1

Calam, R., Peters, S. (in press) Assessing Expressed Emotion: The association
between Camberwell Family Interview and Five Minute Speech Sample ratings for
mothers of children with behavioural problems. International Journal of Methods in
Psychiatric Research.

Assessing Expressed Emotion: Comparing Camberwell Family Interview and Five

Minute Speech Sample ratings for mothers of children with behavior problems

Rachel Calam1,

Sarah Peters2

1
School of Psychological Sciences, University of Manchester
2
Division of Psychiatry, University of Liverpool

Corresponding author:

Dr Rachel Calam

Academic Division of Clinical Psychology

University of Manchester

Second Floor, Education and Research Centre

Wythenshawe Hospital

Manchester M23 9LT, UK

Tel +44 161 291 5881

Fax +44 161 291 5882

E-mail [email protected]

Running head: Comparing Expressed Emotion on the CFI and FMSS

Abstract word count 150

Word count 3,830

 2

Assessing Expressed Emotion: Comparing Camberwell Family Interview and Five

Minute Speech Sample ratings for mothers of children with behavior problems

ABSTRACT

Little is known of the concordance between ratings of expressed emotion (EE) derived from

the Camberwell Family Interview (CFI) and Five Minute Speech Sample (FMSS) for

parents of children with behaviour problems. Concordance between CFI and FMSS ratings

of EE was assessed prior to intervention and compared to parent-rated behaviour after

intervention, at follow-up, 12 months later. Female primary caretakers of 75 children (3-10

years) showing behavioral difficulties were interviewed using FMSS and CFI; interviews

were coded independently by criterion-standard raters. Using CFI, 57 families were

classified high, and 18, low EE. Using FMSS, 65 families classified high, and 10, low EE.

55/75(73%) pairs of ratings were the same (high, n = 51: low, n = 4). 20(27%) mothers

were allocated different EE status (Kappa = .14, ns). FMSS at initial interview appeared

more closely related to behaviour rating at follow-up than CFI. Further investigation is

required to establish comparability of CFI and FMSS results for carers of children.

KEYWORDS: expressed emotion; child; behavior; parent; family

 3

ABBREVIATIONS

CD Conduct Disorder
CFI Camberwell Family Interview
EE Expressed Emotion
EOI Emotional Over-Involvement
ECBI Eyberg Child Behaviour Inventory
FMSS Five Minute Speech Sample
ODD Oppositional Defiant Disorder
 4

INTRODUCTION

Expressed emotion (EE) is used to describe the emotional quality of relationships, and is

generally assessed using the Camberwell Family Interview (CFI) or Five Minute Speech

Sample (FMSS). The CFI-EE rating (Vaughn and Leff, 1976a) has been extensively

researched (Wearden, Tarrier, Barrowclough et al, 2000), but is lengthy to interview and

code; the FMSS (Magana, Goldstein, Karno et al, 1986) is quicker. Although the majority of

studies of EE in families of children and adolescents use FMSS methodology, (eg. Daley,

Sonuga-Barke, & Thompson, 2003; Jacobson, Hibbs & Ziegenhaim 2000; Peris & Baker,

2000), the extent to which these procedures yield the same EE classification for parents

talking about children is not clear. A parenting intervention (Harrington, Peters, Green et al,

2000) where both CFI and FMSS were used enabled examination of the association between

the measures. Relationship of EE rating to parent rated behaviour scores at follow-up was

examined.
 5

METHODS

Participants

The National Health Service Local Research Ethics Committee granted approval.

Consecutive referrals of 75 children reaching DSM IV criteria for Conduct Disorder (CD) or

Oppositional Defiant Disorder (ODD) were identified by child mental health service team

members in the North West of England, UK. The decision on whether children met criteria

for ODD or CD was made by a senior psychiatrist or clinical psychologist, prompted by a

standardised checklist of DSM IV criteria including symptoms (Harrington et al, 2000).

Primary diagnosis was a follows: ODD, 35 children (46.7%), ADD, 5 (6.7%), ADD and

ODD, 10, (13.3%), CD and ODD, 5, (6.7%), CD and ADD, 5, (6.7%), “behaviour/parenting

difficulties”, 15 (20%). Thirty three children (44%) were originally primary care referrals to

a child psychiatry outpatient department at a major teaching hospital; 23 (31%) were

referred by clinical psychology services; 8 (11%) by community paediatrician; 7 (9%) by

social services, 3 were referred directly to the study by their health visitor and one by

educational services. All the families who were approached agreed to take part (N = 75).

