Journal of Pediatric Psychology, Vol. 25, No. 5, 2000, pp. 323-329
© 2000 Society of Pediatric Psychology
The Impact-on-Family Scale : A Test of Invariance Across Culture
1 University of Amsterdam, 2 University of Genoa
All correspondence should be sent to Annemarie M. Kolk, Department of Clinical Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands. E-mail : kp_kolk{at}macmail.psy.uva.nl
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionReferences |
|---|
Objective : To test the cultural invariance of the Impact-on-Family Scale in order to make cross-cultural comparisons.
Methods : The Italian version of the scale was administered to mothers of children with chronic illnesses. Factorial invariance was examined to investigate whether the four factors found with the original United States (U.S.) scale could be replicated.
Results : The results clearly demonstrate the replicability of the first three factors, Financial Burden, Familial/Social Impact, and Personal Strain. In addition, internal consistency and the homogeneity of the items of the corresponding scales are satisfactory. However, the fourth factor, Mastery, could not be replicated, and the reliability of the corresponding scale is poor. Italian mothers scored significantly lower on the Financial Burden and on the Familial/Social Impact dimension, compared to the American sample.
Conclusions : Three of the four factors of the Impact-on-Family Scale are useful for cross-cultural comparisons between U.S. and Italian samples.
Key words: children; mothers; chronic illness; family; methodology; Impact-on-Family Scale.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Chronic childhood illness has considerable consequences for the family system. It is a distressing experience and requires coping and adaptation (Kazak, 1992
;
Kazak, Segal-Andrews, & Johnson,
1995). To determine the effects of chronic childhood illness on
the family, Stein and Riessman
(1980) developed the
Impact-on-Family Scale. Although they considered positive effects, they focus
on negative effects. Negative influences of illness are conceptualized in
terms of losses : financial burden, restrictions in social life, decreased
interaction with significant others, less time for other family members, and
increased subjective distress or strain.
The Impact-on-Family Scale is a 24-item questionnaire measuring four
dimensions of impact. Financial Burden refers to the economic consequences for
the family. Familial/Social impact concerns the disruption of social
interaction. Personal Strain assesses the psychological burden experienced by
the primary caretaker. Mastery refers to the coping strategies employed by the
family. Since its publication in 1980, the Impact-on-Family Scale has been
used in a considerable number of studies concerning a range of chronic
childhood illnesses, mostly physical diseases. However, the psychometric
properties of the instrument are less studied and consequently not completely
understood. Stein and Jessop
(1985) corroborated the
reliability and construct validity of the scale. Carpiniello et al.
(1995) evaluated the
reliability and validity of the Italian version
(Casari & Fantino, 1991) of
the Impact-on-Family Scale. They reported a satisfactory test-retest
reliability. With respect to validity, the scale proved to discriminate
between chronic illness (cancer and nonneoplastic illnesses (cardiopathy,
hepatopathy, bronchopneumopathy, or rheumatic pathology) and acute illnesses
(gastro-intestinal, respiratory, genito-urinary, or skin disorders). Several
studies indirectly confirm the validity of the instrument. Scores on the
Impact-on-Family Scale of caregivers of patients with asthma are shown
associated with scores on the Pediatric Asthma Caregiver's Quality of Life
Questionnaire (Juniper et al.,
1996). In addition, it has been found that the more severe and/or
debilitating the illness, the greater the impact on the family
(Coster, Haley, & Baryza,
1994 ; Cronin, Shapiro, Casiro,
& Cheang, 1995 ; Gennaro,
1995 ; Lesar & Maldonado,
1997 ; Zahr, Khoury, &
Saoud, 1994).
The Impact-on-Family Scale's utility for crosscultural comparisons has not
yet been established. Even after a thorough translation and backtranslation of
the items, researchers must still ascertain that respondents in various
cultures similarly interpret the items and that the scale is, therefore,
equally applicable across cultures (Drotar,
Stein, & Perrin, 1995 ;
Foster & Martinez, 1995).
For example, Zahr et al.
(1994) compared the scores on
the Impact-on-Family Scale of their Lebanese sample with those of the U.S.
sample from the Stein and Riessman
(1980) study to examine
different cultural responses to chronic illness. They attributed the higher
impact-scores of the Lebanese sample to a tendency of Middle Eastern people
"to dramatize and exaggerate their sufferings," and "to the
stress associated with the civil war in Lebanon" (p. 400). However, no
one knows whether the factors that account for the major part of the variance
in the U.S. population act similarly in the Lebanese population. Therefore,
analysis of invariance and item analysis must first ascertain whether certain
dimensions or factors can be meaningfully measured in different cultures
(Eysenck & Eysenck, 1983).
