exacerbates feelings of separa- tion and loss, and arouses anxiety, preventing integration of novelty and causing a perseveration of the old routine, a focus on home, and eventually homesick- ness. Expectations of high homesickness frequency in others who are separated from home and primary caregivers predicts people's actual self-reported homesickness frequency. "An abandonment of certain patterns of stimuli which are accompanied by pleas- ant . . . feelings and which are associated with the cranial division of the auto- nomic nervous system" causes homesickness (paraphrased by Mccam, 1941,

p. 178).

"An individual's response to change and/or an abstraction thereof . . . in form of a symbolic return to, or reinstatement of, those features of his Erlebrlisracrnlh which, at the time, are perceived as having (had) the greatest gratificational value" causes homesickness (p. 308).

Positive reinforcement (e.g., love) stimulates not only a positive brain process, but also a negative "opponent process." When the positive process stops abruptly, as it does in separation, the negative process produces felt affect (e.g., homesick- ness) that dissipates only gradually.

Peer social support and affiliation in the novel environment mediate people's home- sickness more than do close affective relationships with absent family members; without strong peer support in the novel environment, homesickness worsens.

One or more aspects of the novel environment (e.g., interpersonal, aesthetic, profes- sional, recreational, vocational) fail to meet a person's needs and expectations, en- hancing feelings of separation, loss, and disappointment, resulting in homesick- ness.

"Overprotected superego development" makes young people feel nervous, guilty, and homesick when they enjoy separations, gain independence while away from home and primary caregivers, or forget temporarily about their families.

A history of rejecting and neglectful interactions with parents, combined with little previous social experience outside of the family, results in feelings of inade- quacy, difficulty establishing social contacts in the new environment, and home- sickness.

"heory references do not necessarily imply that the theorists themselves have discussed homesickness, only that others have cited these theories as possible explanations for homesickness. Erlebnisracrnl is from the German Erlebnis, meaning "experience," and Raun~,meaning "space." In this context, Erleb~zisraumis perhaps best translated as "experiential space" or "psychological world."

homesickness is caused by parental overinvolvement and overprotectiveness (Burrow, 1992; Doherty, 1940; Hartwig & Myers, 1961; Joy, 1957; Meeks, 1968; Mitchell & Crawford, 1965; van Wagenen, 1933; cf. Schwab, 1925, cited in Zwingmann, 1959); parental underinvolvement (Burrow, 1992); contagion from other homesick children (Davidson, 1983; Hartwig & Myers, 1961; Mitchell & Crawford, 1965; Winland- Brown & Maheady, 1990); and mention of the word "homesick (Burrow, 1992).

Some of the early conventional wisdom in the camping literature about homesickness etiology dovetails with recent empirical theories. For exam- ple, Berg (1958) proposed that children's homesick- ness at summer camp was due to unfamiliar sur- roundings, an idea later echoed in environmental strangeness theories (e.g., Wickland, 1975). Doherty (1940) proposed that children's homesickness was due to a lack of friends, a central idea in peer social support theories (e.g., Takahashi & Majima, 1994) and social flexibility models of homesickness (Rose, 1947, 1948). Other conventional wisdom is contra- dicted by empirical research. For example, a recent study suggested that repeatedly mentioning the word "homesick" and questioning children daily about their emotions does not cause or exacerbate homesickness (Thurber, 1998).

Several factors are thought to exacerbate preex- isting homesickness, such as physical injuries and ill- nesses (Doherty, 1940; Hofer, 1688; Taylor, 1986), per- ceived distance from home (Mooney, Sherman, & Lo Presto, 1991), poor social skills (Rose, 1947,1948), and rumination, stimulated by a lack of distracting, struc- tured activity (Harris, 1989; Harris & Guz, 1986).' In addition, there is substantial evidence that negative, pessimistic appraisal of negative life events (which may include separation from home) prolongs associ- ated depressive symptoms (which may include homesickness) (Hilsman & Garber, 1995; Nolen- Hoeksema, Girgus, & Seligman, 1992; Robinson, Garber, & Hilsman, 1995). Finally, if the circum- stances surrounding the separation are coercive, vio- lent, or traumatic, as may be the case with refugees, foster children, and hospitalized children, then resul- tant homesickness may be intensified (Alapack & Alapack, 1968; Arrendondo-Dowd, 1981; Cox, 1988; Nicassio & Pate, 1984).

Homesickness has both applied and theoretical

1. Many of these etiologic factors are mentioned in case stud- ies of homesickness in European soldiers. See Rosen (1975) and especially Zwingmann (1959) for reviews of the seventeenth-, eighteenth-, and nineteenth-century medical literature on home- sickness. See van Tilburg, Vingerhoets, and van Heck (1996) for a recent review.

Thurber and Sigman 905

clinical importance. From a theoretical standpoint, homesickness (insofar as its expression is a reaction to separation from loved ones) reflects the security of child-caregiver attachment and informs us about the goal-corrected behavioral systems that children ac- quire (Bowlby, 1969). Understanding homesickness better will also help us make informed improvements in the nosology of childhood internalizing disorders. The fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; APA, 1994) lists "extreme" homesickness and discomfort "to the point of misery" as a possible manifestation of sepa- ration anxiety disorder (p. 110). As with other child- hood anxiety disorders, empirical support for this classification is limited (Thurber, 1996).

From an applied clinical standpoint, understand- ing homesickness will guide prevention and inter- vention. Indeed, Rosen (1975) argued that military physicians have been striving to prevent suicide, de- pression, and debilitating physical illness secondary to homesickness since the late 1600s. Contemporary medical historians note that the prevalence of depres- sion and suicide are often highest among peoples "dislocated by social change" (Kleinman & Cohen, 1997, p. 87). Granted, suicide is a dramatic and rare consequence of severe homesickness. Clinically sig- nificant sequelae more commonly associated with homesickness include severe depressive and anxious symptoms (Brewin et al., 1989; Eurelings-Bontekoe et al., 1994; Fisher et al., 1985; Fisher & Hood, 1987, 1988; Hojat & Herman, 1985; Thurber, 1995); somatic complaints (Thurber, 1995); nontraumatic ailments (Fisher & Hood, 1987, 1988); internalizing and exter- nalizing behavior problems (Thurber, 1995, 1998; Thurber et al., 1998); academic difficulties (Burt, 1993; Fisher et al., 1985; Furnham & Bochner, 1986); absent- mindedness (Fisher et al., 1985; Fisher & Hood, 1988; but cf. Burt, 1993; Fisher & Hood, 1987); low self- esteem (Hojat & Herman, 1985; but cf. Fisher et al., 1985); and obsessional thoughts and behaviors (Fisher & Hood, 1987,1988). Less severe sequelae in- clude a proclivity to romanticize home (Fisher, 1989; Zwingmann, 1959); short-sightedness about the fu- ture (Platt & Taylor, 1967); increased confiding be- havior with peers (Brewin et al., 1989) and parents (Woulff, 1975); less socializing with native residents (Hojat & Herman, 1985; Pruitt, 1978; cf. Eisenbruch, 1990); decreased social satisfaction (Burt, 1993; Fisher et al., 1985; Kane, 1987); and decreased environrnen- tal satisfaction (Fisher et al., 1985; Kane, 1987).

The potentially severe sequelae of homesickness warrant prevention. Yet despite rich theoretical underpinnings and cross-cultural prevalence, research- ers have begun only recently to measure key risk and

906 Child Development

protective factors for homesickness. Table 2 summa- rizes 26 published, empirical studies that have di- rectly measured homesickness and potential predictors. The table is ordered by the approximate mean age of each study's sample.

The variety of theories and empirical findings in Tables 1and 2 suggests that testable models of home- sickness could be constructed by simultaneously measuring some of the most common risk factors. One set of testable models may come from integrat- ing the theories of attachment and perceived control. Both insecure relationships with attachment figures and low perceived control have been theoretically implicated in the etiology of homesickness, and both have been empirically associated with negative emo- tion.

Building on observations (Bowlby, 1969) and em- pirical data from infants (Ainsworth, Blehar, Wa- ters, & Wall, 1978; Ainsworth & Wittig, 1969), recent research with children, adolescents, and adults has substantiated a relation between depression and anxiety and insecure interpersonal relationships (Armsden & Greenberg, 1987; Armsden, McCauley, Greenberg, Burke, & Mitchell, 1990; Carnelley, Pie- tromonaco, & Jaffe, 1994; Persons, Burns, Perloff, & Miranda, 1993; Pattem, West, Mahoney, & Keller, 1993; West, Rose, & Sheldon, 1993).

Control beliefs theory (Weisz, 1990) has focused on children's cognitions and has proposed another explanation for some kinds of negative emotion. Building on theories of competence (White, 1959) and learned helplessness (Abramson, Seligman, & Teas-dale, 1978), recent control beliefs research with chil- dren and adolescents has substantiated a relation be- tween depression and low perceived control in social, academic, and behavioral domains (e.g., Weisz, Sweeney, Proffitt, & Carr, 1993; Weisz, Weiss, Was- serman, & Rintoul, 1987; Weisz et al., 1989).

Interestingly, Bowlby's original conceptualization of attachment theory portended control beliefs the- ory. Bowlby (1969) theorized that "control systems" explained the association between negative emotion and insecure attachment. He argued that children modulated their attachment behavior (e.g., crying, talking, approaching) according to the effect it had on their goal: proximity to caregivers. If caregivers responded negatively and/or inconsistently to chil- dren's bids for attention, this diminished children's interpersonal expectations of others and made them feel unworthy of others' love. It also made them per- ceive low interpersonal control and feel depressed and/or anxious.

Thus, interpersonal attitudes and perceived con- trol could possibly develop symbiotically-each enhancing or diminishing the other, based on the child's experiences with the world, and each provid- ing children with working models about their world. Some research with adolescents has demonstrated links between attachment and self-reported compe- tence (Papini & Roggman, 1992). However, interper- sonal attitudes have not been empirically linked to perceived control, despite their possible association. One integral goal of the present study was to deter- mine whether these two constructs were empirically related.

Regarding homesickness and other distress that some children experience when separated from home and primary caregivers, at least two hypotheses exist: (HI) Interpersonal attitudes and perceived control arnplih the relation between separation expectations and the negative emotion and homesickness that can result from separation. According to this hypothesis, children may have preseparation beliefs that they will become homesick (a "homesick disposition"), as well as negative impressions of the novel environ- ment, that diminish their expectations of how much fun the separation from home will be. These risks are then amplified by negative interpersonal attitudes and low perceived control. Such children lack confi- dence to negotiate the new environment because they feel incompetent, doubtful of the contingent relation between their behavior and positive outcomes, and doubtful that surrogate caregivers will assist them if they become distressed. Hence, they experience homesickness, depression, and anxiety.

An alternate hypothesis (H2) is that interpersonal attitudes and perceived control generate separation expectations, which in turn predict the valence of children's emotions and the intensity of their home- sickness during a separation. According to this hy- pothesis, children may have negative interpersonal attitudes and low perceptions of control, which in- crease their preseparation beliefs that they will be- come homesick and diminish their expectations of how much fun the separation from home will be. In turn, a homesick disposition and negative separation expectations predict homesickness and negative emotion. Figure 1 shows how a hypothetical "Atti- tudes and Control" factor could play either an ampli- fying (HI) or generative (H2) role in the etiology of homesickness. In either case, little previous experi- ence away from home is thought to augment a child's homesick disposition and the homesickness felt dur- ing a separation.

In either case, (HI) or (H2), previous research sug- gests additional paths, not pictured in Figure 1. Namely, homesickness will predict negative emotion and internalizing behavior (Thurber, 1995, 1998) as

Table 2 Contemporary Empirical Studies of Risk and Protective Factors for Homesickness

Study Author(s)
Thurber (1995)

Thurber & Weisz (1997)

Zimmerman & Bijur (1995)

Population and
Environment

329 boys at an all-boys overnight summer camp in the U.S. (mean age = 12.4; range: 8.0 to

16.7 years)

315 boys and 717 girls at single-sex overnight summer camps in the

U.S. (mean age = 12.6; range: 8.0 to 16.8 years)

87 boys and 100 girls at a modern Orthodox Jew- ish coeducational over- night summer camp in the U.S. (approx. age range: 8-12)

Measure(s) of
Homesickness

Beginning on thefirst night clf the separation: 14 con- secutive days of a sin- gle item, "I felt home- sick," administered as part of the self-report adjustment scale, Rate Your Day (RYD; Thur- ber, 1995), rated on an 11 point Likert scale

Beginning on thefirst night of the sc,pamtion: 14 con- secutive days of a three item homesickness scale, rated on a 30 point continuous scale, administered as part of the self-report adjust- ment scale, Rate Your Day-Revised (RYUR; 'Thurber, 1997)

At the conclusion of each child's 4 or 8 WLT~camp stay: cabin leader's re- port (based on the final 4 weeks of camp) on (1)a five item negative mood scale, (2) yes/no recall of child having complained of home- sickness, and (3) yes / no opinion about whether child was homesick

Measure(s) of Risk and/or Protective Factor (s)

2 months prror to the separa- tion: self-report and camp records of demo- graphics and previous separation experience

2 months prior to the scpara- tzon self-report and camp records of demo- graphics and prevlous separation experience

Sornctzmp durlng thefirst week of the scparatlon self-report of perceived control or the Perceived Control Scale (Weisz, Proffitt,& Sweeney, 1991)

Sometime during the second week of the separatron: self-report of method and goal of coping on the Ways of Coping with Homesickness Scale (WOCH; Thur- ber & Weisz, 1997)

Daily recordings by medical staff: number of visits to the camp infirmary; rea- son for visit to the camp infirmary

At the completion of camp: child gender and grade, but not exact age

Maln Flnding(s)"

Boys who reported high levels of homesickness were younger, had less previous separation cxperi- cncc, and had more previous homesickness than the boys who reported low homesickness.

