Enhancing attachment security in the infants of women in a jail-diversion program

Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/44804553

Enhancingattachmentsecurityintheinfantsofwomeninajail-diversionprogram

ArticleinAttachment&HumanDevelopment·July2010

DOI:10.1080/14616730903416955·Source:PubMed

CITATIONS

39

READS

306

9authors,including:

BrandiStupica

AlmaCollege

8PUBLICATIONS166CITATIONS

SEEPROFILE

KentHoffman

GonzagaUniversity

6PUBLICATIONS585CITATIONS

SEEPROFILE

AllcontentfollowingthispagewasuploadedbyBertPowellon25June2015.

Theuserhasrequestedenhancementofthedownloadedfile.

Enhancing attachment security in the infants of women in a jail-diversion

program

Jude Cassidya*, Yair Zivb, Brandi Stupicaa, Laura J. Shermana, Heidi Butlera,Andrea Karfginc, Glen Cooperd, Kent T. Hoffmand and Bert Powelld

aUniversity of Maryland; bWestat; cTamar’s Children, Inc.; dMarycliff Institute

(Received 8 August 2008; final version received 30 July 2009)

Pregnant female offenders face multiple adversities that make successful parentingdifficult. As a result, their children are at risk of developing insecure attachmentand attachment disorganization, both of which are associated with an increasedlikelihood of poor developmental outcomes. We evaluated the outcomes ofparticipants in Tamar’s Children, a 15-month jail-diversion intervention forpregnant, nonviolent offenders with a history of substance abuse. All womenreceived extensive wrap-around social services as well as the Circle of SecurityPerinatal Protocol (Cooper, Hoffman, & Powell, 2003). We present data on 20women and their infants who completed the full dosage of treatment (aresidential-living phase from pregnancy until infant age six months andcommunity-living phase until 12 months). Results indicated that (1) programinfants had rates of attachment security and attachment disorganizationcomparable to rates typically found in low-risk samples (and more favorablethan those typically found in high-risk samples); (2) program mothers had levelsof maternal sensitivity comparable to mothers in an existing communitycomparison group; and (3) improvement over time emerged for maternaldepressive symptomatology, but not other aspects of maternal functioning.Given the lack of a randomized control group, results are discussed in terms ofthe exploratory, program-development nature of the study.

Keywords: attachment; incarcerated parents; intervention; maternal sensitivity;female offenders

Introduction

Since the start of the federal War on Drugs in 1986, women offenders have becomethe fastest growing population in the criminal justice system (US Department ofJustice, 1998; for a discussion of the effects of the federal War on Drugs on women,see Wiewel & Mosley, 2006). An emergent body of research has revealed thatincarcerated women are likely to be single mothers with a history of substance abuseand victimization who are poor, uneducated, and traumatized (Myers, Smarsh,Amlund-Hagan, & Kennon, 1999). An estimated 8–10% of incarcerated women arepregnant when they enter prison (Bloom, 1995). Despite increased interest in therapidly growing female prison population, there is a dearth of research on women

*Corresponding author. Email: [email protected]

Attachment & Human Development

Vol. 12, No. 4, July 2010, 333–353

ISSN 1461-6734 print/ISSN 1469-2988 online

� 2010 Taylor & Francis

DOI: 10.1080/14616730903416955

http://www.informaworld.com

who give birth while serving their sentences and virtually no research on thedevelopmental outcomes of the infants born to these women (for an exception, seeByrne, Goshin, & Joestl, 2010).

Mirroring the reports of the psychosocial risk factors characterizing incarceratedwomen, the literature on pregnant inmates reveals that they face multiple adversitiesthat render successful parenting difficult. Among these difficulties are poverty andlow educational attainment. For example, Shelton and Gill (1989) reported that theaverage annual income for pregnant inmates in their sample fell well below thepoverty line for a single person. This level of impoverishment is not surprising giventhat most incarcerated pregnant women have not graduated from high school(Barkauskas, Low, & Pimlott, 2002; Fogel & Belyea, 2001; Hairston, Bates, &Lawrence-Wills, 2003; Kubiak, Young, Siefert, & Stewart, 2004; Schroeder & Bell,2005). In addition, the overwhelming majority must also cope with single parenting:Several studies have revealed that at least 80% of pregnant inmates are single,separated, or divorced (Barkauskas et al., 2002; Fogel & Belyea, 2001; Hairstonet al., 2003; Hutchinson, Moore, Propper, & Mariaskin, 2008; Williams & Schulte-Day, 2006).