Age when parent first noticed a problem ranged from birth (n = 11, 15%) to 84 months (7

years); (n = 2, 3%). Median age of onset was 24 months (2 years). Time between onset and

date of referral to the study ranged from 8 months to 114 months (9 ½ years) with a mean of

54 months (4 ½ years). Older children were rated as having had a behavioral problem for

longer (r = .44, p<.001). Age of onset was unrelated to child’s current age (rho = .23, ns).
 6

The majority of the children (n = 71) were either in full-time mainstream education or

attended a pre-school nursery. Two children had been excluded from school because of

behavioral difficulties. Two children were too young to attend school. Six families (8%)

were receiving disability living allowance because of the severity of the child’s behavioral

difficulties. Eight children (11%) were being prescribed methylphenidate (Ritalin) as

treatment for attention deficit disorder at the time of referral.

Family structure For most children (n = 73, 98%) the primary carer was their biological

mother. One stepmother and one adoptive mother participated. Forty-two mothers (56%)

were currently married or cohabiting. Of the 33 single mothers, 14 (19%) were divorced or

separated; one was widowed. Most (83%) children lived with another sibling. Most

commonly, this was one sibling (42%); 22 (29%) with two siblings; 9 (12%) 3 or more

siblings, 13 (17%) no siblings and one child was a twin. Thirty (40%) were eldest; 15 (20%)

youngest, and 16 (21%), middle children. Thirty-five children had always lived with

biological father (47%); 16 (21%) had regular contact with their own father. Of these, 2

children lived with a stepfather. Seven children lived with a step- or adopted father, but had

no contact with biological father. Seventeen (23%) children had no regular contact with

father or a stepfather.

Age of mothers and children Mothers’ ages ranged from 22 to 44 years (mean 31 years; sd =

5). Mean age of the children was 83 months (sd = 19) with a range of 36 to 120 months.

Eighty-one percent (N = 61) were male, 19% (N = 14) female.

 7

Parental work and socio-economic status The study took place in an area with high

deprivation indices. Socio-economic status (SES) was calculated for each family by

converting the highest-ranking current or last main paid occupation of the mother or father

to a classification of social class using the Standard Occupational Classification (Office of

Population, Censuses and Surveys, 1995) . (See Table 1).

Table 1 about here

Maternal interview

Mothers were interviewed at home; interviews were tape-recorded. Mothers were

interviewed first using the FMSS procedure, then the CFI. Mothers were interviewed where

possible in the absence of other family members; no interviews were recorded in the

presence of the index child. Assessments were repeated at follow-up, 12 months later.

Assessment of CFI-EE

As suggested by Scott and Campbell (2001) the CFI interview schedule was modified to be

applicable to younger children (Bolton, Calam, Barrowclough, Peters, Roberts, Wearden &

Morris, 2003). Questions about daily routine were altered in order to make these age appropriate

for the children in the study. This included for example, waking, dressing, mealtimes, after-

school activities and bedtime. Symptoms in the original CFI were changed to reflect common

emotional and behavioral disorders in children, for example sleep-related problems, non-

compliance, aggression or violence, temper tantrums, over- or under-activity, and anxiety and

depression. Sections on household tasks and finances were removed. Areas from the
 8

"relationship” section of the CFI were made age appropriate. Relationships with close adults

other than parents were also explored. Guidelines for rating EE from the CFI are summarised

below (see Leff and Vaughn, 1985).

Rating EE from the CFI

Criticism The criticism scale comprises a frequency count of critical comments. A critical

comment is defined as “a statement which, by the manner in which it is expressed,

constitutes an unfavourable comment upon the behavior or personality of the person to

whom it refers” (Leff and Vaughn, 1985, p38). Critical comments are determined by the

content of the statement e.g. She acts like a horrible spoilt brat. If tone is sufficiently

critical, a statement can qualify as a criticism regardless of content. In this study the

conventional threshold of six critical comments was used.

Hostility Hostility is rated on a four-point scale. Although evidence for the rating comes

from the entire interview, a single hostile comment is sufficient to justify a rating. Hostility

indicates dislike or rejection of the child as a person. Statements that are rated for hostility as

rejection indicate a frank dislike about the child e.g. I leave him with his dad ‘cause at times

I can’t bear to be near him.