Mean homesickness inten- sity was inversely corre- lated with age for boys (r = .35), but not for girls, and with previous separation experience for boys (r = -SO) and for girls (r = p.21).

Mean homesickness inten- sity was correlated with low general perceived con- trol (boys:r = .35; girls: r = .23), low perceived control over honrcsickness (boys: r = .49; girls: r = .40),low perceived control over separation (boys: r = .13; girls: r == .21), and a coping style char- acterized by rt~linquishcd control.

Girls were homesick more often than boys, based on leader opinion and recall of child com- plaints, but not based on mood scale.

A greater percentage of homesick children had multiple visits to the camp infirmary than non- homesick children.

Null Fmding(s)

Homesickness intensity was unrelated to geographic distance from howre or socioeconomic scale.

Homesickness intensity was unrelated to geographic distance from home or socioeconomic scale.

No gender differences in homesickness preva- lence or intensity.

Homesickness incidence, as measured by any of the three leader-report in- dices, did not differ be- tween the 8- to 9-year- old group and the 10- to 12-year-old group.

Table 2 (Continued)

Study Author(s)

Fisher, Frazer, & Murray (1986) (Study 1)

a

Fisher et al. (1986) (Study 3)

Fisher, Frazer, & Murray (1984)

Population and
Environment

117 first-year students at primarily single-sex boarding schools in the

U.K. (50% female; age range: 11-14 years)

21 first-year students at an all-boys board~ng school in the U.K. (age range: 12-16 years)

50 first-year boarding school students in the

U.K. (48% female; age range: 13-16 years)

Measure(s) of
Homesickness

Sonletinlc during the first school term: retrospective dichotomous self- report of homesickness incidence during the past school year

Sonletlrne durzng thefirst school term: self-report of homesickness inci- dence, frequency, con- text, and related worries in diaries com- pleted daily for 2 weeks

Beginning during thefirst few days of school: self-reported incidence of homesickness in diaries completed daily for 2 weeks

Measure(s) of Risk and /or Protective Factor(s)

Sometime during thc,first school term: self-report of demographics; per- ceived control over the decision to go to board- ing school; attitude about separation, health, and recent life events

Sometzmc during the first school term: self-report in diaries of demograph- ics, perceived control over the decision to go to boarding school, and recent life events

Beginning during the first few days of school: self-report in diaries of de- mographics, perceived control over the deci- sion to go to boarding school, and recent life events

Main Finding(s)"

Students who reported homesickness had less positlve anticipation about the separation, less previous separation experience, and more non-traumatic ailments than the students who re- ported no homesick- ness.

Homesickness frequency correlated with re- porting of fewer positive life events (r = .51).-

Most homesickness epi- sodes (70%) "were asso- ciated with periods of mental and physical pas- sizlity."

Students with siblings at the same school re- ported more problems, in general, than stu- dents without siblings at the school.

Null Finding(s)

Incidence of homesickness was unrel~ted to perceived control over the decision to attend board- ing school, geographical distance, or the presence of a sibling at the school.

No age or gender differences in homesickness incidence.

Homesickness frequency was unrelated to per-CP~VL~~

control over the decision to attend board- ing school, geographical distance, the presence of a sibling at the school, or previous separatii~n ex- perience.

No age or gender differences in homesickness severity.

Homesickness incidence was unrelated to previous separation experience at boarding schools, perceived control over the decision to attend board- ing school, gt.ographical distance, or recent life enents.

No age or gender differences in homesickness incidence.

Fisher et al. (1986) (Study 2)

Fisher, Elder, & Peacock (1990)

Eisenbruch, 1990

17 first-year students at a coed boarding school in the U.K. (47% female; age range: 13-1 7 years)

111 boys and girls at a coed outward bound boarding school in Aus- tralia (31 % female; age range: 14-16 years)

62 previous boarders; 17 previous day pupils; 32 new pupils

79 unaccompanied Cambo- dian refugee adoles- cents consisting of:

47 living in "group care" in Australia (resettled for an average of 28 months)

32 living with Cambodian and American families in the U.S. (resettled for an average of 16 months)

(gender composition, mean ages, and age ranges not reported)

Sornetinte during the first school ternt: self-report of homesickness inci- dence, frequency, and related worries in dia- ries completed daily for 2 weeks

6 weeks after arrival: retrospective self-report of homesickness incidence and frequency

Sorn<,tiin<,during the first 32 rnotlths after arrival: self-report on 15item, Likert scale questionnaire of cultural bereavement (memories of family, tra- ditions, and wartime trauma, as well as survi- vor guilt, regret, and PTSD symptoms), in- clud~ng three homesick- ness items, e.g., "Think- ing often about family and homeland"

Sorn~tirne during the first school t<>rrn: self-report in diaries of demograph- ics, perceived control over the decision to go to boarding school, and recent life events

6 weeks after arrival: retrospective self-report of perceived control and demand at school, pre- ferred control and de- mand at school, recent life events, and demo- graphics

Concurrent with the assess- ment of hont<,sickness: period of settlement, classi- fied into two duration groups:

Newly arrived group (less than or equal to 16 months)

Experienced group (be- tween 17 and 32 months)

Homesickness frequency was inversely correlated with previous sepamtion experience at boarding schools (r = p.58).

For females only, home- sickness frequency was inversely correlated with negative life eveiits (r = p.72).

Homesickness incidence and frequency varied by gender, being highest for girls.

Homesickness incidence and frequency were low- est for the two groups with previous separation experience as boarders or day pupils at this board- ing school.

Frequent homesickness was associated with a high discrepancy be- tween preferred demand and control and perceived dentand and control.

Cross-sectional compari- son between the newly arrived group and the experienced group sug- gested that acculturated adolescents living with

U.S. families decrease in homesickness over time, while adolescents living in Cambodian group care in Australia in- creased in homesickness over time.

Homesickness frequency was unrelated to pcrccivcd control over the decision to attend board- ing school, geographical distance, or the presence of a sibling at the school.

No age or gender differences in homesickness severity.

Relation between recent negative life events and homesickness unclear.

No age differences in homesickness severity.

None reported.

Table 2 (Continued)

Study Author(s)

Ward & Kennedy (1993)

Fisher & Hood (1987)

Carden & Feicht (1991)

Population and Environment

178 New Zealand second- ary school students liv- ing abroad for 1 or 2 years with host families in one of 23 different foreign countries (76% female; mean age = 17.4; range: 16-19 years)

66 first-year resident and home-based university students in the U.K. (about 35% female; mean age -17.9)

75 American and 69 Turk- ish first-year students at an American liberal arts college for women (100% female; mean age = 17.8; range: 16-

22)

Measure(s) of

Homesickness

2-77 weeks after arrzual. 12 cognitive and motiva- tional (but not emo- tional) items from the Dundee Relocation In- ventory (DRI; Fisher, 1989), each rated on a 4 point Likert scale

6 weeks after arrival:

5 week retrospective and current self-report of homesickness intensity on a 4 point Likert scale

Dundee Relocation Inven- tory (DRI; Fisher et al., 1987)

During the fol~rth week of the first academic se- m?ster:

Retrospective self-report of homesickness inten- sity during the initial 3 weeks at college, admin- istered as part of a 60 item I-iomesickness Questionnaire (HSQ; Carden & Feicht, 1991), rated on a 5 point Likert scale

Measure(s) of Risk and /or Protective Factor(s)

2-77 weeks after arrival.

12 item cultural identity questionnaire (Ward & Kennedy, 1992), rated on a 7 point Likert scale

7 item social contact ques- tionnaire, rated on a 4 point Likert scale 15 item locus of control questionnaire, adopted from Collins (1974), rated on a 5 point Likert scale

5-8 ZLIEP~Sprior to separa- tion:

Cognitive Failures Ques- tionnaire of absentmind- edness (CFQ; Broadbent et al., 1982)

Middlesex Hospital Ques- tionnaire of mental health (MHQ; Crown & Crisp, 1966)

Self-reported demo- graphics

During the fourth uieek of the first acad~niic se- ntes ter:

Self-report of demograph- ics and previous separa- tion experience

self-rePoit on the short form of the California Psychological Inven- tory (Burger, 1975)

Self-report on the adult form of the Eysenck Per- sonality Questionnaire (EPQ; Eysenck & Eysenck, 1975)

Main Finding(s),'

Homesickness intensity correlated with strong cultltral identity (r = .20) and external locus of con- trol (r = .25), and in- versely with quality of so- cial interaction with host nationals (r = -.28).

Students who reported some homesickness had higher preseparation lev- els of obsessionality, so- matic complaints, and deprpssion on the MHQ than the nonhomesick students.

Turkish students were more homesick than American students.

All students who reported high levels of homesick- ness were more deppndent on family advice, less socially skilled, and less socially mature than the students who re- ported low homesick- ness.

Turkish students who re- ported high levels of homesickness were less flexible than Turkish stu- dents who reported low homesickness.

Null Finding(s)

None reported.

Students who reported current homesickness did not have higher pre-separation levels of absent- niindedness on the CFQ than the nonhomesick students.

No gender differences in homesickness intensity.

Homesickness intensity was unrelated to geographic distance, previous s~paration ~xperi~nce, nurnber of tirnes fantily moved, family size, or number of friends at home or at school.

Fisher & Hood (1 988)

Takahashi & Majima (1994)

Fisher, Murray, & Frazer (1985)

Porritt & Taylor (1981)

196 first-year resident uni- versity students in the

U.K. (49% female; mean age = 18.5; age range: 17-24 years)

37 female, first-year, resi- dent students at a pri- vate, women's, Catho- lic, 4 year college in Japan (age range: 18-19 years)

100 first-year resident uni- versity students in the

U.K. (40% female; mean age = 18.6)

185 mostly resident first- year student nurses in Australia (95% female; 88% under age 20) con- sisting of 100 who were assessed after 4 months; 85 who were assessed after 10 months

6 weeks after arrlval: single-

question self-report of homesickness intensity, rated on a 4 point Likert scale

At the 2d, 9th, and 26th week of the separation: one dichotomous ques- tion about missing fam- ily, administered as part of an 11 item ad- justment scale

6 weeks after arrival: 1to 5 week retrospective self- report of homesickness frequency, severity, and episode duration, rated on a series of 3 point Likert scales

4 or 10 rnonths after arrlval: retrospective self-report on a single dichoto- mous question of home- sickness incidence

6 weeks after arrival:

Cognitive Failures Ques- tionnaire of absentmind- edness (CFQ; Broadbent et al., 1982)

Middlesex Hospital Ques- tionnaire (MHQ, Crown & Crisp, 1966)

Self-reported demograph- ics and previous separa- tion experience

At the 2d week of separa- tion: Affective Relation- ships Questionnaire (AlZQ; Takahashi, 1990) to measure relative so- cia1 dependence on fam- ily or peers

6 weeks after arrival:

Cognitive Failures Ques- tionnaire of absentmind- edness (CFQ; Broadbent et al., 1982)

Self-Esteem Questionnaire (Bachman & O'Malley, 1977)

Self-reported demograph- ics and recent life events

4 or 10 ntonths after arrlval:

Self-report on an 18 item dichotomous response questionnaire of grief re- actions, symptoms of homesickness, and con- tact with attachment figures

Self-reported demograph- ics and previous separa- tion experience

Students who reported some homesickness had less previous separation experience than the non- homesick students.

I-iomesick students had el- evated levels of anxiety, phobias, obsessive symp- toms, somatic complaints, and depression on the MHQ.

Homesick students had more absentmindedness than nonhomesick stu- dents on the CFQ.

The 14 women classified during the 2d week as 'yatriily-dominan t " on the ARQ reported more homesickness at the 9th and the 26th week than the 23 women classified as "agemate-dominant" on the ARQ.

Students who reported some incidence of home- sickness lived at greater geographic distance, were more absentminded, and reported less decision control than nonhome- sick students.

Incidence of homesickness symptoms decreased with the duration of sepa- ration (in the initial 6 weeks of the separa- tion).

lncidence of homesickness increased for students who recently separated from home prior to starting nursing school.

Students living at home reported fewer symptoms of homesickness.

No age or gender differences in homesickness intensity.

None reported

Homesickness incidence was unrelated to recent life events, the presence of a confiding relation- ship, and self-esteem.

No gender differences in homesickness fre- quency.

lncidence of homesickness symptoms was not asso- ciated with frequency or difficulty of contact zr~ith attachment figures, nor was it associated with urban versus rural background.

Table 2 (Continued)

Study Author($ McCann (1943)

Rose (1947, 1948)

Woulff (1975)

Population and
Environment

200 resident students at a public university in the

U.S. (50% female) (mean age and age range not reported)

149 female, first-year, resi- dent students at a pri- vate university in the

U.S. (mean age and age range not reported)

280 first-year resident uni- versity students in the

U.S. (61% female)
(mean age and age
range not reported)

Measure(s) of
Homesickness

Sontetinze durlng the school year: dichotomous self- report of homesickness incidence resulting in two groups: 100 stu- dents (50% female) who were or who recently had been homesick, and 100 students (50% fe- male) with no home- sickness

Between 1 and 6 weeks after arrival: dichotomous self-report of homesick- ness incidence resulting in 3 groups: Type A (never homesick, n = 66); Type B (one home- sick ep~sode, n = 45); Type C (chronically homesick, n = 38)

3 and 11 weeks after arrival: 15 item Woulff Home- sickness Scale (WHS; Woulff, 1975), rated on a 5 point Likert scale

Measure(s) of Risk and /or Protective Factor (s)

Concurrent with the assess- ntent of hotnesickness inci- dence: self-report on a 186 item questionnaire covering symptomatol- ogy, etiology, preven- tion, and treatment of homesickness.