Only then can cultural differences have validity. Subsequently, these
differences can be attributed to sociodemographic, social, or psychological
differences between cultures, pertaining to, for example, the characteristics
of the health care system, or the structure of family life.
This study aimed to investigate the suitability of the Impact-on-Family Scale for use with Italian mothers of children with chronic illnesses. Our main interests were (1) the replicability of the original factor structure of the Impact-on-Family Scale, (2) the degree of internal consistency of the subscales, the extent of homogeneity of the items of each scale, and the age independence of impactscores. Subsequently, we investigated (3) whether there are cultural differences in impact-scores.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Participants and Procedure
The mothers of 387 children with chronic diseases participated in the study. The children had had symptoms of illness interfering with daily functioning for more than 12 months, or required hospitalization or extensive home- or community-based health services for more than 1 month during a 12-month period. Informed consent was obtained from the mothers during one of their regular visits to the outpatient department of the G. Gaslini Hospital (Genoa, Italy). Directly after the visit to the medical specialist, one of the clinical psychologists of the outpatient department administered the Italian version of the Impact-on-Family Scale (Casari & Fantino, 1991
) to
the mothers. The age of the mothers ranged from 20 to 50, with a mean of 34.9 years (SD = 6.5). Most mothers were not employed outside the home, and about one-third had graduated from high school. The children were between 2 and 16 years old (M = 7.6, SD = 4.1). Seventy percent of the children had spina bifida. Most children lived with both their (biological) parents. The families came from different parts of Italy (see Table I).
|
Measure
The Impact-on-Family Scale contains 24 statements. Respondents are asked to
rate on a 4-point Likert scale to what degree the statements apply to the
parent or the family (strongly agree to strongly disagree). Stein and Riessman
(1980) conducted a principal
components analysis on the ratings of 100 mothers and found four dimensions of
impact. Factor 1, Financial Burden (four items) refers to the economic
consequences for the family. Factor 2, Familial/Social Impact (nine items)
concerns the disruption of social interaction. Factor 3, Personal Strain (six
items) assesses the psychological burden experienced by the primary caretaker.
Factor 4, Mastery (five items) refers to the coping strategies employed by the
family. Internal consistency reliabilities (Cronbach's 
s) were.72 for
Financial Burden,.86 for Familial/Social Impact,.81 for Personal Strain,.60
for Mastery. A total score was obtained by summation of all 24 scores. A high
score indicates greater impact. Cronbach's was.88 for the total
scale.
In 1985, Stein and Jessop conducted a second principal components analysis,
this time on the ratings of 205 primary caretakers, including the mothers of
the first analysis. Again, four dimensions were found. In this second
analysis, the items "time is lost from work" and "cut down
hours of work" were replaced by the items "worry about the
future" and "relatives interfere." Therefore, the two
resulting factor structures have 22 identical items
(Stein & Jessop,
1985).
Casari and Fantino (1991)
developed the Italian version of the Impact-on-Family Scale. The items of the
Impact-on-Family Scale as described in Stein and Riessman
(1980) were translated from
English (U.S.) into Italian by three Italian clinical psychologists and then
translated back into English by two native English speakers. A small-scale
test of the Italian translation proved to be satisfactory
(Brislin, 1993).
Data Analyses
In this study, we used a technique of confirmatory analysis similar to that
used by Arrindell, Hanewald, and Kolk
(1989). This method enables the
interpretation of the similarity between sets of factors derived for identical
items from different populations in terms of factor invariance. Principal
components analysis followed by orthogonal Varimax rotation was performed on
the ratings of the Italian sample. To examine factor invariance, we compared
the Varimax rotated matrix of factor loadings to the Stein and Riessman
(1980) factor structure. For a
measure of factorial invariance, Tucker's coefficient of of congruence [UNK]
was utilized. Tucker's [UNK] ranges from -1.00 via 0 to +1.00. Coefficients
of.80 to 1.00 represent good to perfect similarity of factors and a
coefficient of.70 indicates poor similarity. Good similarity requires a
coefficient of 
.80. We inspected the coefficients between nonanalogous
factors as well. The strength of each rotated factor (in terms of explained
variance and eigenvalue) in the Italian sample should at least approximate
that of the factors obtained in the original sample.
