A Treatment without Empirical Support
Jean Mercer - Richard Stockton College
Correspondence concerning this article should be addressed to Jean Mercer, Ph.D., Richard Stockton College, Pomona, NJ 08240; E-mail: firstname.lastname@example.org.
Attachment therapy (AT) is a mental health intervention for children that involves physical restraint and discomfort. Practitioners base its use on the assumption that rage resulting from early frustration and mistreatment must be provoked and released in order for the child to form an emotional attachment and become affectionate and obedient. Death and injury have resulted from AT, which has nevertheless been supported by some state agencies. AT practitioners have claimed that research evidence supports the effectiveness of their techniques. In the present paper, the research evidence is examined with respect to research design and statistical analysis, and it is concluded that AT remains without empirical validation.
Few events have raised so many questions about the validity of a mental health intervention as the death of 10-year-old Candace Newmaker during a therapy session in April 2000 (Crowder, 2000). The conviction in a Colorado court of the two principal therapists in the case made national news a year later. Connell Watkins and Julie Ponder were each sentenced to 16 years' imprisonment on charges related to Candace's death (Lowe, 2001).
Watkins and Ponder were carrying out an exercise called "rebirthing" as part of their practice of attachment therapy (AT). Candace's adoptive mother, Jeane Newmaker, had brought Candace to Colorado on the advice of therapists in North Carolina, seeking treatment for symptoms she believed to be caused by an attachment disorder. According to testimony at the Watkins-Ponder trial, a therapist in North Carolina had diagnosed Candace with Reactive Attachment Disorder (RAD) on the basis of a questionnaire filled out by Jeane Newmaker.
At the trial, as well as in published material and on associated Web sites (Randolph, 1997a), practitioners and advocates of AT asserted that research evidence existed supporting the treatment. Some government agencies and insurance companies have apparently agreed with this claim. States have appropriated funds for the practice and teaching of AT (New Hampshire Executive Council Minutes, 1999), and testimony at the trial referred to payment for treatment through health insurance. There has been little formal opposition to AT. Only a few clinicians (Hanson & Spratt, 2000; James, 1994; Lieberman & Zeanah, 1999) have published criticisms of AT; legislation attempting to control the practice has been passed in only one state, Colorado, and, as of this writing, has been proposed in Utah.
The present paper will (a) present a brief description of the practice and theoretical rationale of AT, (b) note some inherent problems AT presents with respect to empirical validation, (c) summarize the research evidence offered by AT advocates, and (d) argue that AT should be identified as an unvalidated treatment (Mercer, 2001) as well as a potentially dangerous one.
Attachment Therapy: Its Purpose, Practice, and Rationale
AT (also known as holding therapy, rage-reduction therapy, and Z-process therapy, among other terms) was initially presented as a treatment for autistic children, although some practitioners claimed success with a variety of conditions, including acne (Zaslow & Menta, 1975). Currently, AT is used for children who are considered to be emotionally disturbed as a consequence of early difficulties with attachment experiences, which AT practitioners believe include premature birth. Many of the children are adopted; AT therapists consider all adopted children to need AT (Levy & Orlans, 2000). The children have most often been diagnosed with RAD, for which there are DSM-IV criteria, but some practitioners believe that there is a more severe and different form of attachment disorder (AD, in their terms), involving a combination of RAD and other features such as Oppositional Defiant Disorder (Randolph, 2000). Descriptions of the children often include details that imply severe disturbance. Fire setting and such acts of cruelty as tearing the heads off puppies are frequently mentioned (although, as we will see later, few if any complete case histories have been reported). Some proponents of AT have predicted that without this treatment, affected children will become serial killers, like Ted Bundy (Thomas, 2000).
The actual practice of AT differs among practitioners, but certain features appear to be consistently present. One is holding therapy, in which the child is restrained in the arms of one or more therapists, who attempt to trigger the expression of rage. As this was shown in a therapy videotape during the Watkins-Ponder trial, the child's face was grabbed, her head was shaken and bounced, and a therapist shouted into the child's face and demanded that she shout back. The therapist was provocative and insulting, calling the child a "twerp" and a liar and threatening her with abandonment by her adoptive mother. Each of these sessions lasted an hour or more.