Between 1 and 6 weeks after arrival: retrospective au- tobiographical reports of mobility history, fam- ily history, and adjust- ment, as measured by the Bell Adjustment In- ventory (Student Form, 1934)

2 months prior to the separa- tion:

196 item self-report ques- tionnaire measuring sep- aration attitudes and history, demographics, "family dependency," and confiding behavior

24 item self-report ques- tionnaire of future time perspective

20 item Zung Depression Scale (ZDS; Zung, 1965)

Main Finding(s)"

Homesickness incidence was associated with lit- tle previous separation Mperience, little positive comnulnication from horne, and little social involvement (for women) or athletic involuement (for men).

Interpretation of the auto- biographic reports and adjustments data sug- gested that Type A women had a family history of stability, secu- rity, and social opportu- nity; Type B women had a family history of stability and security, but few social opportuni- ties; Type C women had a family history of rejection and few social oppor- tunities.

Students who had re- ported high levels of positive and negative verbal disclosure to par- ents prior to the separa- tion were more likely to be homesick at week 3 than those who had re- ported low levels of ver- bal disclosure.

Homesick males reported slightly lower presepara- tion self-esteent than non- homesick males.

Null Finding(s)

Homesickness incidence was not associated with urban versus rural background, geographical dis- tance from home, fantily size or composition, aca- demic or religious back- ground, birth order, or previous separation experi- ence at residential sum- mer camp.

No age or gender differences in homesickness incidence.

Incidence of homesickness was unrelated to previous separation experience, either at boarding school or residential summer camp.

Homesickness intensity was not correlated with preseparation levels of self-reported depression.

Homesickness severity was unrelated to family size, birth order, geograph- ical distance from home, size of hometown, or number of parents living in the household.

No age or gender differences in homesickness intensity.

Brewin, Furnham, & Howes (1989)

Eurelings-Bontekoe, Vin- gerhoets, & Fontijn (1 994)

64 first-year resident uni- versity students in the

U.K. (about 66% female, mean age = 20; age range: 18-39 years)

738 male conscripts in the Dutch military con- sisting of:

131 veteran soldiers who reported homesickness (mean age = 20.6)

111 veteran soldiers who reported other psycho- logical disturbance (mean age = 20.6)

496 new conscripts with- out any army experi- ence yet

6 zrleeks after arrival:

Two self-report questions about expected home- sickness frequency, rated on a 4 point Likert scale

Two self-report questions about actual homesick- ness, rated on 3 and 4 point Likert scales

At the time of referral to the military psychology clinlc: self-report on nine di- chotomous questions about homesickness symptoms

6 uieeks after arrival:

Single-question self-report of the percent of stu- dents one would expect to be homesick

Dependency subscale (10 items) of the Depressive Experiences Question- naire (Blatt, D'Afflitti, & Quinlan, 1979)

Self-reported demo- graphics

At the time of referral to the military psychology clinic: retrospective self-report on:

Dutch Personality Inven- tory (DPI; Luteijn, Star- ren, & Van Dijk, 1985)

Dutch shortened version of the MMPI (MMPI-S; Luteijn & Kok, 1985)

Inventory for Interper- sonal Behavior (IIB; Van Dam-Baggen & Kraaimaat, 1987) to measure social skills and social anxiety

Questionnaire of Organi- zational Stress (QOS; Reiche & Dijkhuizen, 1980) to measure social support

Self-report of mobility his- tory and demographics

Homesickness intensity was correlated with esti rnated homesickness fre- quency (r = .43) and self-reported dependence (r = .31), and inversely with age (r = -.34).

Compared to psychiatric controls and new con- scripts: Homesick soldiers had less previous separation experience and fewer socializing experiences.

Homesick soldiers scored higher on the rigidity and lower on the dominance and sey-esteem scales of the DPI.

Homesick soldiers scored lower on the extroversion and higher on the sontatizing and negativism scales of the MMPI-S.

Both the homesick sol- diers and the psychiat- ric controls had more frequent histories of home- sickness, and were more likely to return homt7 than new conscripts.

Homesickness intensity was not correlated with geographic distance from home or previous separa- tion experience.

No gendt~rdifferences in homesickness incidence or intensity.

Homesick soldiers were not differentiated from psychiatric controls on the basis of social anxi- ety, as measured by the 11B.

Table 2 (Continued)

Study Author(s) Burt (1993)

Populat~on and
Env~ronment

93 first-year resident uni- versity students in Aus- tralia (72% female; mean age = 21.2; age range: 17-46 years)

49 Chinese students and visiting scholars at a

U.K. university (18% fe- male; mean age = 30.2)

Measure(s) of
Homesickness

Approxinzately 1-3 nzonths after arrival: self-report on the Dundee Reloca- tion Inventory, version 2-F (DRI; Fisher, 1989), a 26 item questionnaire rated on a 3 point Likert scale

2 weeks after arrioal: retrospective self-report on a 4 item questionnaire of initial homesickness in- tensity, rated on a 5 point Likert scale

2 months after arrival: self-report on single 5 point Likert-scale question: "Do you feel homesick now?" (Present Home- sickness Scale)

Measure(s) of Risk and /or Protect~ve Factor(s)

Approxzmately 7-3 months after arrival:

Self-reported demograph- ~cs,moblllty h~story, past academlc perfor- mance, and a single~temperce~ved control over separation ques- tion, rated on a 7 point Likert scale

Cognitive Failures Ques- tionnaire of absentmind- edness (Broadbent et al., 1982) Digit Span subtest of the Wechsler Adult Intelli- gence Scale (WAIS)

Vividness of Visual Imag- ery Questionnaire (Marks, 1973)

2 zueeks after arrival:

Sphere of Control Battery (Paulhus & Christie, 1981)

Cognitive Failures Ques- tionnaire of absentmind- edness-short form (Broadbent et al., 1982)

2 weeks and 2 months after arrzl~alMlddlesex Hos- pltal Quest~onna~re

of mental health (Crown & Cr~sp, 1966)

2 months after arrzval. 14 Item, 5 point Llkert scale questlonna~re of perceived academlc and soc~al demand

Mam Fmdmg(s)"

For males only, frequency of homesickness symp- toms was correlated with low perceived control over the separation (r = 49) and absentnundedness (r = .49).

Present homesickness in- tensity correlated with present perceived academic demand (r = .32), perceived suclal demand (r -.45), and initial homesickness (r = .57).

Null Fmdmg(s)

Homesickness intensity was not correlated with

age, recency of separation from home, presence of friends, ltrban versus ru- ral home environment, ability to form visual im- ages, or geographical dis- tance from home.

No age or gender differences in homesickness intensity.

Initial homesickness inten- sity was uncorrelated with absentmindedness, locus of control, or mental health.

No age or gender differences in homesickness intensity.

Hojat & Herman (1985) 1,593 male physicians Sometlnze durlr~g thejrst 20 Sometime during the first 20 Homes~ckness severity None reported
  practicing in the US., years after arrival In the years after arrival in the was correlated with so-  
  consisting of: US self-report on sin- U.S.: self-report on two c~allz~ngzulth same-  
  695 Iranian physicians gle questlon about social activity questions, natzonallty peers (r = 15)  
  (meart age = 43.1) homesickness, as part of as part of a 104 item and negatively corre-  
  898 Filipino physicians a 104 item question- questionnaire on adjust- lated with socializirlg  
  (mean age = 43.3) naire on adjustment ment and acculturation, with American colleagues  
    and acculturation, rated rated on a 4 point (r = -.21).  
    on a 4 point Likert scale Likert scale    
van Tilburg, Vingerhoets, 10 Dutch females with a 5, 3, and 1 days prlor to holi- 5, 3, and 7 days prlor to holi- Depression, anger, fatigue, Homesickness presence
van Heck, & Kirsch self-reported history of day trips zuith family: day trlp zuith family. and tensior~were higher and intensity were unre-
baum (1996) becoming severely self-report on the 9 item self-report on a shortened, on days when women lated to levels of sali
  homesick while on holi- Homesickness Decision Dutch translation of the reported homesickness vary cortisol (nmol/ 1).
  day trips with their fam- Tree, rated on a 6 point Profile of Mood States than on days of no  
  ilies (age range: 29-59 Likert scale (Eurelings- (POMS; McNair, homesickness; vigor was  
  years) Bontekoe et al., 1994) Loor, & Droppleman, higher on days of no  
    self-report on the 1item 1971) homesickness.  
    Subjective Homesick- salivary cortisol assays Homesickness intensity  
    ness Rating, rated on a once every 2 days during hol had an idiosyncratic lon  
    10 point Likert Scale iday trips if 1--3 weeks: gitudinal course. For  
    (van Tilburg et al., same as above some women, intensity  
    1996) 7, 3, and 5 days after re- was worse before leav-  
    once every 2 days during hol- turning honiefvoni holi- ing home; for others, it  
    iday trips of 1-3 weeks: day trip: same as above was most intense dur-  
    same as above   ing the holiday trip; for  
    1, 3, and 5 days after re-   still others, homesick-  
    turning homefrom ho11-   ness was equally in-  
    day trlps: same as above   tense prior to and dur-  

ing separation.
" Unless study authors explicitly identified symptom correlates of homesickness (e.g., depression, anxiety) as risk or protective factors, these correlates are not reported in this table.

916 Child Development

H2: Generative

.....................
..........

ATTITUDES ';

;. &CONTROL .................

HI: Amplifying

SEPARATION
EXPERIENCE

Figure 1 Comparison of (HI)-ATTITUDES AND CONTROL BS amplifiers of SEPARATION EXPECTATIONS-and(Hz)-ATTITUDES AND CONTROL as generators of SEPARATION EXPECTATIONS and HOMESICK DISPOSITION.

well as dissatisfaction with the novel environment (Carden & Feicht, 1991; Fisher et al., 1985). Further- more, negative interpersonal attitudes and low per- ceived control will directly predict negative emotion (Armsden et al., 1990; Weisz et al., 1993). Finally, con- ventional wisdom (e.g., Burrow, 1992; Doherty, 1940; van Wagenen, 1933) suggests that parental separa- tion anxiety may increase a child's "homesick dispo- sition"-enduring beliefs that strong feelings of homesickness are likely.

The aims of this study were (1)to measure a large but finite number of empirically and/or popularly supported risk and protective variables for homesick- ness; (2) to aggregate those variables into conceptu- ally sensible factors, particularly the link between interpersonal attitudes and perceived control; and

(3) to test alternative models of how these risk and protective factors predict homesickness and its se- quelae.

The study improved on previous homesickness re- search in several ways. First, it included extensive measurement of children's interpersonal attitudes and perceived control. Second, predictors, sequelae, and homesickness itself were measured longitudi- nally, permitting stronger causal interpretation of re- sults. Third, the study included factor analysis of pre- dictor variables, enabling conceptual and empirical integration of disparate risk and protector variables. Finally, structural equation modeling (SEM) with la- tent variables was used to compare the two hypothe- ses about the roles of interpersonal attitudes and per- ceived control in the etiology of homesickness.

METHOD

Participants

Participants were all campers at a residential boys' sports camp. Several months prior to the start of camp, all children registered for the first month of the 1994 summer were asked to take part, with their parents, in research on adjustment to summer camp. Of 453 registered children, 317 (70%) consented, along with their parents, to participate. Ethnic minor- ities represented 7% of the sample. Two boys were eliminated from the sample because their question- naires showed evidence of haphazard responding. Some 22 additional participants and their parents did not complete precamp measures. Thus, the final Nof complete cases was 293.

The sample was intended to represent a normal cross-section of English-speaking, nondisabled, non- clinic-referred, preadolescent and adolescent boys who attend residential summer camps in the United States. Camp records indicated that most children were from suburban and rural neighborhoods in the northeast United States. The modal child's parents were technicians, semiprofessionals, or small busi- ness owners, but parents' occupational status ranged from executives and major professionals to manual and service workers to unemployed. Scholarship funding permitted admission to this camp without regard to financial means. There were no significant differences between participants and nonparticipants in age (M = 12.5 years versus 12.5), distance to camp from home (145 miles versus 206), parents' occupa- tional status (M = 63 on the Hollingshead [I9751 scale versus 60), or number of previous summers spent at this camp (M = 2.6 versus 2.5) or any camp (M = 3.2 versus 3.1). According to these demographic indica- tors, the participant group was representative of the population of campers attending this camp.

Living quarters for the 260 boys who attended this camp during a given 2 week session are divided into five equal-sized age divisions, each in its own loca- tion on the 177 acre wooded site. Divisions com- prised six cabins, each with room for eight or nine children and one or two cabin leaders. Boys attended this camp for one or two 2 week sessions. Data were collected daily during the first 2 weeks of each boy's stay. Summer camp was an appropriate setting in which to study homesickness and adjustment to sep- aration because the separation from primary caregiv- ers was ethical yet prolonged and discrete. Boys' only contact with home was letter writing, except in rare emergencies. Male cabin leaders functioned as surro- gate caregivers. They were quite experienced, rang- ing in age from 17 to 23 years (M = 19 years; SD = 1.1), and having worked with children for between three and seven summers at this camp (M = 4.06 summers; SD = 1.2).

