Children's Friendship Training

Fred D. Frankel, Robert J. Myatt

• Publisher: Routledge
• Year: 2013
• ISBN:
  9781135451516
• Edition: 1
• Topic: Psychologie
• Subtopic: Psychologie appliquée

First published in 2003. Children's Friendship Training is a complete manualized guide for therapists treating children with peer problems. This unique, empirically validated treatment is the first to integrate parents into the therapy process to ensure generalization to school and home.Representing over twelve years of research, Children's Friendship Training presents the comprehensive social skills training program developed by these pioneering authors. Step-by-step interventions help children develop the skills to initiate mutually satisfying social interactions. These interactions can lead to higher regard within the peer group and the development of satisfying dyadic relationships that will, in turn, serve to enhance overall well being. Clinical and empirical rationales, illustrative case examples and parent handouts that educate parents and give specific guidelines for homework assignments are presented for each treatment module. Brief relevant reviews of the child development literature and selective reviews of assessment techniques and other approached to children's social skills training are presented to sufficiently acquaint therapists interested in implementing children's friendship training.

P A R T

I

BACKGROUND

C H A P T E R

1

Reasons to Treat Peer Relationship Problems

Operating within an outpatient child psychiatry clinical training program, we are constantly struck by the relatively little attention paid by many clinicians to the peer relationships of their patients. Children in therapy are twice as likely to have peer problems than children not in therapy (Malik & Furman, 1993; Rutter & Garmezy, 1983). Peer problems have been only peripherally included in diagnostic schema. Often, when the presenting psychopathology has been treated, the peer problem doesn’t go away and yet is usually not addressed by further treatment. It is estimated that as many as 10% of children without any known risk factors have problems making and keeping friendships (Asher, 1990). In a 1-year follow-up study of 10year-olds, Bukowski, Hoza, and Newcomb (1991, cited in Hartup, 1996) found that having friends had subsequent positive effects on self-esteem. Peer problems in childhood have effects on functioning in later life. In their 12-year follow-up study, Bagwell, Newcomb, and Bukowski (1998) found that peer rejection and lack of at least one best friend in childhood contributed equally to psychopathology of young adults.

We define a quality friendship as “a mutual relationship formed with affection and commitment between two individuals who consider themselves as equals.”¹ Maintaining quality friendships is perhaps the single most salient measure of a child’s successful adjustment. In order to have quality friendships, children must suspend egoism, treat the friend as an equal, and deal effectively with conflict (Hartup, 1996). Friends share a “climate of agreement” much greater than that among nonfriends (Gottman, 1983; Hartup & Laursen, 1993). Children’s problem solving is generally better when done with friends as compared with nonfriends (Azmitia & Montgomery, 1993). Friendships may moderate the negative impact of divorce (Wasserstein & La Greca, 1996). Friendships are the context for learning social skills, learning about and feeling good about oneself, and providing resources for support (Hartup, 1993). According to Malik and Furman (1993), “peers are not only playmates but also confidantes, allies and sources of support in times of stress” (p. 1303).

---

¹The best definition of friendship we have encountered.Unfortunately, the original author for this has been lost to us.

C H A P T E R

2

Assessment of Outcomes

It is difficult to evaluate the research literature on social skills training without first examining the measures used to demonstrate outcome. A frequent misconception among parents and therapists about formal outcome assessment within a clinical context is that assessment is always associated with some kind of research. Although rarely done, formal outcome assessment should play an integral part in clinical practice for the following reasons:

1. We and others have noted that even the best clinicians are subject to clinician drift over time. Perhaps because creative clinicians get tired of delivering the same intervention repeatedly, they are constantly injecting new features. They may consider these new features acceptable variations on a standard treatment. We have also noticed that our clinicians can successfully present an approach 40 or 50 times but are thrown when it doesn’t work quite as well just once. They may then modify their time-worn procedure without consideration of how successfully it usually has worked. We have detected these variations when outcome on teacher measures for a particular group drops (parent measures seem to be less sensitive to procedural variations).

2. We have attempted treatment upgrades on three occasions. On one attempt, we noticed that teacher outcome substantially worsened when compared to the previous groups. We reverted back to our original approach and recaptured previous levels of positive teacher outcome.

3. Posttreatment teacher calls are completed usually within 3 weeks after the last treatment session. We have found that about 30% of parents will call for this feedback. When parents call for this posttreatment feedback, we use the occasion to encourage parents to maintain treatment gains.

Types of Outcome Assessment

There are six types of outcome assessment procedures used by researchers and clinicians: peer assessment, informal “clinical” indexes, ratings from the child patients, behavioral ratings, teacher reports, and parent ratings.