Although holding therapy is an important feature of AT, practitioners vary greatly in their opinions about details of the treatment. Some, like Welch (1989), have held that holding should be done every day by parents of normal children as well as for therapeutic purposes. Others, such as Delaney and Kunstal (1993), have regarded holding as a treatment of last resort and have cautioned practitioners to check on legal and insurance guidelines, and never to threaten a child with abandonment or use other excessive provocation.
Therapeutic Foster Parenting
A second consistent component of AT is therapeutic foster parenting (Thomas, 2000). Children undergoing this aspect of the treatment are separated from their parents and live in foster homes, where foster parents drill them in compliance to orders and in such practices as "strong sitting," which involves sitting tailor-fashion on the floor, without moving, for up to two hours or longer. Withholding of food, performance of heavy chores, and other "boot camp" practices form part of the child's experience.
Rebirthing, the practice during which Candace Newmaker died, is not invariably used in AT. In Candace's case, rebirthing involved being wrapped in a flannel sheet while lying on the floor, having pillows placed on her, and being leaned on by four or five adults. She was to emerge from the sheet by her own efforts and thus "experience a rebirth" as the child of the adoptive mother, who was present and participating. Candace could not escape and was not released despite her screams and pleas for help. Her efforts were apparently blocked in some way, for the enveloping sheet sustained a long tear as a result of her struggles. When she was unwrapped after 70 minutes, the last 30 without a sound or movement, she was found to have suffocated. (According to courtroom testimony, other children given this treatment had been kept wrapped for only about 5 minutes.)
Despite frequent assertions of AT writers and their use of certain vocabulary, there is no demonstrable connection between AT and the attachment theory of John Bowlby (1982). The theory underlying attachment therapy has connections with the ideas of Wilhelm Reich (1945), who stressed eye contact and physical manipulation of the patient, as well as with some of the later Transactional Analysts (Schiff, 1970) and other advocates of New Age thinking (Emerson, 1996). As described by such AT writers as Foster Cline (1992), this approach assumes that the emotional attachment of a child to a parent begins before birth and is continued postnatally by a lengthy cycle of experienced frustrations followed by gratification when the parent feeds or cares for the baby and makes eye contact at the same time. Prenatal rejection or postnatal lack of gratification are thought to cause affectional attachment to be blocked by a buildup of unexpressed rage (see Tavris, 1989, for a better-supported opposing view of anger). Not only the child's lack of affection for the parent, but many problems (disobedience, poor language development, school failure, writing reversals, Candace's inability to emerge from the sheet) are attributed to blocked rage; that is also associated with failure to make eye contact when the parent wants it (Cline, 1992; Welch, 1989; Zaslow, 1966; Zaslow & Menta, 1975). The pain and terror experienced during holding are thought to trigger the catharsis of rage and to unblock the capacity for attachment, after which the child will be affectionate, cheerful, grateful, and obedient. If the child resists, complains, cries, coughs, or vomits during treatment, these behaviors are regarded as aspects of resistance and are believed to demonstrate the need to maintain or increase the discomfort until catharsis occurs (Reber, 1996).
Conceptual Problems for Empirical Validation
There are several barriers to empirical validation inherent in the theory and practice of AT. For example, the generally accepted requirement that older children as well as their parents provide informed consent before research is fundable or publishable conflicts with AT thinking.
Information and Consent
An important theme of AT is that children are helped to recognize adult authority and form an attachment when knowledge is withheld from them (Thomas, 2000). Children as young as preschool age are not told when they are to be separated from their parents and taken to a foster home. Second, children are said not to have a right to stay sick; the idea that a child has the right to refuse holding is considered as unwarranted as the idea that chemotherapy can be refused (Hage, 1997). Third, a child's refusal would be simply considered resistance and additional evidence of the need for holding therapy.
A second conceptual issue involves the usual assumption that data should be capable of communication to others and verifiable by confirmatory measurement. This is generally considered one of the foundations of scientific method, but it is rejected by AT writers who assert that the only valid measure of the symptoms of an attachment disorder is the mother's report. The signature of these disorders is said to be the child's ability to conceal emotional problems from the most experienced observers, but to behave with vicious hostility when alone with the mother (Randolph, 2000; it is considered rare for the child's anger to be directed toward the father, although the possibility of cruelty to animals or younger children is stressed). In the absence of a concealed recording system or hidden observer, which never seem to be used in AT practice, there is no public verifiability of the mother's report.