Procedure

Preseparation. Data collection began by mail 3 months prior to children's arrival at camp. Precamp measures included four questionnaires designed to measure parents' and children's preseparation emo- tions and cognitions. To measure parental separation anxiety, parents completed a version of the Maternal Separation Anxiety Scale (MSAS; Hock, McBride, & Gnezda, 1989). Originally designed for mothers to complete about their infants, the MSAS was adapted and validated by Capps, Sigman, Sena, Henker, and Whalen (1996) for use by mothers or fathers of school-age children and adolescents. Capps et al. re- ported a high 4 week test-retest reliability for the MSAS separation anxiety subscale, r(15) = .75. In the present study, the MSAS separation anxiety subscale had high internal consistency, as measured by coef- ficient alpha (a = .89/21 items).

To measure boys' preseparation levels of distress, each boy completed the Rate Your Day-Revised mood checklist (RYDR; Thurber, 1995, 1998), which

Thurber and Sigman 917

has three subscales: homesickness, negative emotion, and positive emotion. Although usually adminis- tered as a daily questionnaire, the RYDR that boys completed prior to coming to camp asked them to report on how they had been feeling "during the last 2 weeks." The RYDR is reproduced in Appendix A. Reliability and validity are reported below.

Boys also completed two mini-questionnaires de- signed to measure their expectations of homesickness and their expectations of summer camp. Both had high internal consistency (expectations of homesick- ness scale a = .78/3 items; expectations of summer camp scale a = .85 /5 items). Precamp measures were returned by boys and their parents between 12 and 6 weeks prior to the start of camp.

Daily measurement. Homesickness at camp was measured daily on the RYDR. The homesickness sub- scale is a sum of 11 point Likert scale endorsements on each of three items: 1felt homesick, 1missed myfamily, and 1wanted to be home. The RYDR had high inter- nal consistency (negative emotion scale a = .96/6 items; positive emotion scale a = .92/6 items; home- sickness scale a = .94/3 items) and has been vali- dated with clinical interviews (Thurber, 1996), stan- dardized self-reports of negative emotion, and observer reports of homesickness (Thurber, 1995). Al- though boys completed a RYDR before coming to camp and 14 consecutive RYDRs during camp, only the RYDRs completed on days 4-9 were used to in- dex homesickness for this study. Using the RYDRs before day 4 (when the various predictor variables were measured) or after day 9 (when the various se- quelae variables were measured) would naturally have muddled longitudinal causal modeling. In addi- tion to a daily RYDR, boys completed a variety of other questionnaires on certain days.

Day 1.To measure interpersonal attitudes, partici- pants completed two measures: one about their care- givers and one about general self-perceptions. The first, the Children's Expectations of Social Behavior Questionnaire (CESBQ; Rudolph, Hammen, & Burge, 1995), focused on perceptions of caregivers. Children were read each of 15 parent-child interaction scenar- ios and asked to choose how their caregiver would react from among three options describing accepting, dismissing, and rejecting behavior. The higher the overall score, the more accepting the child perceives his primary caregivers to be; the lower the score, the more dismissing and rejecting he or she perceives them to be. The CESBQ had high internal consistency for this sample (a = .77/ 15 items).

The second measure, the Self-Report Attachment Style Prototypes (SRASP; Bartholomew & Horowitz, 1991), focused on participants' self-perceptions of

918 Child Development

their relationship style. The SRASP was adapted by Thurber, Bombar, and Sigman (1994) for use with children and adolescents by slightly simplifying the wording. The adaptation is reprinted in Appendix B. This adapted SRASP asked participants to make Likert scale ratings of how much each of four short paragraphs was an accurate description of them. The paragraphs, each read aloud twice by cabin leaders, described four empirically supported relationship styles: secure, preoccupied, fearful, and dismissing. A composite score of self-perceived relationship security was derived by subtracting participants' endorsements of the preoccupied, fearful, and dis- missing paragraphs from their endorsement of the paragraph describing a secure style. Thus, the higher the composite score on the SRASP, the more the child perceives secure interpersonal attachment.

Day 2. To measure attitudes about separation, par- ticipants completed one experimental questionnaire called About Me. This measure included a subscale of decision control (i.e., perceived control over the de- cision to attend summer camp) (a= .79/ 6 items) and an in-camp subscale of expectations of summer camp (a = 37/12 items). Participants also reported the number of previous overnight separation experiences of at least 1 week's duration. Exact chronological age and years of attendance at this summer camp were extracted from camp records. Finally, participants were asked to rate their perceived distance from home on an 11 point Likert scale from not far at all to very far.

Day 3. To measure perceived control in domains other than decision control, participants completed the Perceived Control Questionnaire (PCQ; Weisz, Proffitt, & Sweeney, 1991). The original PCQ is com- posed of three 8 item scales assessing perceived con- trol in academic, social, and conduct domains (e.g., "I can get good grades if I really try"). For the pur- poses of this study, two identically constructed 8 item scales were added to assess children's perceived con- trol of caregiving (e.g., "When I'm upset, I can get someone to help me") and their own emotions ("I can get in a good mood if I really try"). All five scales had high internal consistency (academic a = 34; so- cial a = 33; conduct a = 32; caregiving a = 34; emo- tions a = 24). The five scales were summed to pro- duce an aggregate score for perceived control (a = .93 /40 items).

To measure early experiences, participants were asked to identify and rate the best and worst things that had happened to them since arriving at camp. Many boys were able to identify only the best thing they had experienced. Therefore, only their rating of their most positive initial experience, on an 11 point Likert scale from it made me feel terrible to it made me feel excellent, was used to index early experience.

Day 10. To measure depressive and anxious symp- toms, participants completed the Children's Depres- sion Inventory (CDI; Kovacs, 1980) and the Revised Children's Manifest Anxiety Scale (RCMAS; Rey- nolds & Richmond, 1978), both of which have sever- ity cutoff scores based on large standardization sam- ples. The questionnaires were administered in their entirety, with the exception of the one question about suicide on the CDI. Both measures have good validity (e.g., Lee, Piersel, Friedlander, & Collamer, 1988; Saylor, Finch, Baskin, Furey, & Kelly, 1984) and proved to have high internal consistency in the pres- ent sample (CDI a = 39/27 items; RCMAS a = .89/ 28 items).

Day 14. To measure boys' satisfaction with their separation experience, participants completed a ques- tionnaire (My Time at Camp) on the day before their departure. This measure asked boys to rate the over- all quality of their stay on an 11 point Likert scale from terrible to excellent and to complete scales of so- cial and environmental satisfaction. Examples from the social satisfaction scale include, "In general, my cabin leader was . . ." and "In general, the kids in my cabin were . . . ." Examples from the environmental satisfaction scale include, "How did you like living in a cabin in the woods?" and "How much did bugs or poison ivy bother you?" Both scales had high in- ternal consistency (social a = .84/8 items; environ- mental a = .73/9 items). High variability in ratings suggested that the sample as a whole was not biased toward positive reports as a function of imminent re- union with parents.

To measure internalizing behavior, cabin leaders completed a Child Behavior Checklist (CBCL; Achen- bach & Edelbrock, 1991) for each participant in their cabins. Previous phenomenological research suggested that the three narrow-band internalizing sub- scales on the CBCL-withdrawn behavior, anxious / depressed behavior, and somatic complaints-were most consistently related to homesickness in boys (Thurber, 1995, 1998). However, to maximize objec- tive reporting, cabin leaders completed the CBCL in its entirety, except for the scale including sexual problems.

Data integrity. Boys completed questionnaires each evening in the privacy of their own bunk beds. Senior cabin leaders assisted in the reading and administra- tion of each questionnaire. The PI verified complete- ness of the questionnaires each evening. On average, 1% of the questionnaires was incomplete or blank. Boys who had not completed questionnaires, either due to carelessness or illness, were asked to complete them the following morning. Further procedural de- tails are reported in Thurber (1995, 1998).

Summary of predictor and outcome variables. To sum- marize, this study included the following predictor variables: (1) exact chronological age; (2) number of summers at camp; (3) number of previous separa- tions; (4) perceived distance from home; (5) presepa- ration negative emotion; (6) preseparation expecta- tions of homesickness; (7) preseparation expectations of summer camp; (8) in-camp expectations of sum- mer camp; (9) initial positive experience; (10) per- ceptions of caregiver acceptance; (11) perceived relationship security; (12) perceived decision con- trol; (13) multidimensional perceived control; and

(14) parental separation anxiety.

In addition to measuring homesickness on a daily basis, using the RYDR, the following sets of outcome variables were measured: (1) negative emotion (de- pressive symptoms and anxious symptoms); (2) in- ternalizing problems (withdrawn behavior, anxious / depressed behavior, and somatic complaints); and

(3) satisfaction (overall, social, and environmental).

Consolidation of Predictors

Whereas previous research and conventional wis- dom have proposed a wide variety of risk and protec- tive factors for homesickness, no empirical study has suggested how these predictors might cluster. There- fore, the first step in developing a plausible patho- genic model of homesickness was to factor analyze the 14 predictor variables that we measured. A prin- cipal components extraction with varimax rotation produced five easily interpretable factors with eigen- values greater than 1.0. This five factor solution ac- counted for 62.8% of the variance in predictors. The rotated factor pattern matrix is presented in Table 3. As predicted, measures of interpersonal attitudes and perceived control loaded on the same fa~tor.~ Following convention, capital letters will be used to designate factors; lowercase letters to designate mea- sured variables.

Hypothesized Relations among Predictors

Research reviewed above suggested ways that these factors might be related. Little previous SEPARATION EXPERIENCE may lead directly to boys' self-report

2. A second factor analysis, using the original version of the PCQ (i.e., without the caregiving and emotion subscales), pro- duced an identical factor structure. Therefore, the empirical link between interpersonal attitudes and perceived control in this case is not an artifact of the new caregiving and emotion control subscales of the PCQ.

Thurber and Sigman 919

of HOMESICKNESS and indirectly to HOMESICKNESS through the development of a preseparation HOMESICK DISPOSITION. AS previously stated, parental sepa- ration anxiety may augment a HOMESICK DISPOSITION. Next, according to (HI), negative SEPARATION EXPECTATIONS and weak ATTITUDES AND CONTROL may am- plify the negative anticipation inherent in boys' HOMESICK DISPOSITION. The right side of Figure 1 shows the relations among predictors according to (H1)-the amplifying model. By contrast, according to (H2), ATTITUDES AND CONTROL predict HOMESICK DISPOSITION and SEPARATION EXPECTATIONS,both of which, in turn, predict HOMESICKNESS.

The left side of Figure 1 shows the relations among predictors ac- cording to (H2)-the generative model.

Consolidation of Outcomes

The outcome variables were grouped into a priori factors based on previous research (Achenbach & Edelbrock, 1991; Ollendick, Yule, & Ollier, 1991; Saylor, Finch, Spirito, & Bennett, 1984; Thurber, 1998). As noted above, the HOMESICKNESS factor was composed of boys' mean ratings on three RYDR vari- ables (feeling homesick, missing family, and wanting to be home) for the six consecutive days (4-9) be- tween the measurement of predictors and sequelae. Next, a NEGATIVE EMOTION factor was composed of the two standardized measures of depressive and anx- ious symptoms, the CDI and the RCMAS, adminis- tered on day 10. An INTERNALIZING BEHAVIOR factor was composed of T-scores from the three internaliz- ing narrow-band clinical scales on the CBCL, admin- istered on day 14. Finally, a SATISFACTION factor was composed of the three satisfaction subscales-over- all, social, and environmental-also administered on day 14.

Hypothesized Relations among Predictors and Outcomes

Previous research suggested testable relations among the predictors and sequelae of homesickness. First, congruent with either (HI) or (H2) are the es- tablished relations between perceived control and negative emotion (e.g., Weisz et al., 1993), and inter- personal attitudes and negative emotion (e.g., Pa- pini & Roggman, 1992). Therefore, a direct path from ATTITUDES AND CONTROL to NEGATIVE EMOTION was hy- pothesized. Second, HOMESICKNESS should predict the three outcome factors of NEGATIVE EMOTION,

SATISFACTION, and INTERNALIZING BEHAVIOR. Naturally, NEGATIVE EMOTION itself was hypothesized to predict INTERNALIZING BEHAVIOR.

Congruent with Fisher's theories,

920 Child Development Table 3 Factor Pattern Matrix of Homesickness Predictor Variables

Measured Predictor Variable

Chronological age Previous summers at camp Previous separations Perceived distance Preseparation negative emotion Preseparation expectations of homesickness Preseparation expectations of summer camp In-camp expectations of summer camp Initial positive experience Caregiver acceptance Perceived relationship security Perceived decision control Perceived multidimension control Parental separation anxiety

Factor
PARENTAL
SEPARATION HOMESICK ATTIWDES SEPARATION SEPARATION
EXPERIENCE DlSPOSlTION AND CONTROL EXPECTATIONS ANXIETY

Note: Strongest loadings appear in italic to illustrate final factor composition

HOMESICK DISPOSITION, SEPARATION EXPECTATIONS, NEGATIVE EMOTION,

and HOMESICKNESS were hypothesized to predict boys' SATIsFACnoN with their separation experience.

RESULTS

Clearly, with its factor analysis of predictors and comparison of alternate models, the following SEM analysis was both exploratory and confirmatory. Moreover, with its 25 measured variables and a subject-to-parameter ratio of 5:1, this sample of 293 boys fell short of the recommended 10:l subject-to- parameter ratio for nonnormal data (Bentler, 1995). Although this limitation promised to constrain any model's goodness-of-fit, SEM had a clear advantage over ANOVAs and multiple regressions because it permitted simultaneous estimation of many parame- ters, with the variance of each effect controlled for all other effects in the model.