Emotional overinvolvement (EOI) is rated on a six-point scale based on evidence provided

during the interview. Scores of 3 or above lead to a high EE rating. Evidence for EOI comes

from behaviors or statements of attitude that the mother reports and from her behavior as

observed during the interview. Reported behavior includes: i) exaggerated emotional

 9

response e.g. I couldn’t sleep because I kept worrying if he would get enough to eat at

school and about how he’d cope with those new strange children. ii) self-sacrificing

behavior and iii) overprotective behavior. Observed over concern is also rated and includes

emotional display and dramatisation.

Adaptation of EE-CFI methodology for current study In addition to ensuring the CFI

schedule was age appropriate, modifications were made to the guidelines for rating EOI

(Bolton et al 2003). These modifications were made following discussions with Christine

Vaughn (first author of the original abbreviated CFI) and are similar to those reported by

(Hodes, Dare, Dodge and Eisler, 1999) when assessing EE in children using the CFI

methodology (details available on request). An example of age-specific behavior was young

children sleeping in their mother’s bed. This was such common practice within this sample

that it was decided not to include this evidence. Emphasis was placed on parental

preoccupation and style of speech (e.g. melodramatic speech and excessive detail) rather

than behavior towards the child. Guidelines were adapted to ensure raters agreed on the

modified rating of EOI.

Although warmth and positive comments were rated, they do not contribute towards the

final EE index, and are not described here. Reaching threshold on criticisms, hostility or

EOI results in a high EE index; otherwise a low EE rating is made.

 10

Inter-rater reliability of CFI-EE methodology Inter-rater reliability was assessed in two

stages. Firstly, following training in the traditional rating procedures and secondly, with

another trained CFI rater to ensure reliability with the modified interview and rating system.

Training in administration and rating CFI-EE One author (SP) was trained by Christine

Vaughn in the administration of the CFI and EE coding system. Inter-rater reliability for

criterion scales were all above 0.84; good reliability was achieved for all scales and full

agreement for overall EE classification (Table 2).

Table 2 about here

Inter-rater reliability of adaptations (N = 11) In order to ensure acceptable inter-rater

reliability using the modified interview schedule and the changed guidelines for rating EOI,

eleven interviews with mothers from the current sample were re-coded by a second trained

rater using the same system in a similar population (Bolton et al, 2003). Satisfactory inter-

rater reliability for the criterion scales was achieved (see Table 2).

Assessment of FMSS-EE

FMSS was always collected before the CFI following the exact administration guidelines

(Magana-Amato, 1989), where the participant is asked to speak without prompting for 5

minutes. All speech was audiotaped and subsequently transcribed. Although not all mothers

spoke for the full five minutes, all samples contained sufficient material to permit coding.
 11

Rating EE from the FMSS

The FMSS is rated along the following scales: i) initial statement (positive, neutral or

negative); ii) critical comments (frequency count); iii) quality of relationship (positive,

neutral or negative); iv) statements of attitude (frequency count); v) emotional display

(presence/absence); vi) evidence of self-sacrificing, overprotective behavior or a lack of

objectivity; vii) excessive detail and viii) frequency of positive remarks. Presence of

dissatisfaction is also noted and contributes towards a borderline rating of critical EE though

in isolation is insufficient to result in a high EE rating. The definitions of critical comments,

positive comments and evidence for EOI are the same as those described earlier for rating

EE using the CFI. Individual scales have been used as continuous variables (Lenior,

Dingemans and Linszen, 1997).

All FMSSs were transcribed by one of the authors (SP) and rated (from audio and transcript)

by an experienced criterion rater (Zaden, University of California, US) who was blind to all

information about the mothers and the study hypotheses.

Inter-rater reliability of FMSS-EE ratings Many of the mothers in this study had strong

regional accents and used colloquialisms that were unfamiliar to the US coder. A subset of

20 speech samples were randomly selected and re-coded by a second approved rater for

purposes of reliability. Raters achieved 100% agreement of high/low status and for

classification of CFI subgroups (low, high-critical, high-EOI or high critical and EOI),

(Kappa = .97, CI = 0.66,1.13).

 12

Behaviour rating

Eyberg Child Behavior Inventory (ECBI, Robinson, Eyberg and Ross, 1980) is made up of a

list of 36 common behavioural problems. Parents respond yes or no to each item to record

whether each of these is currently a problem with their child, and rate how often the specific

behavior occurs (1-7, never to always respectively). Scores are totaled for each scale,

yielding separate Problem and Intensity scores.