This problem with verifiability has become exacerbated over the years. For example, one symptom of attachment disorders was initially defined as "failure to make eye contact"; now it is trouble "making eye contact when adults want him/her to" (Randolph, 2000). Presumably, only the adults themselves can know whether they wanted the child to make eye contact at a given moment.
A third conceptual issue, related to the second, involves the need for a reliable and valid outcome measure to be used in evaluation of the treatment (Chambless & Hollon, 1998). If the primary outcome measure is the mother's unsupported report, independent testing of its validity is difficult. It would be possible to establish predictive validity, especially in light of the fact that some easily measured behaviors such as serial killing are predicted for untreated AD children, but this has not been done. The establishment of concurrent validity is complicated by the claim that the disorder is unrecognized by most therapists. As we will see in the next section, AT writers have attempted to establish the reliability of a questionnaire measure of the mother's report, but this does not solve the problem of an independent measure against which to gauge validity. Serious limitations for the establishment of the questionnaire's reliability and validity thus emerge from the test's reliance on the mothers' judgments.
Research on AT
Most of the research discussed here was carried out by the staff of the Attachment Center at Evergreen (ACE) in Colorado. This organization is also the source of the major book discussing AT (Levy, 2000). Connell Watkins was at one time affiliated with ACE, but the organization's Web site was quick to repudiate her techniques after Candace Newmaker's death.
As commercial ventures, AT clinics can benefit from the use of a checklist that demonstrates to parents their children's need for treatment. AT writers began with such a checklist and have attempted to develop a questionnaire that could be described as reliable and valid.
Efforts toward questionnaire development began with an Attachment Disorder Symptom Checklist, which is still posted on some AT Web sites (for example, www.attachment-ga.com/html/ADSX2.html). This checklist shows a remarkable overlap with similar checklists presented in the past as indicators of sexual abuse (Dawes, 1994; Underwager & Wakefield, 1990). A peculiarity of the checklist is its inclusion of statements about the parent's feelings toward the child as well as statements about the child's behavior. For example, parental feelings are evaluated through responses to such statements as "Parent feels used" and "is wary of the child's motives if affection is expressed," and "Parents feel more angry and frustrated with this child than with other children." The child's behavior is referred to in such statements as "Child has a grandiose sense of self-importance" and "Child 'forgets' parental instructions or directives."
The 30-item Randolph Attachment Disorder Questionnaire (RADQ) (Randolph, 2000) was developed from the Attachment Disorder Symptom Checklist. The RADQ is presented not as an assessment of RAD but rather of attachment disorder (AD), a diagnosis not "yet" in the DSM. This posited condition involves both RAD and either Conduct Disorder or many symptoms of Oppositional Defiant Disorder.
The RADQ manual emphasizes that the RADQ score alone should not be used to make the AD diagnosis. The test is always to be completed by the adult female who knows the child best and has been living continuously with the child, but she should be carefully guided by an AD expert to provide accurate answers (Randolph, 2000).
The RADQ is at the 5th-grade reading level. Only one form appears to exist, although repeated measures are used. The mother ranks her responses to statements from 1 (rarely) to 5 (usually), and is instructed that these ranks are to be linked to specific frequencies with which the behaviors occur. All items are set up with 5 on the reader's left and 1 on the right. Positive statements are never converted to negative forms as a check for an acquiescence or a counteracquiescence response set. Some statements are cast in unusually dramatic or emotional ways: "My child has a tremendous need to have control . . ."; "My child acts amazingly innocent . . ." (emphasis in original).
Randolph (2000) reported data based on the completion of the RADQ by 350 parents. She noted high reliability, reporting correlation coefficients of over .80 for test-retest stability and internal consistency (as measured by odd-even correlations, which are rarely used today by psychometricians).
As noted earlier, validity is inherently the more serious problem with the RADQ. Randolph claimed content validity because the RADQ items were based on the Attachment Disorder Symptom Checklist mentioned earlier. She also presented significant correlations with two of six selected subscales from the Personality Inventory for Children (Lachar, 1979) and two of eight subscales of the Child Behavior Checklist (Achenbach, 1991). There was a significant correlation with 1 of 12 subscales of the Millon Adolescent Personality Inventory (Millon, 1982). Randolph also noted the importance of Rorschach information for increased validity in the assessment of certain types of children (but see Wood & Lilienfeld, 1999, for a critique). There was no discussion of any direct measure of behavior independent of the mother's report, although such items as stealing, fire setting, cruelty, and frequent injuries should be amenable to independent corroboration.