The factor structure of the Italian sample was also compared with that of
the sample of Stein and Jessop
(1985) with respect to their
22 identical items. The latter comparison is complicated because of item
differences. As mentioned earlier, two items are different. Nevertheless, we
performed it to check whether substantial differences from the first
comparison would occur.
Differences with respect to individual items in the current and original
samples can be discovered by analyzing the internal consistency of each scale
(Cronbach's coefficient ) in conjunction with the item-remainder
correlation for each item. The lowest acceptable item-remainder correlation
was set at.2 (Nunnally & Bernstein,
1994, p. 306). The homogeneity indices for each set of
theoretically relevant items are also of importance. According to Briggs and
Cheek (1986), mean inter-item
correlations between.2 and.4 indicate an optimal level of homogeneity
(Briggs & Cheek, 1986, p.
115). The association between the age of the mother and the age of the child
and the impact-scores was determined by means of analysis of variance (ANOVA).
Mothers as well as children were divided in three age groups of approximately
equal frequency, respectively, 20-31 (n = 131), 32-37 (n =
125), and 38-50 (n = 131), and 2-5 (n = 148), 6-9
(n = 115), and 10-16 (n = 124) years of age.
To investigate cultural differences on the impact on the family, the
Italian sample was compared with the U.S. sample. The mean scores of the
Italian sample were compared with the data from the Stein and Riessman
(1980) study by means of
Student's t test.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Replicability
A principal components analysis followed by orthogonal Varimax rotation was performed on the responses of the Italian sample. Four factors were extracted. At face value, these factors show a marked resemblance to those described in the Stein and Riessman (1980
)
study.
Table II shows the
[UNK]-coefficients that were obtained from comparing the Italian factor
structure with the Stein and Riessman
(1980) factor structure. For
the factor Financial Burden, the [UNK]-coefficient was.91, which indicates a
high degree of similarity to the U.S. factor. With respect to the factor
Familial/Social Impact and the factor Personal Strain, [UNK]-coefficients of,
respectively,.79 and.78 were found. Both are just below.80, which reflects
sufficient similarity. The Mastery factor with a [UNK]-coefficient of.73,
being close to.70, has a rather poor similarity to the original U.S. Mastery
factor. Coefficients of congruence for incompatible components were all well
below.70, ranging from -.06 to.62.
|
The pattern of [UNK]-coefficients pertaining to the comparison with the
factor structure of Stein and Jessop
(1985) proved to be
essentially the same. The coefficients concerning the factor Familial/Social
Impact (.87) and the factor Personal Strain (.80) stand for good similarity.
The Mastery factor (.70) again has a rather poor similarity. With respect to
Financial Burden, the [UNK]-coefficient dropped from.91 (highly similar) to.79
(sufficiently similar). Because two factor-relevant items were deleted in the
analysis, this drop was expected. Coefficients of congruence for incompatible
components were all below.70, ranging from -.01 to.67.
Based on the factor loadings of the items, the percentages of variance
explained by the four factors as well as eigenvalues were calculated for the
U.S. sample and for the Italian sample (see
Table III). In terms of
explained variance, the factor Financial Burden was of equal strength compared
to the original factor. The factors Familial/Social Impact, Personal Strain,
and Mastery explained less of the variance than in the Stein and Riessman
(1980) sample. A considerable
difference in explained variance can be seen on the factor Personal Strain.
The reason for this is that in the U.S. sample there are some theoretically
irrelevant items with nonnegligible .40) loadings on this factor.
|
Again, the results of the comparison with the data of Stein and Jessop
(1985) were essentially the
same. Percentages of explained variance and eigenvalues of the factors
Financial Burden (90%/1.98) and Familial/Social Impact (16.5%/3.52) in the
Italian sample closely resembled those in the Stein and Jessop sample
(respectively, 8%/1.76 and 16%/3.52). The factors Mastery (6%/1.54) and
Personal Strain (10%/2.15) in particular explained less of the variance than
in the Stein and Jessop sample (respectively, 8%/1.76 and 17%/3.74).