 13

RESULTS

Of the 75 pairs of ratings, 55 (73%) were allocated the same rating of either high (n = 51) or

low (n = 4) EE using the two methodologies. Twenty mothers (27%) were allocated a

different EE status depending on which method of rating was used. This level of agreement

between ratings was not statistically significant (Kappa = .14, ns, CI = 0.1, 0.38). Table 3

presents overall ratings for each method.

Table 3 about here

Using the CFI methodology (CFI-EE), 57 families were classified as high EE and 18

families were rated as low EE. Using the FMSS methodology (FMSS-EE), 65 families were

classified as high and 10 were low. Six of the 57 mothers who were rated as high CFI-EE

were classified as low FMSS-EE whilst 14 low CFI-EE mothers were rated as high FMSS-

EE.

Due to the small numbers of low EE families, numbers were too small to test for predictive

validation with respect to outcome. However, simple t tests showed that ECBI Intensity at

follow-up was significantly lower in families initially classified as low FMSS-EE, compared

to those classified high EE (t = -2.08, df 59, p = 0.042). A significant difference was also

seen when a similar analysis was carried out using CFI-EE (t = -2.419, df 56, p = .019).

Both approaches therefore identified differences in EE at initial interview that was

associated with maternal ratings of outcome at follow-up. For descriptive purposes, Table 3
 14

shows the mean ECBI Intensity scores for follow-up for those families for whom ECBI data

were available, 12 months later. Initial interview data are included to show baseline. These

data again show that both approaches are related to clinical outcome, but that a high FMSS-

EE/low CFI-EE is more likely to be associated with higher ECBI scores than is the converse,

perhaps indicating higher sensitivity of the FMSS.

The correspondence between CFI and FMSS ratings was further examined by investigating

the level of agreement in each of the subtypes of the EE rating (Table 4). The coding

schemes for the FMSS and CFI are not identical e.g. in the CFI methodology hostility

contributes towards a classification of high EE. Two subscales are unique to the FMSS

methodology but contribute towards a high EE rating (initial statement and quality of

relationship). Only subscales that were common to both measures were included in the

analysis: critical, EOI and critical and EOI. Figure 1 presents the distribution of critical

comments rated from CFIs for the sample.

Figure 1 about here

For 28 cases (37%), subscale ratings were the same. Misses on the FMSS (n = 22) were 7

critical cases and 15 EOI cases. In contrast, false positives (n = 23) made using the FMSS

were 14 critical cases and 9 EOI cases.

Table 4 about here

 15

Thirteen cases classified on the FMSS as high EE on the basis of criticism were identified as

low EE by the CFI. Because most criticisms are made during the earlier parts of the CFI

(Leff and Vaughn; 1985), an analysis was conducted where criticisms rated during the

FMSS were included as critical comments using the CFI rating. On this analysis, 7 mothers

then reached threshold on critical comments and were reclassified as showing high EE on

the CFI. Four (5.3%) were classified low by both systems; 58 (77.3%) were classified high

by both, with 7 (9.3%) rated low on CFI but high on FMSS and 6 (8.0%) rated high on CFI

and low on FMSS. This level of agreement between ratings (62/75, 83%) then reached

statistical significance (Kappa = .28, p<.05, CI = 0.01, 0.55) though this remains a low

level of reliability.
 16

DISCUSSION

This study aimed to examine the extent to which EE ratings derived from the CFI and the

FMSS were equivalent. The study used a sample of primary female caregivers of children

referred for behavioural difficulties drawn from a deprived urban environment, who would

be expected to represent a commonly occurring group presenting for child clinical services.

CFI-EE ratings were made using the established CFI procedure and coding scheme, with

some modifications in order to make it age appropriate for parents of 3 to 10-year-old

children (Bolton et al 2003). Modifications were approved by one of the originators of the

approach, and ratings were made by coders trained to criterion standards. FMSS-EE ratings

were made by an independent criterion rater blind to the purpose of the study.

On the basis of the CFI approach, over three quarters of mothers were classified as high EE.

Ratings were primarily made for criticism with only one mother rated high CFI-EE on the

basis of EOI only. Using the FMSS methodology, almost 90% of the sample was rated as

high EE. This finding is contradictory to previously published work that found either good

levels of agreement (Magana et al, 1986) or a trend towards under-reporting in the FMSS

(Malla, Kazarian, Barnes & Cold, 1991; Van Humbeeck, Van Audenhove, De Hert et al,

2002). We found the FMSS methodology had a lower, rather than higher threshold for EE.