Randolph (2000) reported that the RADQ clearly distinguished, with nonoverlapping distributions (N = 186), between children who have an AD diagnosis and groups who (a) had been maltreated, (b) had disruptive behavior disorders, and (c) were classed with other disorders, primarily anxiety and depression. There were large differences among the groups in living situations (birth home, foster home, adoptive home, group home) and therefore presumably differences in the types of persons responding to the RADQ. The groups also differed, sometimes dramatically, in the children's ethnicity and gender.
Randolph also presented a division of the RADQ into subscales that she considered associated with four posited subtypes of AD. These four subtypes were given names associated with categories of toddler attachment behavior seen in the Strange Situation (Main & Solomon, 1990), but Randolph specifically noted that she did not mean to imply any connection between the two; she had, she said, been unable to think of any other names (Randolph, 2000, p. 52). Randolph's attempts to present the subscale scores as linked to subtypes, based on measures from 160 children, were weakened by the fact that she herself apparently assigned the children to subtype groups and did not seek an independent evaluation.
Randolph's extensive work on the RADQ suggests that this test should be the instrument of choice in the evaluation of AT outcomes, but, as we will soon see, this has not been the case.
Empirical Evaluation of AT
In written material and in testimony at the Watkins-Ponder trial, AT practitioners have repeatedly asserted that AT is not only effective, but should be considered "best practice" for a disorder that other treatments do not ameliorate. This claim would be difficult to substantiate using the APA Task Force criteria for empirically supported treatments, because of the absence of a clearly valid outcome measure. In addition, however, there are no reported studies of the effect of AT on children using random assignment to groups, viz., the Class I evidence (in terms of the evidence-based approach [Patrick, Mozzoni, & Patrick, 2000]) that allows best practice to be determined. The absence of randomized trials is not surprising, because AT appears to be performed in poorly organized, commercially oriented clinic arrangements, rather than in the institutional settings in which randomized trials are commonly planned and overseen.
These methodological limitations notwithstanding, we should examine the less stringent approaches that some practitioners have adopted for the evaluation of AT. Preliminary steps in research can be valuable guides to the development of randomized, controlled studies. To think in terms of evidence-based care, such work may not tell us about best practice, but it can sometimes provide treatment guidelines or options (Patrick, Mozzoni, & Patrick, 2000).
In the following section, I review what appears to be all of the available evidence regarding the efficacy of AT. Two dissertations mentioned by AT writers do not appear in Dissertation Abstracts International and were not located. (One dissertation whose title refers to AT is, in fact, on a different topic. ) The material discussed here amounts to a dissertation, a journal article based on this dissertation, a Web site article apparently taken from an organization's newsletter, and other materials from the ACE Web site.
Considering the entirely clinical emphasis of AT practitioners, we might expect some carefully described case histories or similar clinical reports. The practice of videotaping therapy sessions, shown at the Watkins-Ponder trial, lends itself to detailed descriptions. However, although many brief anecdotes are presented, there seem to be no complete case reports. A detailed case report in one book on AT (Cline,1992) is in fact the work of Erickson (1962), whose thinking was not generally based on AT principles.
Three quasi-experimental studies have investigated the efficacy of AT. The first of these, the only one not associated with ACE, is a simple study whose interpretation does not outstrip the data. Lester (1997) examined 12 families whose adopted children received AT. The children, whose age range was broad, experienced different levels of treatment, many with 3-hour sessions daily for weeks. The parents responded to the Devereux Scale of Mental Disorders, a rating scale for which evidence of reliability has been published, as well as another scale in the process of development, on four occasions (before the child's initial assessment, at the time of the initial assessment, after the assessment but before therapy began, and at least 4 weeks after therapy began). Average scores were presented, but there were no statistical analyses. Lester reported that all scores improved over time, but that the greatest improvement occurred before therapy had begun. She noted that the parents might simply have felt better after talking to a sympathetic person. Like other AT writers, Lester apparently did not consider regression to the mean as a possible source of improvement, although it is likely that the parents sought treatment when the children's symptoms were at their worst and that some degree of spontaneous remission was likely.