Close inspection of each factor loading disclosed that the loadings of the
Italian sample were smaller than in the original Stein and Riessman
(1980) sample (see
Table IV). Although factor
loadings in the Italian sample were proportionally smaller than those in the
original study, 17 of the 24 items (U.S. : 22 of the 24) still had a loading
above.40 on their theoretically relevant factor. The loadings of the items of
the dimension Financial Burden were all well above.40 (see
Table IV). The factors
Familial/Social Impact and Personal Strain had five items with loadings
smaller than.40. Four of them can be considered as just acceptable, being
between.33 and.39. The fifth item of the Familial/Social dimension, "I
think about not having any more children becuase of the illness," had a
loading of.10 on Familial/Social Impact, but.57 on Personal Strain. Two items
of the dimension Mastery ("Treat child as if normal" and
"Relatives understand") had unacceptable loadings (.10
and.11).
|
Internal Consistency and Homogeneity
Table V shows the internal
consistency coefficients (Cronbach's ) of each scale, the ranges of
item-remainder correlations, and the homogeneity indices for each scale.
Cronbach's for the Mastery scale is very low (.32). Stein and Riessman
(1980) also reported a low
alpha coefficient for this scale (.60). Cronbach's for Personal Strain
is also rather low (.65), but acceptable for research with large samples. The
range of the item-remainder correlations pertaining to each factor indicates
that the correlations are all well within the acceptable limits, except for
Mastery. The mean inter-item correlations were of optimal magnitude, again
except for Mastery. Analysis of variance was carried out to assess whether the
impact-scores were independent of age. No significant differences were found ;
the impact of the illness on the family was dependent neither on the age of
the mother or the age of the child.
|
Cultural Differences
The Italian sample scored significantly lower (lower impact) on the
Financial Burden dimension than the U.S. sample, t (485) = 3.91,
p <.001, (see Table
VI). In addition, Italian mothers scored lower on the
Familial/Social Impact dimension, t (485) = 13.06, p <.
001. No differences were found on the dimension Personal Strain, t
(485) = -.87, p >.05, and on Total Impact, t (485) =.11,
p >.05. Italian mothers scored much higher on the Mastery scale,
t (485) = 31.28, p <.001. However, considering the poor
qualities of this factor, no conclusions concerning this result are drawn.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
This study demonstrates the partial suitability of the Impact-on-Family Scale for use with Italian mothers of children with chronic illnesses. The results support the replicability of the dimensions of Financial Burden, Personal Strain, and Familial/Social Impact as well as the internal consistency and homogeneity of the corresponding scales. The impact-scores on these scales were independent of the age of the mother as well as of the age of the child. However, the Mastery dimension does not meet the criteria for replicability, internal consistency, and homogeneity.
We suggest not using the Mastery scale as it is now because the reliability
of the Italian version and the U.S. version is too low and similarity is poor.
The psychometric properties of the Mastery scale should be improved by
including more items. As the scale concerns the positive impact of illness,
the inclusion of more items as well as a reconsideration of the concept of
positive impact would also increase the balance between positive and negative
impact within the scale as a whole. For example, starting with the items
"closer family because of sharing" and "discuss problems
with partner," one could develop a scale that assesses the unifying
impact of illness on the family (Stein
& Riessman, 1980). Moreover, we recommend considering the
removal of the item "I think about not having any more children because
of the illness" from Familial/Social Impact scale, and to add it to the
Personal Strain scale. Then, the Italian Impact-on-Family Scale would contain
the dimensions Financial Burden (four items), Personal Strain (seven items)
and familial/Social Impact (eight items), and cross-cultural comparisons of
Italian with U.S. data would be feasible.
Significant cultural differences were found with respect to Financial Burden and Familial/Social Impact. The chronic illness of a child seems to have less economic consequence for Italian families than for families from the United States. As the explanation of possible cultural differences was not the main object of this study, relevant data were not collected. However, the most obvious interpretation is that, due to the national health care system (Sistema Sanitario Nazionale), health care was free in Italy at the time of the study. Moreover, as most Italian families have close relatives living in their immediate vicinity, spending money on child care is less necessary. The accessibility and availability of family members might explain the difference in the disruption of social interactions as well.
In general, this study underlines the importance of research into the invariance of measures. The Impact-on-Family Scale is used in studies in different cultures, occasionally with the aim of making cross-cultural comparisons. Without a test of invariance, the difference between Italian and American mothers on the Mastery scale could easily lead to unreliable and invalid interpretations of this difference and to erroneous beliefs about cultural differences. Like reliability and validity, the invariance of measures should be routinely tested and retested in the development of instruments used in different cultures or in large countries with different subcultures.
|
Acknowledgments |
|---|
We thank Dr. Ruth Stein (Department of Pediatrics, Albert Einstein College of Medicine, New York) for granting permission to translate the original version of the Impact-on-Family Scale into Italian.