Because the numbers of families classified as showing low EE were small, and we could not

therefore undertake analysis, there was some indication that families rated as high on the
 17

FMSS-EE but low on CFI-EE were more likely to rate their children as showing higher

levels of behavioural difficulty at follow-up than the converse. In choosing between the two

approaches, therefore, researchers should be aware that the pattern of findings across studies

is not consistent for families with children. The mixed evidence for the predictive power of

the FMSS for families with children (Asarnow, Goldstein, Tompson & Guthrie, 1993);

Kershner, Cohen and Coyne, 1996; Van Humbeeck et al, 2002) is a factor that needs

consideration in decisions on methodology.

One aspect of the current study which should be noted is the very high levels of EE we

found. When compared to other published samples of parents of younger children, it is clear

that regardless of methodology used for assessment, the level of high EE in this sample was

considerable. Some studies of families with children and adolescents have lowered the cut-

offs used for the classification of EE. For example, Vostanis, Nicholls and Harrington

(1994) used a threshold to five or more critical comments and Hodes, Garralda, Rose and

Schwartz (1999) reduced theirs further to four or more critical comments. If we had

followed the models adopted by these studies, the levels of EE and the discrepancy in

classification between the CFI and FMSS would have been even greater.

The study has a number of strengths. All the children had been referred, assessed and

diagnosed independently of the study. We had a complete inclusion rate for parents invited

to take part in the study. The minor modifications to the CFI interview to take account of the

age of the children were approved by one of the originators of the CFI. The CFI ratings were

made by psychologists trained to reliability by criterion raters, and the FMSS was fully
 18

independently coded by a criterion rater in the USA. This gives a high level of confidence

that the ratings yielded by the procedures were conducted to the highest possible standard,

and highly consistent with literature published by other experts in the field.

The level of agreement between the two approaches to the measurement of EE was such that

the methodologies should not be assumed to be interchangeable. It is important to examine

possible reasons for this. One suggestion for the low level of agreement found is that it was

to some extent an artefact of the research procedure. On close examination of the pattern of

agreement of subtypes of ratings, it is evident that the largest number of discrepancies was

on the category of criticism. In order to prevent contamination from interview or

questionnaires, it had been decided to always administer the FMSS first, immediately

followed by the CFI. This may have led to a lowering the number of critical comments

spoken during the CFI. Some support for this explanation was found when the level of

agreement between the two methodologies was re-examined including FMSS-EE criticisms

in the CFI-EE rating. Analysed this way, there was agreement on classification in 83% of

families. It is important to note however, that while the level of agreement reached

statistical significance, reliability was still relatively low. Repeating this study, but

collecting the two EE assessments on separate occasions, could clarify this question.

It would be valuable to know more about the performance of these two measures across different

samples. Based on our ratings, regardless of methodology, the percentage of families falling into

the high EE category was extremely high. Other studies using child psychology and psychiatry

referrals have found lower levels. It would be valuable to compare these measures on a larger
 19

sample which incorporates a wider range of families with different levels of EE. There may also

be differences depending on the nature of the referred problems. The parents in this sample were

very keen to emphasise the level of difficulties that they were experiencing with their children,

and often made extremely negative comments about them. They were also drawn from an area of

high deprivation, which limits the generalizability of our findings, and indicates a need for

replication with other samples, with a good sized follow-up. Normative data on EE across

different demographic characteristics, including age of the child, membership of cultural group,

and socio-economic status would be helpful in identifying groups which are more likely to show

elevated levels of EE, and would provide a framework for developing systematic rules if it is

necessary to employ different thresholds to use within different populations. Certainly, the high

levels found in our sample compared to other mental health clinic samples would indicate that

this kind of framework for classifying families as high EE is needed, particularly as the

application of the approach is widened out and modified for use with samples other than patients

with presentations of schizophrenia for which it was originally designed.

An important assumption is that the way in which parents talk about their child reflects

important aspects of the way that they interact with their child in everyday life. Studies

should establish the extent to which the various dimensions of EE relate to actual parenting

behavior. For example, McCarty, Lau, Valeri & Weisz (2004) found support for the validity

of criticism but not EOI when observing actual parent child interactions. Calam, Bolton,

Barrowclough & Roberts (2002) reported that CFI criticism ratings were significantly

correlated with independent clinician ratings of perceived parent maltreatment potential.