Incarcerated pregnant women are also at risk because of high rates of substanceabuse; approximately 65 to 94% of pregnant inmates report histories of substanceabuse (Cordero, Hines, Shibley, & Landon, 1992; Fogel & Belyea, 2001; Hairstonet al., 2003; Kubiak et al., 2004; Williams & Schulte-Day, 2006). In addition tohaving drug problems, the majority of pregnant inmates report depressivesymptomatology at levels indicative of clinical depression (Fogel & Belyea, 2001;Hutchinson et al., 2008; Kubiak et al., 2004). The psychiatric problems faced bythese women may be magnified by their histories of victimization. Fogel and Belyea(2001) found that 60% of the pregnant inmates they interviewed experienced familyviolence during childhood, with some women reporting having been beaten (21%) orhaving a knife or gun used on them (6%). Additionally, 24% reported experiencingsexual abuse before adulthood. Similarly, Sable, Fieberg, Martin, and Kupper (1999)reported that incarcerated pregnant women, compared to non-incarcerated pregnantwomen from similar demographic backgrounds, were more than three times as likelyto have experienced a combination of both physical and sexual violence.Furthermore, most pregnant incarcerated women recalled their primary caregiverduring their own childhood as having been cold, rejecting, intrusive, and over-controlling (Hutchinson et al., 2008).

The multiple psychosocial problems presented by pregnant inmates have beenwell-established as risk factors for poor parenting. For instance, Belsky, Bell,Bradley, Stallard, and Stewart-Brown (2007) found that lower income, lower levelsof maternal education, and more time spent as a single parent, were associated withless maternal warmth and positive control and with greater maternal negativity.Substance abuse and psychopathology have also been linked to poor parenting (for areview see Hans, Bernstein, & Henson, 1999). In particular, Hans et al. found thatopioid-dependent mothers were less responsive and more negative during interac-tions with their infants than non-drug-dependent mothers. Maternal psychopathol-ogy was also related to less responsive and more negative interactions with children,even after controlling for substance abuse. In addition, childhood experiences ofabusive and harsh parenting also put incarcerated women at risk of using abusiveand harsh parenting techniques with their own children (Hans et al., 1999; seePutallaz, Costanzo, Grimes, & Sherman, 1998, and van IJzendoorn, 1992, for

334 J. Cassidy et al.

reviews of the intergenerational transmission of parenting). The factors that placeincarcerated women at risk for insensitive parenting and maltreatment are troubling,given the large body of research and meta-analytic evidence indicating thatinsensitive parenting and maltreatment place children at risk for developing insecureattachments and attachment disorganization (Cyr, Euser, Bakermans-Kranenburg,& van IJzendoorn, in press; De Wolff & van IJzendoorn, 1997; van IJzendoorn,Schuengel, & Bakermans-Kranenburg, 1999; for reviews, see Cassidy & Shaver,2008). Consistent with the high-risk status of children of incarcerated parents, onestudy found that 63% of these children had insecure attachment representations oftheir mothers (Poehlmann, 2005). These findings establish a direct link betweenmaternal incarceration and increased risk of attachment insecurity.

Most concerning, however, is the constellation of poor developmental outcomesassociated with attachment insecurity and disorganization. Insecure attachment ininfancy has been linked to more negative affect, anger, and aggression in preschool(Sroufe, 1983), problems in showing appropriate empathy toward classmates(Kestenbaum, Farber, & Sroufe, 1989), and less successful and satisfying peerrelations and friendships (Elicker, Englund, & Sroufe, 1992). In addition, children’sinsecure attachment-related experiences are also a risk factor for various forms ofpsychopathology during childhood and adulthood (for reviews, see DeKlyen &Greenberg, 2008, and Dozier, Stovall-McClough, & Albus, 2008). Faring the worstin terms of psychosocial outcomes, however, are infants classified as insecure-disorganized. Two studies in particular have shown that disorganized infants areleast able to cope with the stress of separating and reuniting with their mothers, asevidenced by elevated cortisol levels following the Strange Situation procedure(Hertsgaard, Gunnar, Erikson, & Nachmias, 1995; Spangler & Grossman, 1993).Furthermore, several studies have confirmed that disorganization in infancy predictslater aggressive and externalizing problem behavior (Carlson, 1998; Goldberg, 1997;Lyons-Ruth, Easterbrooks, & Cibelli, 1997; Radke-Yarrow, McCann, DeMulder, &Belmont, 1995; Shaw, Owens, Vondra, & Keenan, 1996; for a review, see Lyons-Ruth & Jacobvitz, 2008).