Parameters of two alternate models, including the measurement of latent variables, were tested with EQS (Bentler, 1995), an SEM program. In an attempt to match the observed variances and covariances among measured variables, SEM estimates the factor loadings between latent variables and their indica- tors, the regression coefficients among factors, and the variances and covariances of independent vari- ables. Because most of the measured variables were not normally distributed, the models were evaluated using maximum likelihood path estimation with ro- bust test statistics and standard errors (Hu, Bentler, & Kano, 1992). Table 4 is the correlation matrix of mea- sured variables3

Several goodness-of-fit indexes were used to eval- uate the two alternate models. First, Bentler's (1990) Robust Comparative Fit Index (RCFI), which is normed from 0to 1, measures the relative adequacy of a model on a continuum of models ranging from a null model (all variables are unrelated) to a saturated model (all variables are related). The RCFI is robust against nonnormal data and minimizes Type I error when test statistics are not chi-square distributed. RCFI values 2.90 are desirable and suggest that a model is adequately reproducing the data (Bentler, 1995). A second fit index is the magnitude and distri- bution of the residuals produced by comparing the original covariance matrix of measured variables to the covariance matrix reproduced by a hypothesized model. The smaller the absolute value of the largest off-diagonal residual, and the more symmetric and centered around 0the residuals are, the better the model reproduces the data (Bentler, 1995). Finally, a third fit index is the Satorra-Bentler chi-square statis- tic (Satorra & Bentler, 1988). This statistic tests the hypothesized model against a saturated model. Sig-

3. Where necessary, the variance of variables was scaled up or down by a factor of 10 to make them roughly equal. This pro- cess can help the determinant of the covariance matrix converge at a minimum, but does not affect the sign or magnitude of re- gression coefficients.

Table 4 Correlation Matrix of Measured Variables (with Standard Deviations-nd Measure Names)

  1. Chronological age (SD = 6.26) (About Me)
  2. Previous summers at camp (SD = 5.14) (About Me)
  3. Previous separations (SD = 7.05) (About Me)
  4. Parental separation anxiety (SD = 9.69) (MSAS revised)
  5. Preseparation expectations of HS (SD = 6.84) (RYDR)
  6. Preseparation negative emotion (SD = 9.33) (RYDR)
  7. Perceived distance from home (SD = 5.49) (About Me)
  8. Preseparation expectations of camp (SD = 7.25) (RYDR)
  9. Initial positive experience (SD = 5.12) (RYDR)
  10. In-camp expectations of camp (SD = 5.76) (About Me)
  11. Perceived relationship security (SD = 7.06) (SRASP revised)
  12. Caregiver acceptance (SD = 3.50) (CESBQ)
  13. Perceived decision control (SD = 4.98) (About Me)
  14. Multidimensional perceived control (SD = 5.35) (PCQ revised)
  15. Feeling homesick (SD = 9.18) (RYDR day 4-9)
  16. Missing family (SD = 10.8) (RYDR day 4-9)
  17. Wanting to be home (SD = 9.85) (RYDR day 4-9)
  18. Depressive symptoms (SD = 6.22) (CDI)
  19. Anxious symptoms (SD = 6.15) (RCMAS)
  20. Withdrawn behavior (SD = 5.58) (CBCL T-score)
  21. Somatic complaints (SD = 3.90) (CBCL T-score)
  22. Anxious/depressed behavior (SD = 3.61) (CBCL T-score)
  23. Overall satisfaction (SD = 5.92) (My Time . . .)
  24. Social satisfaction (SD = 10.30) (My Time . . .)
  25. Environmental satisfaction (SD = 11.79) (My Time . . .)
  1. Preseparation negative emotion
  2. Perceived distance from home
  3. Preseparation expectations of camp
  4. Initial positive experience
  5. In-camp expectations of camp
  6. Perceived relationship security
  7. Caregiver acceptance
  8. Perceived decision control
  9. Multidimensional perceived control
  10. Feeling homesick
  11. Missing family
  12. Wanting to be home
  13. Depressive symptoms
  14. Anxious symptoms
  15. Withdrawn behavior
  16. Somatic complaints
  17. Anxious/ depressed behavior
  18. Overall satisfaction
  19. Social satisfaction
  20. Environmental satisfaction
  1. Perceived relationship security
  2. Caregiver acceptance
  3. Perceived decision control
  4. Multidimensional perceived control
  5. Feeling homesick
  6. Missing family
  7. Wanting to be home
  8. Depressive symptoms
  9. Anxious symptoms

922 Child Development Table 4 (Continued)

  1. Withdrawn behavior
  2. Somatic complaints
  3. Anxious/ depressed behavior
  4. Overall satisfaction
  5. Social satisfaction
  6. Environmental satisfaction
  1. Missing family
  2. Wanting to be home
    1. ~e~ressive
    2. symptoms
  3. Anxious symptoms
  4. Withdrawn behavior
  5. Somatic complaints
  6. Anxious /depressed behavior
  7. Overall satisfaction
  8. Social satisfaction
  9. Environmental satisfaction
  1. Somatic complaints
  2. Anxious/ depressed behavior
  3. Overall satisfaction
  4. Social satisfaction
  5. Environmental satisfaction

1 2 3 4 5

"Standard deviations and correlations were computed after some variables were scaled up or down by a factor of 10 to equalize variance

for EQS modeling.

* p < .05; ** p < .01; *** p < ,005; **** p < ,001.

nificant differences indicate that the model is not per- fectly reproducing the data, that is, that there may be additional variables or additional paths in the model that were not measured or estimated a priori. In fairly large samples, the chi-square index is sensitive to even trivial differences between the data and the model.

Evaluation of the Amplifying Hypothesis (HI)

The amplifying version of the model had an RCFI of .90. The absolute value of the largest off-diagonal residual was .27, and the residuals were symmetric and centered around 0. The Satorra-Bentler chi- square was significant, x2(261, N = 293) = 487.9, p < .001, suggesting some differences between this model and the saturated model. Using robust estimation of standard errors, all but one of the hypothesized paths among latent factors were in the predicted direction and significant at p < .05or less. The path from HOMESICKNESS to NEGATIVE EMOTION was not statistically sig- nificant. Figure 2 depicts the amplifying model, in- cluding the standardized parameter estimates and the significance levels for the paths among latent fac- tors.

A Langrangian Multiplier (LM) test was used to suggest paths that may have been added post hoc to improve model fit. The test suggested four theoreti- cally plausible covariances between the error terms of two measured variables. Conceptually, correlated error terms indicate relations between two measured variables' specific variance (i.e., that part of the vari- ance unrelated to the factor that indicates the vari- able). Two of the suggested covariances were be- tween anxiety (RCMAS) and both interpersonal attitudes (SRASP) and preseparation negative emo- tion (RYDR). The second two suggested covariances were between Previous Summers at Camp and both In-camp Expectations of Camp and Environmental Satisfaction. The addition of these paths could be the- oretically justified, but model fit improved only slightly (Robust CFI = .92; absolute value of the

Thurber and Sigman 923

*.64****

D2-.90 R2=.18

HOMESICKNESS

3 MONTHS PRE-SEPARATION TO DAY 3 DAY 4 TO DAY 9 DAY 10 DAY 14 Figure 2 EQS path diagram of the amplifying model of homesickness in boys. Interpersonal attitudes and perceived control

amplify homesickness and negative emotions.

largest standardized off-diagonal residual = .25; Satorra-Bentler x2 (257, N = 293) = 433.0, p < .001). Tweaking the model did not dramatically change the standardized path coefficients' magnitude, sign, or significance.

Evaluation of the Generative Hypothesis (H2)

The generative version of the model had an RCFI of 29. The absolute value of the largest off-diagonal residual was .26, and the residuals were symmetric and centered around 0. The Satorra-Bentler chi- square was significant, x2 (260, N = 293) = 503.4, p < .001. Using robust estimation of standard errors, all but three of the hypothesized paths among latent factors were in the predicted direction and significant at p < .05 or less. The path from Parental Separation Anxiety to HOMESICK DISPOSITION was not statistically significant, nor was the path from SEPARATION EXPECTATIONS to HOMESICKNESS. The parameter change be- tween HOMESICK DISPOSITION and SATISFACTION is dis- cussed below, in the section on effect decomposition. Figure 3 depicts the generative model, including the standardized parameter estimates and the signifi- cance levels for the paths among latent factors.

With the addition of the four correlated error vari- ances suggested by the LM test described above, the robust CFI of the generative version of the a priori model was improved to .91. The absolute value of the largest standardized off-diagonal residual was un- changed at .26, and the Satorra-Bentler chi-square de- creased slightly, x2(256, N = 293) = 447.5.0,p < .001. As with the amplifying version of the model, none of the standardized path coefficients changed dra- matically in magnitude, sign, or significance. Addi- tional LM tests suggested no other theoretically plau- sible paths among the latent factors. Overall, the LM tests were instructive indicators of shared variance, but improved fit only slightly in both models. There- fore, the remainder of this section focuses on the two unmodified versions of the model.

Measurement model. In both versions of the a priori model, latent factors reliably indicated measured variables. Standardized path estimates from factors to variables, summarized in Table 5, were nearly equivalent in the amplifying and generative models.

Effect decomposition. Direct effects in the amplifying and generative models are indicated in Figures 2 and 3, respectively. As noted above, a benefit of SEM is that the magnitude of each standardized direct effect

924 Child Development

D3-.65 R2=.58 SEPARATION

.55 R2=.70---I37

D2-.82 R2=.23

3 MONTHS PRE-SEPARATION TO DAY 3 DAY 4 TO DAY 9 DAY 10 DAY 14 Figure 3 EQS path diagram of the generative model of homesickness in boys. Interpersonal attitudes and perceived control
generate a homesick disposition and separation expectations.

is adjusted simultaneously for all other effects in the model. Indirect effects, computed by multiplying all of the mediating paths that link one factor to another, are reported in Table 6. Total effects can be calculated simply by adding direct and indirect effects. For ex- ample, in the amplifying version of the model, the total effect of SEPARATION EXPERIENCE on HOMESICKNESS is -.46, which is -.I9 (direct effect, from Figure 2) plus -.27 (indirect effect, from Table 6).

Effect decomposition explains the one counterin- tuitive valence in both versions of the model: the ef- fect of HOMESICK DISPOSITION on SATISFACTION. The re- lation was hypothesized to be negative, but the direct effect parameters were positive in both versions of the model. This statistical anomaly occurred because most of the negative effect of HOMESICK DISPOSITION on SATISFACTION is accounted for by indirect effects, through SEPARATION EXPECTATIONS and HOMESICKNESS. For example, in the generative model, the indirect ef- fects of HOMESICK DISPOSITION on SATISFACTION total -.98. Added to the .81 direct effect, the net total effect of HOMESICK DISPOSITION on sAnsFAcnoN is negative (-.I$), as predicted. The overwhelming strength of the indirect effects of HOMESICK DISPOSITION on SATISFACTION also explains why the direct effect is not sig- nificant in the generative version of the model.

Error variance and squared multiple correlation coeffi- cients. The robust CFI and chi-square values obtained above suggest that the alternate versions of the model reproduced the data adequately, but not perfectly. Therefore, it is particularly useful to examine the er- ror variances associated with the latent factors, as the models predict them. Squared multiple correlation coef- ficients (R2) can then be derived by subtracting the square of these error variances from 1. R2 can be inter- preted as the percent of a factor's variance accounted for by its predictors. R2 values are presented in Fig- ures 2 and 3 alongside their corresponding error vari- ances. Both models account for 70% of the variance in NEGATIVE EMOTION and 21% of the variance in INTERNALIZING BEHAVIOR.

However, the generative model accounted for greater variance in HOMESICK DISPOSITION (23% versus I$%), SEPARATION EXPECTATIONS (58% versus 41%), HOMESICKNESS (69% versus 65%), and SATISFACTION (70% versus 66%).

DISCUSSION

A wide variety of personal, circumstantial, and envi- ronmental variables has been hypothesized to predict homesickness. Researchers have measured many of these variables in different populations, with mixed

Thurber and Sigman 925

Table 5 Robust Standardized Parameter Estimates: Measurement Paths from Latent Factors to Measured Variables

Loading

Latent Factor HI HZ Measured Variable

SEPARATION EXPERIENCE .68" .67~hronological age

.89 .90 Previous summers at camp

.63 .63 Previous separations

HOMESICK DISPOSITION .48 .46 Perceived distance

.32 .32 Preseparation negative emotion

.82" .76Vreseparation expectations of homesickness

POSITIVE EXPECTATIONS .65 .64 Preseparation expectations of summer camp

.96\96"n-camp expectations of summer camp

.31 .30 Positive initial experience

ATTITUDES AND CONTROL -.33 -.35 Caregiver acceptance

.43" .43Verceived relationship security

.39 .38 Perceived decision control

.77 .77 Perceived multidimensional control

HOMESICKNESS .94" .93Veeling homesick

.90 .90 Missing family

.94 .93 Wanting to be home

NEGATIVE EMOTION .94" .93"epressive symptoms (CDI)

.74 .74 Anxious symptoms (RCMAS)

INTERNALIZING BEHAVIOR .87 .87 Withdrawn behavior

.31 .31 Somatic complaints

.78" .78" Anxious/depressed behavior

SATISFACTION .81\81~erall quality

.79 .79 Social satisfaction

.82 .82 Environmental satisfaction

Note: All parameter estimates significant at p < ,001, except Somatic complaints, p <
,005, and those parameters marked with a superscript "a."
To fix the scale of their respective latent factors, these parameters were fixed at 1
during estimation; therefore, no standard error is available to calculate parameter sig-
nificance.

results (see Table 2). Some variables, such as little nificantly correlated with the following predictors: previous separation experience, have emerged as (1) young chronological age; (2) few previous sum- fairly consistent predictors of homesickness. Other mers at camp; (3) few previous separations; (4) high variables, such as decision control, are fairly inconsis- parental separation anxiety; (5) preseparation expec- tent predictors. Researchers have suggested at least tations of strong feelings of homesickness; (6) presep- 17 theories to explain correlates of homesickness and aration negative emotion; (7) great perceived dis- group differences between homesick and nonhome- tanceqrom home; (8) negative preseparation sick people (see Table 1). Yet it is not surprising that expectations of camp; (9) few initial positive experi- out of the small, mostly cross-sectional and retrospec- ences at camp; (10) negative in-camp expectations of tive empirical literature, no data-based models of camp; (11) insecure interpersonal attitudes; (12) low childhood homesickness have been developed or perceived decision control; and (13) low multidimen- tested. As a first step in generating such a model, the sional perceived control. Homesickness also corre- present study longitudinally measured a variety of lated with the following sequelae: (14) depressive hypothesized predictors and sequelae of homesick- symptoms; (15) anxious symptoms; (16) withdrawn ness. behavior; (17) anxious / depressed behavior; (18) low

The correlations presented in Table 4 alternately overall satisfaction; (19) low social satisfaction; and confirmed and contradicted the previous research (20) low environmental satisfaction. Homesickness compiled in Table 2. Limited previous experience did not correlate with the predictor of caregiver ac- and a homesick disposition, fueled by insecure inter- personal attitudes and low perceived control,

4. Actual linear geographic distance between home town and emerged as key predictors of homesickness. Specifi- summer camp was uncorrelated with mean self-reported home- cally, self-reported homesickness intensity was sig- sickness intensity, ~(292) = -.02, p = .66.