A more elaborate quasi-experimental study, presented on the ACE Web site, was apparently carried out by Elizabeth Randolph and other ACE staff members (Randolph, 1997a). The Web site report lacks a number of details that would normally be found in published material, such as measures of variability to accompany means or complete analysis of variance tables. This study examined Child Behavior Checklist data from 25 children, 7 to 12 years old, on three occasions. Parents or foster parents completed the checklist (a) before treatment began, (b) following a 2-week intensive treatment period and between 3 and 6 months of long-term treatment in a therapeutic foster home, and (c) following an additional 6 months. No untreated comparison group was included to establish the effects of factors such as maturation that could cause change over such a time period (AT practitioners have asserted that the disorder they treat is so intractable that it does not change over time without treatment, so they would probably argue that maturation was a negligible factor in improvement; Randolph, 1997b).
The Web site report noted mean scores on the 8 subscales of the checklist, but no measures of variability. The statistical analysis was described as a simple analysis of variance (ANOVA), although presumably a repeated measures ANOVA would have been appropriate. One of the 8 subscales showed no change, and 5 showed significant changes over the 12-month period.
A dissertation by Myeroff (1997) at the Union Institute of Ohio, which was subsequently published (Myeroff, Mertlich, & Gross, 1999), was the third quasi-experiment that attempted to evaluate AT. This study included an untreated comparison group. Myeroff and her colleagues collected parents' responses to the Child Behavior Checklist for 23 families who contacted ACE to seek AT for their adopted children. The treatment group was composed of 12 children who were brought for treatment, and the untreated comparison group was composed of 11 children whose parents made contact but were unable to bring them to ACE.
Both groups of parents completed reports on two subscales of the checklist, after their initial contact and again after a 4-week interval. For the treatment group, a 2-week "intensive" occurred midway between the two reports. Myeroff et al. (1999) reported significant differences between the two groups, with the treatment group showing significant improvement on both the aggression and delinquency subscales.
According to Myeroff et al. (1999), the failure of the untreated children to attend the clinic was not due to the condition of either parent or child. Family income, gender, race, and pre-adoption placement did not differ significantly across the two groups. Whether these groups were also matched statistically on initial checklist scores was not mentioned. Myeroff et al. noted that differences in finances might have been responsible for the failure to attend. Such differences could also influence the developmental outcome for an adopted child. Similarly, marital disagreements, number of siblings, physical or mental problems of family members, educational needs of siblings, and job situations of parents could all affect both the development of adopted children and the decision of parents to bring a child to the clinic. Myeroff et al.'s claim that the groups were appropriately matched is thus unwarranted, and the differences reported between the groups may well be due to the numerous factors that determined the families' self-selection. This conclusion is in agreement with the briefer critique offered by Wilson (2001) on some of the weaknesses of AT research.
Many of the statistical analyses performed by AT researchers have been problematic. The substitution of a simple ANOVA for the appropriate repeated measures ANOVA in Randolph's (1997a) work increased the possibility of a Type I error (Ferguson, 1959). In Randolph's (2000) reports on the RADQ, she noted what appear to be large standard deviations in comparison with reported ranges, but did not comment on the normality of the distributions.
Myeroff et al.'s (1999) published work reported significant t comparisons for both pretreatment and posttreatment aggression scores and pre and post delinquency scores for the treatment group, but nonsignificant ts on both measures for the nontreatment group. She also compared pre and post difference scores for the treatment and nontreatment groups on the aggression and the delinquency subscales, with significant ts in both cases. The latter two comparisons are the appropriate ones to conduct rather than the first four, and the use of six rather than two t calculations increases the chances of a Type I error.
Claims for AT's efficacy have not been supported by the sparse research evidence presented by AT advocates. Because this intervention is not only without validation but has been associated with injury and death, insurance companies and state agencies appear to have used poor judgment in supporting it.
AT remains an important topic of study for professional psychology in spite of the absence of empirical support, or perhaps even because of it. AT and similar therapies offer a window into the folk beliefs that cause people to reject conventional psychotherapy in favor of unvalidated mental health interventions. Practitioners of AT may share, or at least speak to, some assumptions about human nature that are foreign to many professional psychologists but that are embedded in American popular culture. An example would be the belief that personality transformations can be produced by such ritual acts as baptism or exorcism. Such a sacramental view of personality change is greatly at odds with cognitive and behavioral approaches to therapy.