Received July 22, 1998; revision received March 5, 1999; revision received August 10, 1999; revision received October 15, 1999; accepted October 15, 1999
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Arrindell, W. A., Hanewald, G. J. F. P., & Kolk, A. M. M. (1989). Cross-national constancy of dimensions of parental rearing style : The Dutch version of the Parental Bonding Instrument (PBI). Personality and Individual Differences, 10, 949-956.
Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis in the development and evaluation of personality scales. Journal of Personality, 54, 106-148.[Web of Science]
Brislin, R. (1993). Understanding culture's influence on behavior. Fort Worth : Harcourt Brace.
Carpiniello, B., Baita, A., Pariante, C. M., Orru, W., Ruvinetti, A. S., & Rudas, N. (1995). II carico familiare nelle malattie croniche. Studio validazione e di riproducibilità del questionario sull'impatto della malattia cronica sulla famiglia (IMPAT) (The family burden in chronic diseases. The study of the validity and reproducibility of the questionnaire on the impact of the chronic disease on the family [IMPAT]). Minerva Psychiatrica, 36, 171-178.
Casari, E. F., & Fantino, A. G. (1991). Valutazione dell'impatto della malattia cronica sulla famiglia (Assessment of the impact of chronic illness on the family). Bollettino di Psicologia Applicata, 198, 19-27.
Coster, W. J., Haley, S., & Baryza, M. J. (1994). Functional performance of young children after traumatic brain injury : A 6-month follow-up study. American Journal of Occupational Therapy, 48, 211-218.
Cronin, C. M., Shapiro, C. R., Casiro, O. G., & Cheang, M. S.
(1995). The impact of very low-birth-weight infants on the family
is long lasting. A matched control study. Archives of Pediatrics
and Adolescent Medicine, 149,
151-158.
Drotar, D., Stein, R. E. K., & Perrin, E. C. (1995). Methodological Issues in using the Child Behavior Checklist and its related instruments in clinical psychology research. Journal of Consulting and Clinical Child Psychology, 24, 184-192.
Eysenck, H. J., & Eysenck, S. B. G. (1983). Recent advances in the cross-cultural study of personality. In J. N. Butcher & C. D. Spielberger (Eds.), Advances in personality assessment (vol. 2, pp. 41-69). Hillsdale, NJ : Erlbaum.
Foster, S. L., & Martinez, C. R. (1995). Ethnicity : Conceptual and methodological issues in child clinical research. Journal of Clinical Child Psychology, 24, 214-226.[Web of Science]
Gennaro, S. (1995). Preterm low-birthweight infants : Health and family outcomes. Family and Community Health, 17, 12-21.
Juniper, E. F., Guyatt, G. H., Feeny, D. H., Ferrie, P. J., Griffith, L. E., & Townsend, M. (1996). Measuring quality of life in the parents of children with asthma. Quality of Life Research, 5, 27-34.
Kazak, A. E. (1992). The social context of coping with childhood chronic illness : Family systems and social support. In A. M. La Greca, L. J. Siegel, J. L. Wallander, & C. E. Walker (Eds.), Stress and coping in child health (pp. 262-279). New York : Guilford Press.
Kazak, A. E., Segal-Andrews, A. M., & Johnson, K. (1995). Pediatric psychology, research and practice : A family/systems approach. In M. C. Roberts (Ed.), Handbook of Pediatric psychology (pp. 84-104). New York : Guilford Press.
Lesar, S., & Maldonado, Y. A. (1997). The impact of children with HIV infection on the family system. Families in Society. The Journal of Contemporary Human Services, 78, 272-279.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory. New York : McGraw-Hill.
Stein, R. K., & Jessop, D. (1985). Tables documenting the psychometric properties of a measure of the impact of chronic illness on the family. New York : Pact papers, Albert Einstein College of Medicine.
Stein, R. K., & Riessman, C. K. (1980). The development of an Impact-on-Family Scale : Preliminary findings. Medical Care, 28, 465-472.
Zahr, L. K., Khoury, M., & Saoud, N. B. (1994). Chronic illness in Lebanese preschoolers : Impact of illness and child temperament on the family. American Journal of Orthopsychiatry, 64, 396-403.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||