926 Child Development

Table 6 Robust Standardized Parameter Estimates: Indirect (Mediated) Effects among Elements of the Amplifying and Generative Models

Latent or Measured Dependent Variable

POSITIVE EXPECTATIONS

ATTITUDES AND CONTROL

HOMESICKNESS

NEGATIVE EMOTION

INTERNALIZING BEHAVIOR

SATISFACTION

Indirect Effect
Latent or Measured
HI H2 Independent Variable

-.I1 .25

-.40

-.07 .15 .06

-.I0 .12 -.27

.38 -.49 -.02

.07 -.I7

.29 -.49 -.02

.03 .07 -.I7 -.86 .17 .27 -.02 .06 .23

ceptance, as measured by the CESBQ, or with the se- quela of somatic complaints, as measured by the CBCL.

The second step in generating a model for child- hood homesickness was to factor analyze the pre- dictors of homesickness. A five factor solution, which accounted for nearly 63% of the variance in pre- dictors, suggested that four latent variables-ATTI- TUDES AND CONTROL,SEPARATION EXPERIENCE, HOMESICK DISPOSITION, and SEPARATION EXPECTATIONS-along with the measured variable of Parental Separation Anxiety, were a parsimonious, theoretically justifi- able grouping. The empirical link between measures of interpersonal attitudes and perceived control con- firmed an important theoretical link between the two constructs. Based on prior empirical findings, sequelae were grouped into factors labeled NEGATIVE EMOTION, and SATISFACTION.

INTERNALIZING BEHAVIOR,

Finally, two alternate models, each with some em- pirical and theoretical support, were tested: one with ATTITUDES AND CONTROL as a final amplifier in the path of the other predictors; another with ATTITUDES AND

Parental separation anxiety

SEPARATION EXPERIENCE ATTITUDES AND CONTROL HOMESICK DISPOSITION

Parental separation anxiety

SEPARATION EXPERIENCE HOMESICK DISPOSITION POSITIVE EXPECTATIONS

Parental separation anxiety

SEPARATION EXPERIENCE ATTITUDES AND CONTROL HOMESICK DISPOSITION POSITIVE EXPECTATIONS ATTITUDES AND CONTROL

Parental separation anxiety

SEPARATION EXPERIENCE HOMESICK DISPOSITION POSITIVE EXPECTATIONS ATTITUDES AND CONTROL HOMESICKNESS

Parental separation anxiety

SEPARATION EXPERlENCE HOMESICK DISPOSITION POSITIVE EXPECTATIONS ATTITUDES AND CONTROL HOMESICKNESS

Parental separation anxiety

SEPARATION EXPERIENCE

CONTROL as a predictor of HOMESICK DISPOSITION and POSITIVE EXPECTATIONS.

Structural equation modeling with EQS suggested that the two models adequately reproduced the data and were equally plausible as systems of equations. However, the hypothesized path from ATTITUDES AND CONTROL to HOMESICKNESS was rather weak to support strongly the amplifying model (HI). By comparison, the generative model (H2)accounted for more variance in four key factors, including HOMESICKNESS. The generative model also predicted a particularly strong relation between HOMESICK DISPOSITION and HOMESICKNESS.

Most impor- tant, if concepts of relationships begin developing in infancy, then the generative model makes more theo- retical sense than the amplifying model. Attachment theory in particular suggests that infant and early childhood experiences with care solicitation and con- trol exertion generate childhood beliefs about how tolerable separations are. It follows that interpersonal attitudes and perceived control influence older chil- dren and adolescents' beliefs about whether separa- tions induce homesickness (HOMESICK DISPOSITION) and whether separations can be pleasurable (SEPARATION EXPECTATIONS).

In the generative model, the ATTITUDES AND CONTROL factor has a broad predictive value. According to this version of the model, an archetypal homesick boy is young and has little prior separation experi- ence, at camp or elsewhere. In addition, negative in- terpersonal attitudes and low perceived control com- bine to augment his preseparation feelings of sadness, nervousness, and expectations of homesick- ness. He is unsure of his power to control the out- come of important situations, including times when he might need assistance from others. These feelings also diminish his positive expectations about his camp stay, make him feel distant from home once he is separated, and diminish his enjoyment of initial experiences. This boy's negative interpersonal atti- tudes and low perceived control are highly predictive of the negative emotion he experiences once sepa- rated from home, but do not predict his level of homesickness. Rather, it is his homesick disposition that predicts his homesickness severity. Contrary to conventional wisdom, parental separation anxiety does little to contribute to a homesick disposition, and homesickness is not a strong predictor of nega- tive emotion, relative to the effects of interpersonal attitudes and perceived control. Finally, the more de- pressed, anxious, and homesick this boy is, the more likely he is to exhibit internalizing problem behav- iors, and the less likely he is to enjoy his experience.

Conceptualized protectively, the generative model suggests that an older boy, with prior separation ex- perience, secure interpersonal attitudes, and a strong sense of perceived control, is unlikely to develop a homesick disposition and subsequent homesickness, negative emotion, and internalizing behavior. Fur- thermore, this boy's secure interpersonal attitudes and high perceived control predict that his initial ex- perience and impressions at summer camp will be positive and leave him feeling satisfied at the end of his stay. These two boys are fictitious archetypes, provided to animate a mathematical model that ac- counts for observed trends in a large sample of indi- viduals. The examples have heuristic value only. Boys' individual experiences naturally vary.

One technical limitation of the generative version of the model is that the measured variables indicated by the ATTITUDES AND CONTROL factor were collected on day 3 of boys' stays and are being used in the gen- erative model to predict variables that were mea- sured prior to camp or during the first 2 days. Still, many researchers have argued that models of rela- tionships have stability across the life span. The at- tachment measures were not included in the pre-

Thurber and Sigman 927

camp packet for brevity and because parental supervision may have influenced boys' honest an- swers to questions about the ways their parents treat them. Nevertheless, one must concede that circum- stantial factors at camp could also have influenced boys' responses to the attitudes and perceived con- trol questionnaires. Still, if models of relationships are stable over time, it is plausible that interpersonal atti- tudes and perceived control play both a generative and an amplifying role in the etiology of homesick- ness. Furthermore, homesick disposition, separation expectations, and other factors may have different predictive value for children in foster care, refugee circumstances, hospitals, and boarding schools. How far this study's results generalize to other homesick boys and girls remains an empirical question.

As noted, the fit indexes for both models suggest that better measurement and better model specifica- tion are necessary to produce a more accurate picture of how homesickness develops and is expressed in children. Clearly, future research needs to include girls, whose pathogenesis and manifestations of homesickness may be different from those of boys. Nevertheless, the current models suggest some ways that factor estimation could be improved. For exam- ple, HOMESICKNESS and NEGATIVE EMOTION have equal predictive power and account for 21% of boys' inter- nalizing behavior. A better fitting model would need to account for more variance. Intuition suggests that boys' preseparation internalizing behavior might predict boys' internalizing behavior at camp. Our pi- lot research found moderate correlations between parents' preseparation CBCL T-scores and cabin leaders' in-camp CBCL T-scores for withdrawn be- havior scale, r (328) = .24, p < .001, and the anxious / depressed behavior scale, r (328)= .16, p < .005. More comprehensive measurement using larger samples should improve future models.

Perhaps more important than the immediate preci- sion of the particular models presented in this article are their intermediate contributions to our theoretical understanding of homesickness. The correlations and structural equations from this study support several of the contemporary theories summarized in Table 1, but in ways not previously hypothesized. For exam- ple, the current findings support predictions from attachment theory and Fisher's control-demand discrepancy theory. Interpersonal attitudes and perceived control act as generators of a homesick dis- position (expectations of homesickness and presepa- ration negative emotion); less so as amplifiers of a boy's initial experiences and expectations. Whereas negative interpersonal attitudes and low perceived control may impede boys' progress to form new rela-

928 Child Development

tionships and exercise control in the novel environ- ment, these factors have a stronger role in shaping a boy's approach to and conceptualization of a separa- tion. Indeed, Brewin et al.'s theory of negative expec- tations appears to play an important role in the devel- opment of boys' homesickness. Boys' expectations of homesickness in the months prior to summer camp correlated higher than any other single variable with self-reported homesickness during the separation, r(292) = .64. By itself, the single question about pre- separation expectations of homesickness accounted for 41% of the variance in the HOMESICKNESS factor. Extensive peripheral measurement notwithstanding, boys themselves are impressive predictors of their fu- ture distress levels. Their accuracy naturally raises the question of whether intervention could target the predictors of a HOMESICK DISPOSITION. The current findings suggest that inexperience, negative interper- sonal attitudes, and low perceived control all contrib- ute to boys' HOMESICK DISPOSITION. Ironically, but per- haps logically, good separation experience may be the best way to enhance a boy's interpersonal atti- tudes, secure his perceptions of caregiving and emo- tional control, and assuage his parents' anxiety about separations. Several contemporary theories used to explain homesickness suggest key ways to create

such positive separation experiences.

Cognitive models, such as the attentional resources and competing demands models, suggest that the crux of a positive separation experience is distracting activity. Environmental strangeness and geographical attachment theories suggest that the crux of a positive separation experience is the incor- poration of familiar elements, such as food and cus- toms, into the novel environment. Preseparation fa- miliarization with meaningful aspects of the novel

-

environment may also enhance a subsequent separa- tion experience. Finally, opponent process, peer so- cial support, and social flexibility theories suggest that the crux of a positive separation experience is the establishment of new close relationships.

The fact that HOMESICKNESS is not an important pre- dictor of NEGATIVE EMOTION,

once the effects of ATTITUDES AND CONTROL are taken into account, raises two key issues. First, homesickness is not synonymous with negative emotion. The present study, for per- haps the first time, suggests how homesickness and negative emotion can co-occur, yet have dissimilar etiologies. A cogent theoretical distinction between homesickness and negative emotion was made by Zwingmann (1959), who articulated the distinguish- ing cognitive feature of homesickness: preoccupying thoughts about the most gratifying aspects of one's home, or of one's past Erlebnisratlrn (experiential space). As one severely homesick boy remarked, "My mom . . . has a nice smell . . . she let me take one of her pillows with [me] . . . and now it makes me feel, 'Wow, I wish I could smell this . . . on her person'."

A second, related issue is the possibility that there exist two or more groups of homesick children. Rose (1947,1948) was the first to argue this point, based on retrospective data from female first-year university students. Her research suggested that "Type B" homesick women had a family history of stability and security, but had not had much social opportu- nity to travel away from home or mix with diverse groups of people. By contrast, "Type C" homesick women had a family history of rejection, as well as little social opportunity. Rose reported that Type B women experienced episodic homesickness, whereas Type C women experienced chronic homesickness. Our preliminary research suggests that recent nega- tive events in boys' lives (e.g., parental divorce, loss of a major attachment figure) do not necessarily pre- dispose them to homesickness. However, results of previous research indicate that different children miss different things about home (Thurber, 1995). Moreover, testimony from individual children has confirmed that enduring a recent negative life event is one pathway to homesickness. Recently, van Til- burg et al. (1996) conducted longitudinal case studies and found evidence for three homesickness chro- nologies in women: "anticipation," "holiday," and "combination." Further investigation of the possibil- ity that certain risk and protective factors affect the chronicity or severity of homesickness in children will be a valuable theoretical step, and will have im- portant implications for prevention.

An additional challenge for future researchers, as well as for surrogate caregivers in camps, schools, hospitals, and foster homes, is to promote children's mental health in environments that distract, stimu- late, and support, without an overwhelming degree of novelty. Although severely homesick boys are less likely to return to summer camp than their nonhome- sick peers, boys who do return to the same camp the next summer experience a significant decrease in their mean level of homesickness (Thurber, 1998). Perhaps these returning boys are those whose posi- tive interpersonal attitudes and perceived control were augmented, whose self-expectations of home- sickness were dramatically lowered, and whose abil- ity to negotiate future separations was increased. Thus, whereas homesickness may be the natural- albeit distressing-result of inexperience and a de- veloping sense of interpersonal security and personal control, it also may be the very phenomenon that children must encounter and overcome to prevent. If so, then preseparation orientation to novel environ- ments and supportive exposure to separations are fundamental to beneficial encounters with homesick- ness.