The number of studies of EE in families with children which make these kinds of
 20

comparisons are extremely limited, and it was beyond the scope of the present study to

investigate this aspect.

Our findings would appear to indicate that, at least in the sample we used, the use of the

FMSS might lead to elevated levels of inclusion of families in the high EE category. This is

potentially important, given that the FMSS has been the approach that has been more widely

adopted across a range of studies of parents and children, both in relation to physical and

mental health difficulties. Further exploration of methodological aspects of the application

of different approaches to classification of EE in families with children would be both

valuable and timely.

 21

STATEMENT ON AUTHORS

Sarah Peters collected the data for this study as part of her PhD while a student in the

Division of Child Psychiatry at the University of Manchester. Rachel Calam and Sarah

Peters were both responsible for the conception and design of the study. Both have

contributed to the drafting and revision of the manuscript, and both have approved the final

version.

ACKNOWLEDGEMENTS

The authors would like to thank Christine Vaughn for providing comment on CFI
modifications; Catherine Bolton and Sibyl Zaden for help with coding, the clinicians for
assistance with the sample, and all the families who gave so generously of their time and
experiences. The contribution of the late Dick Harrington to all aspects of this research is
warmly acknowledged.
 22

REFERENCES

Asarnow JR, Goldstein MJ, Tompson M, Guthrie D. One-year outcomes of depressive

disorders in child psychiatric in-patients: Evaluation of the prognostic power of a brief

measure of expressed emotion J Ch Psychol Psychiat 1993; 34: 129-137.

Bolton C, Calam R, Barrowclough C et al. Expressed emotion and causal attributions in

mothers of children with problem behavior J Ch Psychol Psychiat 2003; 44: 242-254.

Calam R, Bolton C, Barrowclough C, Roberts J. Maternal expressed emotion and clinician

ratings of emotional maltreatment potential. Ch Abuse and Negl 2002; 26: 1101-6.

Daley D, Sonuga-Barke EJ, Thompson M. Assessing expressed emotion in mothers of

preschool AD/HD children: psychometric properties of a modified speech sample. Br J Clin

Psychol 2003; 42: 53-67.

Harrington RC, Peters S, Green J, Byford S, Woods J, McGowan R . Randomized

comparison of community- and hospital-based mental health services for children with

behavioral disorders BMJ 2000; 321: 1047-1050.

Hodes M, Dare C, Dodge E, Eisler I. (a) The assessment of expressed emotion in a

standardised family interview J Ch Psychol Psychiat 1999; 40: 617-625.

 23

Hodes M, Garralda ME, Rose G, Schwartz R (b). Maternal expressed emotion and

adjustment in children with epilepsy JJ Ch Psychol Psychiat 1999; 40: 1083-1093.

Jacobson T, Hibbs E, Ziegenhaim. Maternal expressed emotion related to attachment

disorganization in early childhood: a preliminary report. J Ch Psychol Psychiat 2000; 41:

899-906.

Kershner JG, Cohen NJ, Coyne JC. Expressed emotion in families of clinically referred and

nonreferred children: toward a further understanding of the expressed emotion index J Fam

Psychol 1996; 10: 97-106.

Leff J,& Vaughn C. Expressed Emotion in Families: Its significance for mental illness New

York: Guildford, 1985.

Lenior ME, Dingemans PMAJ, Linszen DH. A quantitative measure for expressed emotion

Psychiat Res 1997; 69: 53-65.

Magana-Amato A. Manual for coding expressed emotion from the Five Minute Speech

Sample. Los Angeles: UCLA Family Project 1989.

Magana AB, Goldstein MJ, Karno M, et al. A brief method for assessing expressed emotion

in the relatives of psychiatric patients. Psychiat Res 1986;17: 203-212.

 24

Malla AK, Kazarian SS, Barnes S,Cold JD. Validation of the Five Minute Speech Sample in

Measuring Expressed Emotion. Can J Psychiat. 1991; 36: 297-299.

McCarty CA, Lau AS, Valeri SM, Weisz, JR. Parent-child interactions in relation to critical

and emotionally overinvolved expressed emotion (EE): Is EE a proxy for behavior? J Ab

Ch Psychol 2004;32: 83-93.