Informal clinical indexes of improvement, such as unstructured parent feedback, clinician ratings of child behavior changes in session, and the clinician’s impressions global impression of change, have very little validity for the assessment of changes in peer acceptance of the rejected child: Parents are usually glad they brought their child to the groups and children almost always have a good time (or they leave and aren’t around for the posttest). When effective behavioral control techniques are used within the group, the children are better behaved. It is the generalization of these changes to the child’s so cial environment that is of paramount importance.

As a training and clinical tool, we ask the child group leaders to rate improvement and then compare these ratings with formal parent and teacher measures. The correlation is close to O. For instance, our clinicians will typically rate a child who disrupts the group or challenges the leader as unimproved. Yet parent and teacher ratings often show some benefit of treatment (this finding has served as impetus for our clinicians to persist with these children). The following subsections briefly review the types of formalized assessments that have been used to measure improvement in social skills training programs.

Assessment Using Peers Without Social. Problems

Peer assessment entails having members of a peer group evaluate each other. According to Gresham and Stuart (1992), peer ratings are the most frequently employed types of peer-referenced assessment in research studies. Many researchers consider this form of assessment as the most valid because peers frequently observe each other and operate from a child’s frame of reference (DanielsBeirness, 1989). As Landau and Moore (1991) pointed out, children are more aware of interpersonal interactions of their peers than are teachers and parents.

Peer assessments are too cumbersome to use in typical clinical outpatient settings. For example, our outpatient program draws from more than 200 different schools. Visiting each school and setting up peer ratings for each patient would be a logistic nightmare. Peer assessments are presented here in order to help the reader to better understand and evaluate research studies and categories of peer acceptance (see Chap. 9) that are derived from them. Two types of peer assessment are ratings and nominations.

In a peer nomination procedure, a classroom of elementary school children are asked to list the other children in their class they would like to play with the most and the least (Dodge, Coie, Pettit, & Price, 1990). Peer status is tabulated for each child using social preference (number liked nominations – number disliked nominations) and social impact scores (total number of nominations; Asher & Hymel, 1981; Coie & Dodge, 1983). Gresham and Stuart (1992) found that liked least nominations showed the most stability with test–retest correlations of .60. Extensive research (e.g., Coie & Dodge, 1983; Coie & Kupersmidt, 1983; Dodge, 1983) has indicated the stability of status and predictability of negative outcomes. There has been some concern that the negative nomination procedure might subsequently lead to more negative behavior toward low-status children (Asher & Hymel, 1981), although two studies have failed to show negative effects (Bell-Dolan, Foster, & Sikora, 1989; Bell-Dolan, Foster, & Christopher, 1992).

A peer rating procedure requires each child to rate each other child in his or her group using predefined criteria. A simple example is the “roster and rating” sociometric procedure (Singleton & Asher, 1977). Children are given rosters of all same-sex classmates and are asked to rate how much they like each one on a 5-point Likert scale with endpoint anchors of like a lot and dislike a lot. Other examples are, “How much does______ get picked on?” and “How much does______cooperate?” (endpoint anchors would be not at all and very much). Average rating is tabulated for each child.

Rating procedures are more apt than nomination procedures to pick up children who have low visibility but are enjoyed as playmates (Asher & Hymel, 1981). The average rating received from classmates (or from same-sex classmates) is highly stable over time even with young children (Asher, Singleton, Tinsley, & Hymel, 1979), is sensitive to the effects of intervention (Ladd, 1981; Oden & Asher, 1977), but is also sensitive to variations in the wording of the criterion (Oden & Asher, 1977; Singleton & Asher, 1977). Another important advantage of the rating-scale method is that children are not required to list anyone as disliked. Due to the advantages and disadvantages of each type of peer assessment, we have used both in our 5-year National Institute for Mental Health (NIMH)-funded study (Frankel & Erhardt, 2001). Individual interviews take about 8 minutes per student and most children seem to enjoy doing them.

Ratings from Child Patients

Ratings from child patients are easily obtained, but show the least correlations with other assessments and may be susceptible to social desirability response set (Ledingham, Younger, Schwartzman, & Bergeron, 1982). Children’s actual behavior may not correspond to their responses to queries about what they would do in hypothetical situations (Bearison & Gass, 1979; Damon, 1977). An example of this was a study by Grenell, Glass, and Katz (1987). They asked 15 boys diagnosed with attention deficit/hyperactivity disorder (ADHD) and 15 comparison boys what they thought they should do in 16 hypothetical situations. Adult judges rated their responses on friendliness, impulse control, assertiveness, and effectiveness in relationship enhancement. Raters also observed and rated them in free play, a cooperative puzzle task (where one boy was the worker and another was the helper), and a persuasion task (where they tried to persuade their partner to play their choice of game). Correlations of ratings of the responses to hypothetical situations with observed prosocial behavior were low (absolute value of correlations ranged from .37 to .43) although statistically significant.