ACKNOWLEDGMENTS

This research was supported in part by grants from Sigma Xi, the Scientific Research Society, the Colin
~rown Fund, a UCLA ~e~artmentalResearch Award, and National Research Service Award MH10920-01 from the National Institutes of Health. Foremost, we wish to thank John R. Weisz for his en- thusiasm and insightful comments about the design and documentation of this research. We also thank

APPENDIX A

RATE YOUR DAY-REVISED

Thurber and Sigman 929

Caryn and Gene Clark, Tom Giggi, Jon Malinowski, Roger Maroni, Steve Reville, and the parents of all the campers for their enormous support of this re- search; the cabin leaders and camp staff, for their hard work and creative contributions; and the campers themselves, for their inspiration, time, and honesty.

ADDRESSES AND AFFILIATIONS

Corresponding author: Christopher A. Thurber, De-
partment of Rehabilitation Medicine, Box 359740,
School of Medicine, University of Washington, Seat-
tle, WA 98195; e-mail: cthurber@u.washington.edu .
Marian D. Sigman is at the University of California,
Los Angeles.

Please read each of the words or phrases below and think about how much you felt that way. Circle an answer for each word or phrase on the number scale. Think of the number scale as a kind of thermometer. You can circle any number between 0and 10, depending on how much you felt that way. There are no right or wrong answers. It's your opinion that counts.

NOT AT   VERY
ALL SORT OF MUCH
................... .....I felt HAPPY ...    
I felt SAD ........................................    
I felt WORRIED .............................    
I felt EXCITED ...............................    
I felt AFRAID ................................    
I felt RELAXED .............................    
I felt HOMESICK ..........................    
I felt DOWN ..................................    
I felt LONELY ...............................    
I felt CALM ................................    
I felt CHEERFUL ..........................    
I missed my family .......................    
I felt PEACEFUL ...........................    
I felt NERVOUS ............................    
I wanted to be home ....................    
930 Child Development

APPENDIX B

SELF-REPORT ATTACHMENT STYLE PROTOTYPES

Participants are provided with copies of each paragraph and six Likert-type scales.

Read by administrator: "This part asks you to think about yourself and other people. I'm going to read you six short paragraphs that describe different thinking. Everyone thinks something different. I'm going to read each para- graph twice, and your job is to choose how much what I read sounds like you. If what I read doesn't sound like you at all, then circle the zero. If it sounds exactly like you, then circle the ten. If what I read sounds sort of like you, then circle a number between zero and ten, depending on how much it sounds like you. You can circle any number you want for each of the six stories-whatever fits the way you think about yourself and other people. There aren't any right or wrong answers. The important thing is that you circle what you think is true for you. Remember, your name isn't on your paper, so all your answers are private. Does anyone have any questions? O.K. First we'll do Practice A at the top of your page. I'll read each description twice. Listen carefully and decide how much this sounds like you."

(Practice A,) I like to play games and sports with other people. I like to be on the team that wins and I don't like it when other people cheat. [READ AGAIN]

"O.K. Now it's your job to decide how much that sounded like you. [PAUSE] O.K. Here's Practice B. I'll read it twice. Listen carefully and circle your answer. Remember, it can be any number between 0 and 10."

(Practice B.) I like to clean up other people's messes. I know that sometimes people don't have time to put all their things away, but that's O.K. be- cause I like to do it for them. If someone came to my house and messed up my room, that would be fine with me, because I would just clean it up. [READ AGAIN]

"Has everyone circled an answer? Don't think about it too long, just circle what seems right. Now we're on NUMBER

1. Listen carefully and circle your answer."

1.Sometimes it's hard for me to like other people a lot and have them like me. I zoant to like other peo- ple and have them like me, but it's hard to trust other people all the way, or really count on them. I some- times worry that my feelings will get hurt if I like other people too much or if they like me too much. [READ AGAIN PLEASE]

"Now we're on NUMBER 2. Listen carefully and circle your answer."

2. It's pretty easy for me to like other people a lot and have them like me. I can do things for myself, but I also feel fine counting on other people for things and having them count on me for things. I don't worry that much about being alone, or about people being my friend. [READ AGAIN PLEASE]

"Now we're on NUMBER 3. Listen carefully and circle your answer."

3. I want to like other people and be with them a lot, but sometimes those people don't like me, or they don't want to be with me as much as I want. It's hard for me to be without people I like and who like me, but I sometimes worry that other people don't like it when I'm around. [READ AGAIN PLEASE]

"Now we're on NUMBER 4. Listen carefully and circle your answer."

4. It's pretty easy for me to be withoi~tpeople I like, or who like me. It is very important to me to do things by myself, and I'd rather not count on other people for things or have them count on me for things. [READ AGAIN PLEASE]

Now look back at your answers to numbers 1, 2, 3, and 4. Which one did you circle the highest number for? If you have a tie for the highest, you'll have to break the tie by picking the paragraph that describes you best overall. At the bottom of your answer sheet, circle the number of the paragraph that sounds the most like you. The practice ones don't count.

REFERENCES

Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D. (1978). Learned helplessness in humans: Critique and reformation. ]oi~rnal of Abnormal Psychology, 87, 49-74.

Abt, T. (1988). Progress zuithoi~t loss of soul: Toward a wholistic approach to modernization planning (B. L. Matthews, Trans.). Wilmette, IL: Chiron. (Original work published 1983).

Achenbach, T. M., & Edelbrock, C. S. (1991). Mani~alfor the Child Behavior Checklist and Revised Child Behavior Profile.

Burlington: University of Vermont.

Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978).Patterns of attachment: A psychological sti~dy of the Strange Situation. Hillsdale, NJ: Erlbaum.

Ainsworth, M. D. S., & Wittig, B. A. (1969). Attachment and the exploratory behavior of one-year-olds in a strange situation. In B. M. Foss (Ed.), Determinants ofinfant behav- ior (Vol. 4, pp. 113-136). London: Methuen.

Alapack, M. C. L., & Alapack, R3J. (1984). The hinge on the door to authentic adulthood: A Kierkegaardian inspired synthesis of the meaning of leaving home. Journal of Phe- nomenological Psychology, 15, 45-69.

American Psychiatric Association. (1994). Diagnostic and statistical mani~al of mental disorders (4th ed.). Washing- ton, DC: Author.

Armsden, G. C., & Greenberg, M. T. (1987). The Inventory of Peer and Parent Attachment: Individual differences and their relationship to psychological well-being in adolescence. Iournal of Youth and Adolescence, 16, 427

454.

Armsden, G. C., McCauley, E., Greenberg, M. T., Burke,

P. M., & Mitchell, J. R. (1990). Parent and peer attach- ment in early adolescent depression. Iournal of Abnormal Child Psychology, 18, 683-697.

Arrendondo-Dowd, P. M. (1981). Personal loss and grief as a result of immigration. Personnel and Guidance Journal, 59, 376-378.

Bachman, J. G., & O'Malley, P. M. (1977). Self-esteem in young men: A longitudinal analysis of the impact of ed- ucation and occupational attainment. Journal of Personal- ity and Social Psychology, 35, 365-380.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. journal of Personality and Social Psychology, 61, 226-244.

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246. Bentler, P. M. (1995). EQS Program Manual. Encino, CA: Multivariate Software. Berg, B. R. (1958). Psychology in children's camping. New York: Vantage.

Blatt, S. J., D'Afflitti, J. P., & Quinlan, D. M. (1979). Depressive Experiences Questionnaire. Unpublished manuscript, Yale University.

Bowlby, J. (1969). Attachment and Loss: Vol. 1. Attachment. New York: Basic.

Brewin, C. R., Furnham, A., & Howes, M. (1989). Demo- graphic and psychological determinants of homesick- ness and confiding among students. British Iournal of Psychology, 80, 467-477.

Broadbent, D. E., Cooper, P. F., Fitzgerald, P., & Parkes,

K. R. (1982). The Cognitive Failures Questionnaire (CFQ) and its correlates. British Journal of Clinical Psy- chology, 21, 1-16.

Burger, G. (1975). A short form of the California Psycholog- ical Inventory. Psychological Reports, 37, 179-182. Burrow, D. (1992). How to be a great camp counselor. Shirley, MA: McElroy.

Burt, C. D. B. (1993). Concentration and academic ability following the transition to university: An investigation of the effects of homesickness. Journal of Environmental Psychology, 13, 333-342.

Capps, L., Sigman, M., Sena, R., Henker, B., & Whalen, C. (1996). Fear, anxiety, and perceived control in children of agoraphobic parents. Journal of Child Psychology and Psychiatry, 37, 445-452.

Carden, A. I., & Feicht, R. (1991). Homesickness among American and Turkish college students. Journal of Cross- Cultural Psychology, 22, 418-428.

Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K. (1994). De- pression, working models of others, and relationship functioning. Journal of Personality and Social Psychology, 66, 127-140.

Collins, B. E. (1974). Four components of the Rotter Internal-External Scale: Belief in a difficult world, a just world, a predictable world, and a politically responsive world. Iournal of Personality and Social Psychol- ogy, 29, 381-391.

Conklin, E. S. (1935). Principles of adolescent psychology. New York: Holt.

Thurber and Sigman 931

Cox, J. L. (1988). The overseas student: Expatriate, sojourner, or settler? Acta Psychiatrica Scandinavica, 78, 179-184.

Crown, S., &Crisp, A. M. (1966). A short clinical diagnostic self-rating scale for psychoneurotic patients: The Mid- dlesex Hospital Questionnaire (MHQ). British Journal of Psychiatry, 112, 917-923.

Davidson, A. J. (1983). How to refresh yourself as a camp nurse. Nursing Life, 3, 60-61. Doherty, J. K. (1940). Solving camp behavior problems. New York: Association Press.

Eisenbruch, M. (1990). Cultural bereavement and home- sickness. In S. Fisher & C. L. Cooper (Eds.), On the move: The psychology of change and transition (pp. 191-205). New York: Wiley.

Eurelings-Bontekoe, E. H. M., Vingerhoets, A., & Fontijn,

T. (1994). Personality and behavioral antecedents of homesickness. Personality and lndividual Differences, 16, 229-235.

Eysenck, H. J., & Eysenck, S. B. G. (1975). The manual of the Eysenck Personality Questionnaire. San Diego, CA: EDITS.

Fisher, S. (1984). Stress and the perception of control. London: Erlbaum Fisher, S. (1988). Leaving home: Homesickness and the psy- chological effects of change and transition. In S. Fisher &

J. Reason (Eds.), Handbook of life stress, cognition, and health (pp. 41-59). New York: Wiley. Fisher, S. (1989). Homesickness, cognition, and health. Hove, Sussex: Erlbaum.

Fisher, S. (1990). The psychological effects of leaving home: Homesickness, health and obsessional thoughts. In S. Fisher & C. L. Cooper (Eds.), On the move: The psychology of change and transition (pp. 153-170). New York: Wiley.

Fisher, S., Elder, L., & Peacock, G. (1990). Homesickness in a school in the Australian Bush. Children's Environments Quarterly, 7, 15-22.

Fisher, S., Frazer, N., & Murray, K. (1984). The transition from home to boarding school: A diary-style analysis of the problems and worries of boarding school pupils. Journal of Environmental Psychology, 4, 211-221.

Fisher, S., Frazer, N., & Murray, K. (1986). Homesickness and health in boarding school children. Iournal of Envi- ronmental Psychology, 6, 35-47.

Fisher, S., & Hood, B. (1987). The stress of the transition to university: A longitudinal study of psychological distur- bance, absent-mindedness and vulnerability to home- sickness. British Iournal of Psychology, 78, 425-441.

Fisher, S., & Hood, B. (1988). Vulnerability factors in the transition to university: Self-reported mobility history and sex differences as factors of psychological distur- bance. British Journal of Psychology, 79, 309-320.

Fisher, S., Murray, K., & Frazer, N. A. (1985). Homesick- ness, health, and efficiency in first year students. journal of Environmental Psychology, 5, 181-195.

French, J. R. P., Caplan, R., &van Harrison, R. (Eds.). (1982). The mechanics of job stress and strain. New York: Wiley.

Freud, S. (1966). Introductory lectures on psycho-analysis (J. Strachey, Trans.). New York: Norton. (Original work published 1920).

932 Child Development

Furnham, A., & Bochner, S. (1986). Culture shock. London: Methuen. Hall, G. S. (1897). Some aspects of the early sense of self. American Journal of Psychology, 9, 351-395. Harris, P. L. (1989). Children and emotion. Oxford: Basil Blackwell.