Given the multitude of risks and the potential for poor outcomes that pregnantinmates and their infants face, and given that many of these women view theirpregnancy as an opportunity for a ‘‘fresh start’’ (Hutchinson et al., 2008), preventionand intervention programs with pregnant inmates are clearly needed. Moreover,despite the need for intervention, fewer than a dozen states currently have prisonnursery programs (i.e., a program that permits infants to remain with their mothersin the prison; US Department of Justice, 2002; see Carlson, 2009, for a review ofcurrent prison nursery programs). In addition, non-prison residential treatmentprograms for mothers involved in the justice system and their infants tend to emergeand end quickly, depending on the political climate and available resources (MaryByrne, personal communication, July 10, 2008). Although most prison nurseriesoffer some form of parenting education, information about the content, design, andeffectiveness of these parenting classes is largely unpublished (see Bruns, 2006, andByrne et al., 2010; see also Baradon, Fonagy, Bland, Lenard, & Sleed, 2008, for adescription of a prison nursery program in the United Kingdom).

This lack of publicly available information also extends to non-prison residentialjail-diversion programs for pregnant women. Of the four articles published aboutsuch jail-diversion programs, two contained formal evaluations (Barkauskas et al.,2002; Kubiak et al., 2004), but the programs did not provide mothers with parenting

Attachment & Human Development 335

education, nor did they assess child development outcomes. Although Hairstonet al. (2003) and Wiewel and Mosley (2006) provided parenting education as partof their jail-diversion programs, their evaluations were limited to individualinterviews with mothers and staff (Hairston et al., 2003) plus a case study of oneparticipating dyad (Wiewel & Mosley, 2006). It is clear that more formalassessments of residential treatment programs (both within and outside of prisonsettings) are needed, especially of programs focused on parenting behavior andchild outcomes.

The present study: enhancing secure attachment in the Tamar’s Children program

Tamar’s Children was a preventive intervention designed to address the multipleneeds of pregnant, substance-abusing women convicted of non-violent offenses andtheir infants through a jail-diversion program offering an integrated andcomprehensive network of services. Because all eligible women were invited toparticipate and were offered the same services, the study design did not allow for arandomized control group drawn from the initial pool of offenders. The firstcomponent of this intervention program addressed the medical and psychiatric needsof the mothers and the well-being of their infants by creating an integrated networkof prenatal and medical care; substance abuse, mental health, and trauma treatment;individual and group psychotherapy; educational enhancement (e.g., GED classes);work skills training; housing assistance; and advocacy. The second component, theCircle of Security Perinatal Protocol (COS-PP; Cooper et al., 2003), providedparenting education designed to promote maternal sensitivity and enhance secureinfant attachment.

The COS-PP derives from the Circle of Security Protocol (COS; Cooper,Hoffman, Powell, & Marvin, 2005; Hoffman, Marvin, Cooper, & Powell, 2006;Marvin, Cooper, Hoffman, & Powell, 2002), originally designed as a 20-weekprotocol for use with groups of parents of preschool children. The COS is designedto improve parental caregiving capacities and enhance secure infant attachment bydeveloping and fostering: (1) the parent’s understanding of her infant’s emotionalcues and relationship needs; (2) the parent’s observational and inferential skills; (3)the parent’s reflective functioning about the reciprocal effects of her and her infant’sbehaviors on each other’s cognitions, emotions, and behaviors; (4) the parent’scapacity for emotion regulation; and (5) the parent’s appropriate responsiveness toher infant’s signals relevant to her infant’s use of her as a secure base for exploringand a safe haven for comfort and safety.

Evidence for the efficacy of the COS intervention with at-risk populations hascome from a study of 65 toddlers and pre-school-aged children assessed forattachment security both before and immediately after the 20-week COSintervention (Hoffman et al., 2006). Participants were recruited from Head Startand Early Head Start programs; all had low incomes, and most parents and some ofthe children had experienced maltreatment or trauma. Prior to the COS intervention,only 20% of the children were classified as secure; following the COS intervention, asignificant shift occurred and 54% were classified as secure. A significant reductionin rates of disorganized attachment (from 60% to 25%) also was found.