Another example is the Preschool Interpersonal Problem Solving measure (PIPS; Spivack & Shure, 1974). The psychometric properties are adequate, although there are no normative data. Similar to the ratings of Grenell et al. (1987), the number and quality of solutions generated to various hypothetical situations are scored. However the content areas of the PIPS has been criticized as not representative of social problems encountered by preschoolers (Brochin & Wasik, 1992) and not adequately covering the domain of peer issues (Getz, Goldman, & Corsini, 1984). Others have suggested enhancements in responses to teasing (Feldman & Dodge, 1987) and management of conflict with other children (Brochin & Wasik, 1992).

Asking child patients to list who likes them seems also to be of limited value. Children without friends can usually list “friends” when asked (Hartup, 1996). Sociometrically rejected children were the least accurate in their judgments of who liked them, when compared to average and neglected status children (see Chap. 9, for more on these sociometric categories; MacDonald & Cohen, 1995). On the other hand, asking child patients to rate peer behaviors yields more valid results. Whalen and Henker (1985) reported similarities between their summer camp cohorts of 24 children diagnosed with ADHD and 24 nonADHD peers: Both cohorts rated their peers‘ negative behaviors similarly. Correlations between the cohorts ranged from .80 for “causes trouble” to .85 for “noisy.” Correlations with teacher ratings were comparable for the cohort diagnosed with ADHD (.69) and those that were not (.65). Hinshaw and Melnick (1995) reported systematic distortions: Boys diagnosed with ADHD were less likely to give positive nominations to non-ADHD peers than the non-ADHD peers were. Behavioral ratings may therefore be useful outcome measures in contexts where children with peer problems are mixed together with peers without social problems.

Although we have avoided the use of ratings from child patients, assessment of self-esteem has been an exception, because the child is uniquely qualified to be asked about perceptions of self-competence. We have used the Piers-Harris Self Concept Scale (PHS).² The PHS is an 80-item yes-no self-report measure that takes about 20 minutes for a child to complete. Among instruments commonly used to measure self-esteem, the PHS has been regarded as the most psychometrically sound. The coverage of the relevant domain is adequate (Ross, 1992), and the factor structure of specific self-esteem scales is well-known (Hughes, 1984; Jeske, 1985). The PHS manual provides factor scores on six scales measuring specific self-esteem (Piers, 1984) and a Global score that is a weighted composite of items from the specific self-esteem factors. Among children scoring in the low range of PHS self-esteem at baseline (only about 20% of children enrolled in our groups), we have noticed that about 80% show some improvement on posttreatment assessment.

Behavioral Ratings

Behavioral ratings of negative peer behaviors by adult observers may provide a objective measure of an important aspect of peer relations. An example is a study by Pelham and Bender (1982). They formed play groups of one child diagnosed with ADHD and four non-ADHD peers. The children played together for 36 minutes of arts and crafts (which involved sharing the same materials) and free play. Results showed that children diagnosed with ADHD were rated by observers as spending significantly more time in conversation and asking questions, and engaging in 2 to 10 times more negative behavior (e.g., loud repeated yelling, hitting, noncompliance, interrupting anther’s activity) than comparison children. Hinshaw and Melnick (1995) compared 101 boys diagnosed with ADHD with 80 non-ADHD peers during their participation in a summer camp. Observers rated several categories of behavior in free-play situations. Results showed that boys diagnosed with ADHD had significantly more rule violations, defiance, and disruptive behaviors than the non-ADHD cohort, regardless of the levels of aggression rated by teachers.

In contrast, behavioral ratings of prosocial events, such as, frequency of peer contact and percentage of time interacting with peers, have long been discredited as valid measures of peer adjustment (cf. Gottman, 1977). Correlations between these behavioral ratings and peer ratings have been low. Rate of peer interaction without consideration of quality of interaction is not related to measures of peer acceptance (Asher, Markell, & Hymel, 1981). For example, rejected children may “bug” others more in unsuccessful attempts at entry, whereas liked children may be successful on the first or second attempt (or take no for an answer, cf. Chap. 12). Behavioral measures of attempts at peer entry might therefore favor the rejected child. Valid behavioral measurement of treatment success may entail focus on “critical events” that occur too rarely to be observed under usual procedures (Bierman, Smoot, & Aumiller, 1993).

We have not employed behavioral observations in judging outcome to our interventions, due to severe limitations on their usefulness. Training observers and data reduction involve an extensive time commitment, with only limited reward within an outpatient clinical setting.

Teacher Reports

Teacher reports of peer relationships correspond more closely than any other informant to peer ratings (Glow & Glow, 1980). Hinshaw and Melnick (1995) found that asking teachers and parents to rate their child’s popula...