Harris, P. L., & Guz, G. R. (1986). Models of emotion: How boys report their emotional reactions upon entering an En- glish boarding school. Unpublished paper, Department of Experimental Psychology, University of Oxford. Hartwig, M., & Myers, B. (1961). Children are human, even at camp. Minneapolis, MN: Burgess. Hilsman, R., & Garber, J. (1995). A test of the cognitive diathesis-stress model of depression in children: Aca- demic stressors, attributional style, perceived compe- tence, and control. Journal of Personality and Social Psy- chology, 69, 370-380. Hock, E., McBride, S., & Gnezda, M. T. (1989). Maternal separation anxiety: Mother-infant separation from the maternal perspective. Child Development, 60, 793-802. Hofer, J. (1688).Dissertatio medica de nostalgia oder heimzuehe [Medical dissertation on nostalgia or homesickness]. Unpublished doctoral dissertation, University of Rauracum, Basil, Switzerland. Introduced and translated by C. K. Anspach (1934) in Bulletin of the Institute of the History of Medicine, 2, 376-391. Hojat, M., & Herman, M. W. (1985). Adjustment and psy- chosocial problems of Iranian and Filipino physicians in the U.S. ]ournal of Clinical Psychology, 41, 130-136. Hollingshead, A. B. (1975). Four Factor Index of Social Status. Unpublished manuscript, Yale University, New Haven, CT. Hu, L., Bentler, P. M., & Kano, Y. (1992). Can test statistics in covariance structure analyses be trusted? Psychological Bulletin, 112, 351-362. Joy, B. E. (1957). Camping. Minneapolis, MN. Burgess. Kane, G. (1987). Studies of coping in stressed populations. Unpublished doctoral dissertation, University of Manches- ter, U.K. Kleinman, A,, & Cohen, A. (1997). Psychiatry's global chal- lenge. Scientific American, 276, 86-89. Kovacs, M. (1980). Rating scales to assess depression in school aged children. Acta Paedopsychiatrica, 46, 305-315. Lee, S. W., Piersel, W. C., Friedlander, R., & Collamer, W. (1988). Concurrent validity of the Revised Children's Manifest Anxiety Scale (RCMAS) for adolescents. Educational and Psychological Measurement, 48, 429-433. Lerner, J. V., Baker, N., & Lerner, R. M. (1985). A person- context goodness of fit model of adjustment. In P. C. Kendall (Ed.), Advances in cognitive-behavioral research and therapy (Vol. 4, pp. 111-136). New York: Academic Press. Levy, C. S. (1952). Guilt and homesickness at summer camp. Journal of Child Psychiaty, 2, 252-256. Lu, L. (1990). Adaptation to British universities: Homesick- ness and mental health of Chinese students. Counseling Psychology Quarterly, 3, 225-232. Luteijn, F., & Kok, A. R. (1985). Nederlandse verkorte MMPI (NVMI. Lisse: Swets & Zeitlinger.

Luteijn, F., Starren, J., & Van Dijk, H. (1985). Handleiding Nederlandse Persoonlijkheidsvragenlijst. Lisse: Swets & Zeitlinger.

Mandler, G. (1975). Mind and emotion. New York: Wiley.

Marks, D. F. (1973). Visual imagery differences in the recall of pictures. British ]ournu1 of Psychology, 64, 17-24.

Matheny, K. B., & Cupp, P. (1983). Control, desirability, and anticipation as moderating variables between life change and illness. Journal of Human Stress, 9, 14-23.

McCann, W.H. (1941). Nostalgia: A review of the literature. Psychological Bulletin, 38, 165-182. McCann, W. H. (1943). Nostalgia: A descriptive and com- parative study. Journal of Genetic Psychology, 62, 97-104.

McNair, D. M., Loor, M., & Droppleman, L. F. (1971). Manual for the Profile of Mood States. San Diego, CA: Educa- tional & Industrial Testing Services.

Mechanic, D. (1983). Adolescent health and illness behav- ior: Review of the literature and a new hypothesis for the study of stress. journal of Human Stress, 9, 4-14.

Meeks, J. E. (1968). Homesickness and the homesickness crisis in residential therapy. ]ournal of the American Acad- emy of Child Psychiatry, 7, 350-365.

Mitchell, A. V., & Crawford, I. B. (1965). Camp counseling. Philadelphia: W. B. Saunders.

Mooney, S. P., Sherman, M. F., & Lo Presto, C. T. (1991). Academic locus of control, self-esteem, and perceived distance from home as predictors of college adjustment. Journal of Counseling and Development, 69, 445-448.

Nicassio, P. M., & Pate, J. K. (1984). An analysis of problems of resettlement of the Indochinese refugees in the United States. Social Psychiatry, 19, 135-141.

Nolen-Hoeksema, S., Girgus, J. S., & Seligman, M. E. P. (1992). Predictors and consequences of childhood de- pressive symptoms: A 5-year longitudinal study. journal of Abnormal Psychology, 101, 405-422.

Ollendick, T. H., Yule, W., & Ollier, K. (1991). Fears in Brit- ish children and their relationship to manifest anxiety and depression. Journal of Child Psychology and Psychia- try, 32, 321-331.

Papini, D. R., & Roggman, L. A. (1992). Adolescent per- ceived attachment to parents in relation to competence, depression, and anxiety: A longitudinal study. ]ournal of Early Adolescence, 12, 420-440.

Pattem, O., West, M., Mahoney, A., & Keller, A. (1993). De- pression and attachment problems. journal of Psychiatry and Neuroscience, 18, 78-81.

Persons, J. B., Burns, D. D., Perloff, J. M., & Miranda, J. (1993). Relationships between symptoms of depression and anxiety and dysfunctional beliefs about achieve- ment and attachment. ]ournu1 of Abnormal Psychology, 102, 518-524.

Platt, J. J., & Taylor, R. E. (1967). Homesickness, future time perspective, and the self-concept. Journal of Individual Psychology, 23, 94-97.

Porritt, D., & Taylor, D. (1981). An exploration of homesick- ness among student nurses. Australian and New Zealand Journal of Psychiatry, 15, 57-62.

Pruitt, F. J. (1978, Spring). The adaptation of African stu- dents to American society. International journal oflnter- cultural Relations, 91-117.

Rather, L. J. (1965).Mind and body in 18th centu y medicine: A study based on Jerome Gaub's "De regimine mentis." Los Angeles: University of California Press.

Redl, F. (1947). Psychopathologic risks of camp life. The Nervous Child, 6, 139-147.

Reiche, H., & van Dijkhuizen, N. (1980). Vragenlijst organi- satiestress; testhandleiding voor testafname. Internal report. Nijmegen: Nijmegen University.

Reynolds, C. R., & Richmond, B. 0.(1978). What I think and feel: A revised measure of children's manifest anxiety. Journal of Abnormal Child Psychology, 6, 271-280.

Robinson, N. S., Garber, J., & Hilsman, R. (1995). Cogni- tions and stress: Direct moderating effects on depressive versus externalizing symptoms during the junior high school transition. Journal of Abnormal Psychology, 104, 453-463.

Rose, A. A. (1947). A study of homesickness in college freshmen. Journal of Social Psychology, 26, 185-202. Rose, A. A. (1948). The homes of homesick girls. Journal of Child Psychiatry, 1, 181-189. Rosen, G. (1975). Nostalgia: a "forgotten" psychological disorder. Psychological Medicine, 5, 340-354.

Rudolph, K. D., Hammen, C., & Burge, D. (1995). Cognitive representations of self, family, and peers in school-age children: Links with social competence and sociometric status. Child Development, 66, 1385-1402.

Satorra, A,, & Bentler, P. M. (1988). Scaling corrections for chi-square statistics in covariance structure analysis. Proceedings ofthe American Statistical Association, 20,308-313.

Saylor, C. F., Finch, A. J., Baskin, C. H., Furey, W., & Kelly,

M. M. (1984). Construct validity for measures of childhood depression: Application of multitraitmultimethod methodology. journal of Consulting and Clinical Psychology, 52, 977-985.

Saylor, C. F., Finch, A. J., Spirito, A., & Bennett, B. (1984). The children's depression inventory: A systematic eval- uation of psychometric properties. Iournal of Consulting and Clinical Psychology, 52, 955-967.

Schwarzer, R., Hahn, A,, & Jerusalem, M. (1993). Negative emotion in East German migrants: Longitudinal effects of unemployment and social support. Anxiety, Stress, and Coping, 6, 57-69.

Seabury, D. (1924). Unmasking our minds. New York: Live- right. Solomon, R. L. (1980). The opponent process theory of ac- quired motivation. American Psychologist, 35, 691-712.

Stokols, D., Schumaker, S. A., & Martinez, J. (1983). Resi- dential mobility and personal well-being. Journal ofEnvi- ronmental Psychology, 3, 5-19.

Takahashi, K. (1990). Affective relationships and their life- long development. In P. B. Baltes, D. L. Featherman, &

R. M. Lerner (Eds.), Lifespan development and behavior (Vol 10, pp. 1-27). Hillsdale, NJ: Erlbaum.

Takahashi, K., & Majima, N. (1994). Transition from home to college dormitory: The role of preestablished affective relationships in adjustment to a new life. Journal of Re- search on Adolescence, 4, 367-384.

Taylor, R. E. (1986, June). Homesickness, melancholy, and blind rehabilitation. Journal of Visual Impairment and Blindness, 800-802.

Thurber and Sigman 933

Thurber, C. A. (1995). The experience and expression of homesickness in preadolescent and adolescent boys. Child Development, 66, 1162-1178.

Thurber, C. A. (1996). Separation anxiety disorder and the collapse of anxiety disorders of childhood and adoles- cence. Anxiety Disorders Practice Journal, 2, 115-135.

Thurber, C. A. (1998). The phenomenology of homesickness in boys: Longitudinal follow-up and extension. Manuscript submitted for publication, University of Washington.

Thurber, C. A,, Bombar, M. L., & Sigman, M. D. (1994).Revision ofthe Self-Report Attachment Style Prototypesfor use zuith children and adolescents. Unpublished manuscript, University of California at Los Angeles.

Thurber, C. A,, Sigman, M. D., Weisz, J. R., & Schmidt, C. (1998).Homesickness in preadolescent and adolescent girls: Riskfactors, behavioral correlates, and sequelae. Manuscript submitted for publication, University of Washington.

Thurber, C. A,, & Weisz, J. R. (1997). "You can try or you can just give up": The impact of perceived control and coping style on childhood homesickness. Developmental Psychology, 33, 508-517.

Van Dam-Baggen, C. M. J., & Kraaimaat, F. W. (1987). Inventarisatielijst Omgaan met Anderen. Lisse: Swets & Zeitlinger.

van Tilburg, M. A. L., Vingerhoets, A. J. J. M., & van Heck,

G. L. (1996). Homesickness: A review of the literature.

Psychological Medicine, 26, 899-912. van Tilburg, M. A. L., Vingerhoets, A. J. J. M., van Heck,

G. L., & Kirschbaum, C. (1996). Mood changes in home- sick persons during a holiday trip. Psychotherapy and Psychosomatics, 65, 91-96.

van Wagenen, B. C. (1933). Summer camps. New York: Child Development Institute.

Vernberg, E. M., & Field, T. (1990). Transitional stress in children and young adolescents moving to new environ- ments. In S. Fisher & C. L. Cooper (Eds.), On the move: The psychology of change and transition (pp. 127-151). New York: Wiley.

Wapner, S., Kaplan, B., & Ciottone, R. (1981). Self-world relationships in critical environmental transitions: Childhood and beyond. In L. S. Liben, A. H. Pat- terson, & N. Newcombe (Eds.), Spatial representation and behavior across the life span (pp. 251-280). New York: Aca- demic Press.

Ward, C., & Kennedy, A. (1992). Locus of control, mood disturbance, and social difficulty during cross-cultural transitions. International journal of Intercultural Relations, 16, 175-194.

Ward, C., & Kennedy, A. (1993). Psychological and socio- cultural adjustment during cross-cultural transitions: A comparison of secondary students overseas and at home. International Jourr~al of Psychology, 28, 129-147.

Weisz, J. R. (1990). Development of control-related beliefs, goals, and styles in childhood and adolescence: A clini- cal perspective. In K. W. Schaie, J. Rodin, & C. Scholler (Eds.), Self-directedness and eficacy: Causes and ejfects throughout the life course (pp. 103-145). Hillsdale, NJ: Erl- baum.

Weisz, J. R., Proffitt, V., & Sweeney, L. (1991).The Perceived Control Scale for Children: Development and validation. Un

934 Child Development

published manuscript, University of California at Los Angeles.

Weisz, J. R., Stevens, J. S., Curry, J. F., Cohen, R., Craighead,

W. E., Burlingame, W. V., Smith, A,, Weiss, B., & Par-melee, D. X. (1989). Control-related cognitions and de- pression among inpatient children and adolescents.

Journal of the American Academy of Child and Adolescent

Psychiatry, 28, 358-363.

Weisz, J. R., Sweeney, L., Proffitt, V., & Carr, T. (1993). Control-related beliefs and self-reported depressive symptoms in late childhood. Journal of Abnormal Psy- chology, 102, 411-418.

Weisz, J. R., Weiss, B., Wasserman, A. A., & Rintoul, B. (1987). Control-related beliefs and depression among clinic-referred children and adolescents. journal of Ab- norinal Psychology, 96, 58-63. West, M., Rose, M. S., & Sheldon, A. (1993). Anxious attach- ment as a determinant of adult psychopathology. Journal of Nervous and Mental Disease, 181, 422-427. White, R. W. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66, 297-333.

Wickland, R. A. (1975). Objective self-awareness. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 8, pp. 233-275). New York: Academic Press.

Winland-Brown, J. E., & Maheady, D. C. (1990). Using intu- ition to define homesickness at a summer camp. journal ofpediatric Health Care, 4, 117-121.

Woulff, N. (1975).Homesickness in college freshmen. Unpublished doctoral dissertation, University of Maine, Orono.

Zimmerman, D. R., & Bijur, P. E. (1995). Homesickness and the use of a camp infirmary: A preliminary report. Developmental and Behavioral Pediatrics, 16, 187-191.

Ziickert, J. F. (1768).Von den Leidenschaften, 2te Aufl., Berlin. Zung, W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70.

Zwingmann, C. A. (1959). "Heimveh" or "nostalgic reaction": A conceptual analysis and interpretation of a medico-psychological phenomenon. Unpublished doctoral disser- tation, Stanford University, Palo Alto, CA.

Comments
  • Recommend Us