Examination of AT theory and research also reveals a tendency toward certain basic cognitive errors that are characteristic of adolescents (Demetriou, Efklides, Papadaki, Papantoniou, & Economou, 1993). For example, our earlier discussion of AT research showed researchers' difficulty in isolating variables. The theory and practice of AT involve the dependent variable error, a tendency to assume that if manipulation of the independent variable causes changes in the dependent variable, the opposite should also be true. For instance, AT theory assumes that lack of eye contact (the dependent variable) is caused by failure of attachment (the independent variable); treatment involves the forcing of eye contact, which is expected to correct attachment. This immature thinking on the part of the AT practitioner may seem comfortable to parents, whereas the reasoning inherent in conventional psychotherapy may be resisted because of its unfamiliar sophistication. Many adults continue to make this type of error, broadly known in logic as "affirming the consequent," and may be reassured by the familiar thought pattern as they try to cope with the diagnosis given to their child and the concern that they may be blamed.
A final point about the importance of attending to AT is the possible connection of this treatment to other dangerous forms of "wild" therapy (see Singer & Lalich, 1996). For example, some deaths of adolescents in wilderness camps may be attributable to a belief system shared with AT. The American director of a camp in Samoa in which severe physical abuse and fraud have been alleged has been quoted as saying, "Kids come in with all sorts of little ways to manipulate, with a lot of anger. We physically stress them out and that breaks down the facades to get to their heart"(Janofsky, 2001, p. A14); this man was acquitted of charges related to the death of a girl in another camp some years ago. An assumption shared by a number of these "crazy therapies" (Singer & Lalich, 1996) involves recapitulation, the notion that it is possible to rework a developmental sequence by mimicking at a later time the factors that the practitioners believe cause normal developmental change in early life.
It would seem desirable for legislation to regulate treatments that are unvalidated and potentially harmful, but a legislative approach brings up First Amendment issues and may be resisted by professional groups wishing to retain the privilege of self-policing. Legislation can also drive undesirable treatments underground and decrease the extent to which they can be controlled. More effective approaches may involve public education. As a first effort in this direction, I suggest withdrawal of the continuing education units currently given by some institutions for AT workshops and seminars.
- Williams, G. G. (1990). Attachment therapy as a method for altering significantly family cohesion and adaptability. Dissertation Abstracts International, 51 (5-A), 1795. (UMI NO. 9026811)
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist and 1991 profile. Burlington: University of Vermont Press.
Bowlby, J. (1982). Attachment. New York: Basic.
Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66, 7-18.
Cline, F. W. (1992). Hope for high risk and rage filled children. Evergreen, CO: EC Publications.
Crowder, C. (2000, July 29). Prosecutors add charges for rebirthing therapist. Rocky Mountain News. Retrieved July 29, 2000, from http://insidedenver.com/news/0729char0.shtml.
Dawes, R. M. (1994) House of cards: Psychology and psychotherapy built upon myth. New York: Free Press.
Delaney, R. J., & Kunstal, F. R. (1993). Troubled transplants. Fort Collins, CO: Horsetooth Press.
Demetriou, A., Efklides, A., Papadaki, M., Papantoniou, G., & Economou, A. (1993). Structure and development of causal-experimental though from early adolescence to youth. Developmental Psychology, 29, 480-497.
Emerson, W. R. (1996). Points of view: The vulnerable prenate. Pre- and Perinatal Psychology Journal, 10(3), 125-142.
Erickson, M. H. (1962). The identification of a secure reality. Family Process, 1(2), 294-303.
Ferguson, G. A. (1959). Statistical analysis in psychology and education. New York: McGraw-Hill.
Hage, D. (1997). Holding therapy: Harmful? . . . or rather beneficial. Roots and Wings Adoption Magazine, 9(1), 46-49. Retrieved June 15, 2000, from http://debrahage.com/pwp.
Hanson, R. F., & Spratt, E. G. (2000). Reactive Attachment Disorder: What we know about the disorder and implications for treatment. Child Maltreatment, 5(2), 137-151.
James, B. (1994). Handbook for treatment of attachment-trauma problems in children. New York: Lexington.
Janofsky, M. (2001, August 9). Boot camps proponent becomes focus of critics. New York Times, A14.