The Tamar’s Children program had many goals; in this report we examine three:(1) enhancing secure infant attachment and reducing the risk of disorganized infantattachment, (2) fostering sensitive maternal behavior, and (3) improving maternal

336 J. Cassidy et al.

psychosocial functioning. Because there was no control group, we comparedoutcome data from the present study to existing norms for low- and high-riskpopulations, to data from an existing community sample, and to the participants’pre-intervention data collected at enrollment. First, in order to examine whether theTamar’s Children program fostered secure infant attachment and reduced the risk ofdisorganized infant attachment, we compared the rates of security and disorganiza-tion obtained for the present sample with rates established in meta-analytic reportsas typical of high- and low-risk samples (Cyr et al., in press; van IJzendoorn et al.,1999). Specifically, we compared the attachment distribution obtained in the presentstudy to those established by van IJzendoorn et al. (1999) for low-SES participants(48% secure, 25% disorganized), infants with depressed parents (41% secure, 21%disorganized), and infants with substance abusing mothers (26% secure, 43%disorganized); we compared the attachment distribution obtained in the presentstudy to that established by Cyr et al. (in press) for infants with maltreating parents(5% secure, 67% disorganized). In addition, we compared the rates of security anddisorganization obtained in the present study to those typical in low-risk samples(i.e., middle-class, non-clinical, North American samples; 62% secure, 15%disorganized; van IJzendoorn et al., 1999). If the Tamar’s Children program waseffective at fostering secure infant attachment and reducing rates of disorganizedattachment, rates of security should be significantly greater than those for high-risksamples, rates of disorganization should be significantly lower than those typical ofhigh-risk samples, and rates of security and disorganization should be comparable tothose typical of a low-risk sample.

Second, we evaluated maternal behavior at program completion bycomparing maternal sensitivity to that obtained from an existing community-based sample. If the Tamar’s Children program was effective at enhancingmaternal sensitivity, we expected ratings of maternal sensitivity not to differ fromthose obtained from the community-based sample (Cassidy, Woodhouse, Sher-man, Stupica, & Lejuez, 2010). Third, we evaluated the extent to which maternalfunctioning improved over time by comparing maternal self-reported functioningobtained at enrollment during pregnancy (pre-intervention) to that obtained atinfant age 12 months (post-intervention). If the Tamar’s Children program waseffective at improving maternal psychosocial functioning, there should besignificant improvement from the pre-intervention assessment to the post-intervention assessment.

Method

Sample selection and participants

To be eligible for the Tamar’s Children program, pregnant women who lived in theBaltimore, MD, metropolitan region had to meet the following inclusion criteria:term of mandatory supervision of greater than three years; history of substanceabuse; no current charge (or conviction within five years) for a violent crime; and nodiagnosis of psychosis. Once eligibility for the Tamar’s Children program wasdetermined, women were provided a description of the program and given theopportunity to participate. Fifty-four pregnant women who met inclusion criteriaentered the program. According to national statistics on mothers in prison asreported by the US Department of Justice (2008), the present sample wascomparable to state and federal prison samples in terms of race, age, education,

Attachment & Human Development 337

and history of abuse. Specifically, the racial/ethnic distribution in our sample was65% African American, 33% Caucasian, 2% Hispanic; median age was 32 years(M ¼ 31.59, SD ¼ 5.01); 45% of the participants completed high school or hadreceived a GED certification; 63% of women reported having been physically orsexually abused before the age of 18 years.

Because an administrative decision led the program to end earlier thananticipated, only 40 women had the opportunity to complete the entire 15-monthprogram. Of those 40 women, 22 (55%) women remained in the program andreceived the full dosage of treatment (from pregnancy until their infants were 12months old). Eighteen women dropped out before receiving the full treatment for thefollowing reasons: transferred to another program (4), drug relapse (4), lack ofcooperation (3), lack of interest (3), moved from the area or working (3), and infantillness/mortality (1). The 22 participants who completed the study did not differ fromthe 18 participants who dropped out on a variety of demographic variables: meanage, racial/ethnic minority status, percent abused before age 18 years, percentwitnessing violence before age 18 years, percent single, percent living with theinfant’s father, percent having other children, percent having previous involvementwith child protective services, or percent who completed high school or equivalency.Mean years of education, however, was significantly lower for study dropouts(M ¼ 10.28, SD ¼ 1.43) than completers (M ¼ 11.97, SD ¼ 1.9; t(29) ¼ 2.73,p 5 .05; see Table 1). In addition, those who completed the study and those whodropped out did not differ on several self-report measures of psychosocialfunctioning: attachment anxiety and avoidance, depression, dissociative experiences,self-esteem, social support, or experiences of parental rejection (see Table 2). Twoparticipants who completed treatment did not complete the laboratory assessment ofinfant attachment. Thus, the final sample size was 20 dyads.