Office of Population, Censuses and Surveys. Standard Occupational Classification Volumes

2 and 3. 2nd Ed London: HMSO 1995.

Peris TS. Baker BL. Applications of the expressed emotion construct to young children with

externalizing behavior: Stability and prediction over time J Ch Psychol Psychiat 2000; 41:

457-462.

Robinson EA, Eyberg SM. Ross AW. The standardization of an inventory of child conduct

problem behaviors. J Ch Psychol Psychiat 1980;) 9: 22-9.

Scott S & Campbell C. Expressed emotion about children: reliability and validity of a

Camberwell Family Interview for Childhood (CFI-C). Int J Meths Psychiat Res 2001; 9: 3-

10.
 25

Van Humbeeck, G, Van Audenhove, Ch, De Hert, M, et al. Expressed emotion: A review of

assessment instruments. Clin Psychol Rev 2002; 22: 321-341

Vaughn CE, Leff JP (a) The measurement of expressed emotion in the families of

psychiatric patients. Brit J Soc & Clin Psychol 1976; 15: 423-429.

Vaughn CE, Leff JP (b) The influence of family and social factors on the course of

psychiatric illness: a comparison of schizophrenic and depressed neurotic patients. B J

Psychiat 1976; 129: 125-137.

Vostanis P, Nicholls J & Harrington R. Maternal expressed emotion in conduct and

emotional disorders of childhood. J Ch Psychol Psychiat 1994;35: 365-376.

Wearden AJ, Tarrier N, Barrowclough C et al. A review of expressed emotion research in

health care. Clin Psychol Rev 2000; 20: 633-666.

 26

LEGENDS

Figure 1: Frequency of critical comments measured from the CFI

Table 1: Employment and socio-economic status

Table 2: Reliability of ratings as assessed with other trained raters

Table 3: Relationship between CFI and FMSS classifications of EE, showing mean ECBI
Intensity scores at initial interview and follow-up

Table 4: Agreement between EE ratings on the CFI and FMSS

 27

Figure 1: Frequency of critical comments measured from the CFI

 28

Table 1: Employment and socio-economic status

n % n %
Mother’s work status (n Family Benefits (n = 75)
= 75)
Full time parent 40 53 Income support/family 41 55
credit
Full time employment 8 11 Sickness benefit 8 11
Part time employment 20 27 No low income benefits 26 34
Sickness/invalidity 7 9
benefit
Father’s work status (n Family Social class (n =
= 42) 75)
Full time parent 3 7 I Professional 1 1
Full time employment 22 52 II 15 20
Managerial/technical
Part time employment 3 7 III Skilled non- 18 24
manual
Self-employed 3 7 III Skilled manual 13 17
Sickness/invalidity 9 22 IV Partly skilled 21 28
benefit
Unemployed 2 5 V Unskilled 5 7
Never worked 2 3
 29

Table 2: Reliability of ratings as assessed with other trained raters

Scale Schizophrenia Child Sample 2

Sample1
N = 12 N = 11
Overall EE 100% agreement 100% agreement
classification
(high/low)
Critical rho = 0.89 rho = 0.87
Comments
Hostility  = 0.84  = 0.67
(High/low) (1 disagreement) (2
disagreements)
EOI (high/low) 100% agreement  = 0.77
(1 disagreement)
EOI (ordinal not assessed rho = 0.64
data)
Warmth r = 0.77 r = 0.77

Reliability undertaken with: 1 C.Vaughn; 2 C. Bolton (Bolton et al, 2003)

 30

Table 3: Relationship between CFI and FMSS classifications of EE, showing mean ECBI
Intensity scores at initial interview (n=71) and follow-up (n=61)

CFI EE Initial
interview
Low High
Mean SD Mean SD
FMSS ECBI Intensity Initial 162.4 26.7 162.4 31.7
Initial interview
interview
low
ECBI Intensity Follow-up 124.9 23.0 135.6 38.3

FMSS ECBI Intensity Initial 173.1 34.7 183.4 30.1

Initial interview
interview
high
ECBI Intensity Follow-up 155.7 45.9 162.0 39.8
 31

Table 4: Agreement between EE ratings on the CFI and FMSS

CFI Ratings
Low Critical EOI Critical N=
and EOI
Low 4 6 0 0 10
FMSS Ratings

Critical 13 22 0 15 50
EOI 1 2 0 1 4
Critical 0 6 1 4 11
and
EOI
N= 18 36 1 20 75