Lachar, D. (1979). Manual for the Personality Inventory for Children. Minneapolis, MN: National Computer Systems.
Lester, V. S. (1997). Behavior change as reported by caregivers of children receiving holding therapy. Retrieved August 4, 2000 from the Association for Treeatment and Training in the Attachment of Children Web site: http:// www.attach.org/lester.htm.
Levy, T. M., & Orlans, M. (2000). Attachment disorder and the adoptive family. In T. M. Levy (Ed.), Handbook of attachment interventions (pp. 244-260). San Diego, CA: Academic.
Lieberman, A., & Zeanah, C. H. (1999). Contributions of attachment theory to infant-parent psychotherapy and other interventions with infants and young children. In J. Cassidy & P. Shaver (Eds.), Handbook of attachment (pp. 555-574). New York: Guilford.
Lowe, P. (2001, June 19). Therapists get 16 years. Rocky Mountain News. Retrieved June 19, 2001, from http://rockymountainnews.com/drmn/0,1299,DRMN_15_6750 52,00.html.
Main, M. B., & Solomon, J. A. (1990). Procedures for identifying infants as disorganized/disoriented during the Ainsworth Strange Situation. In M. Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment in the preschool years (pp.121-160). Chicago: University of Chicago Press.
Mercer, J. (2001). Attachment therapy using deliberate restraint: An object lesson in the identification of unvalidated treatments. Journal of Child and Adolescent Psychiatric Nursing, 14(3), 105-114.
Millon, T. (1982). Millon Adolescent Personality Inventory manual. Minneapolis, MN: National Computer Systems. Myeroff, R. L. (1997). Comparative effectiveness of attachment therapy with the special needs adoptive population. Dissertation Abstracts International, 58(6-B), 3323. (UMI No. AAM9736716)
Myeroff, R. L., Mertlich, G., & Gross, G. (1999). Comparative effectiveness of holding therapy with aggressive children. Child Psychiatry and Human Development, 29(4), 303-313.
New Hampshire Executive Council Minutes. (1999). Item #85A. Retrieved August 19, 2000, from the New Hampshire State Government Web site: http://www.state.nh.us/ council/min/092899.
Patrick, P. D., Mozzoni, M., & Patrick, S. T. (2000). Evidence-based care and the single-subject design. Infants and Young Children, 13(1), 60-73.
Randolph, E. (1997a). Attachment therapy does work! Retrieved January 20, 2001, from the Attachment Center at Evergreen Web site: http://www.attachmentcenter.org/ articles/article015.htm.
Randolph, E. (1997b). Is attachment therapy necessary? Connections, August, 45. Retrieved January 20, 2001 from the Attachment Center at Evergreen Web site: http:// attachment center.org/articles/article029.htm.
Randolph, E. (2000). Manual for the Randolph Attachment Disorder Questionnaire. Evergreen, CO: Attachment Center Press.
Reber, K. (1996). Children at risk for reactive attachment disorder: Assessment, diagnosis, and treatment. Progress: Family Systems Research and Therapy, 5, 83-98.
Reich, W. (1945). Character analysis. Rangely, ME: Orgone Institute Press.
Schiff, J. L. (1970). All my children. New York: M. Evans.
Singer, M. T., & Lalich, J. A. (1996). Crazy therapies. San Francisco: Jossey-Bass.
Tavris, C. (1989). Anger: The misunderstood emotion. New York: Simon & Schuster.
Thomas, N. L. (2000). Parenting children with attachment disorders. In T. Levy (Ed.), Handbook of attachment interventions (pp. 67-111). San Diego, CA: Academic.
Underwager, R. C., & Wakefield, H. (1990). The real world of child interrogations. Springfield, IL: C. C. Thomas.
Welch, M. G. (1989). Holding time. New York: Fireside.
Wilson, S. L. (2001). Attachment disorders: Review and current status. Journal of Psychology, 135(1), 37-51.
Wood, J. M., & Lilienfeld, S. O. (1999). The Rorschach Inkblot Test: A case of overstatement? Assessment, 6, 341-349.
Zaslow, R. W. (1966). Reversals in children as a result of midline body orientation. Journal of Educational Psychology, 57(3), 133-139.
Zaslow, R. W., & Menta, M. (1975). The psychology of the Z-process: Attachment and activity. San Jose, CA: San Jose State University Press.