Table 1. Maternal demographics as a function of study completion status.

Whole sample Completers Dropouts Fisher’sexact test

N % n % n % (p, two-tailed)

Age (530) 12 39 5 30 7 50 .29Minority racial/ethnic status 23 72 12 71 11 78 .69Completed high school 14 45 10 59 4 29 .15Single 14 47 10 62 4 29 .08Living w/infant’s father 4 18 3 27 1 9 .59Abused before age 18 20 63 12 63 8 62 .99Physically 14 42 7 37 7 50 .49Sexually 18 55 11 58 7 50 .73Physically & Sexually 11 34 6 32 5 39 .72

Witnessed violence before age 18 28 82 16 84 12 80 .99Have other children 26 90 15 94 11 85 .57

Previous protectiveservices involvement

8 38 4 33 4 44 .67

Note: Sample sizes vary due to missing data. Fisher’s exact tests examine differences between completersand dropouts. Percentages are calculated based on the n of that subsample (e.g., n/all completers). Withone exception, all racial/ethnic minority women are African American. Single women include those whowere never married, separated, divorced, widowed, or single/unspecified. In addition, completers did notdiffer from dropouts on mean age or mean number of other children.

338 J. Cassidy et al.

Study design

All participants were assigned to the same intervention program (i.e., there was norandomized control group). As such, all mothers had access to (1) individualizedsocial services (e.g., substance abuse treatment, mental health treatment, traumatreatment, individual and group psychotherapy, educational enhancement, GED/work skills training, housing assistance, and advocacy); and (2) the individualizedparenting intervention, the Circle of Security Perinatal Protocol (COS-PP; Cooperet al., 2003).

The Tamar’s Children program included two treatment phases: the residentialphase (when women lived in the treatment facility from enrollment during pregnancyto infant age six months with their infants), and the community-living phase (frominfant age six months to 12 months). Before entering the residential facility,participants gave written informed consent (which included a description of thefederal Certificate of Confidentiality obtained for the study) and they independentlycompleted pre-intervention self-report questionnaires in two sessions within twoweeks of program entry. The residential facility adopted a milieu model whereby allparticipants in the community (i.e., both staff and residents) contributed to dailytherapeutic interventions. All residential staff received extensive education aboutattachment theory and psychological trauma as these issues pertain to thepopulation with whom they worked. The residence was a locked, restricted-accessfacility (e.g., mothers were permitted to leave for special approved activities only,such as work or education). At infant age six months, mothers were released into thecommunity and continued to meet at the residential facility for services until theirinfants were 12 months old.

When infants were 12 months old, mothers came to a university laboratory wherethey participated with their infants in a 60-minute laboratory session that began withthe Strange Situation attachment assessment, followed by a 10-minute unstructured

Table 2. Maternal psychosocial functioning as a function of study completion status:comparison of completers vs. dropouts.

Measure

N ¼ 40Whole sample

n ¼ 22Completers

n ¼ 18Dropouts

t-valueM (SD) M (SD) M (SD)

ECRAvoidance 3.82 (1.09) 3.98 (.84) 3.63 (1.33) t(31) ¼ .92Anxiety 3.29 (1.33) 3.55 (1.15) 2.99 (1.49) t(31) ¼ 1.21

BDI - IA 10.91 (7.37) 12.99 (6.95) 8.41 (7.29) t(31) ¼ 1.85DES 21.62 (17.84) 20.56 (18.59) 22.87 (17.41) t(35) ¼7.39RSES 22.75 (7.29) 23.60 (7.04) 21.77 (7.74) t(26) ¼ .66SSQNo. supports 2.93 (1.64) 2.78 (1.47) 3.09 (1.85) t(34) ¼7.54Satisfaction 5.32 (1.19) 5.51 (.65) 5.11 (1.57) t(33) ¼ 1.00

PARQ 71.88 (29.55) 75.65 (30.13) 67.44 (29.13) t(35) ¼ .84

Note: No significant differences emerged from these tests. Sample sizes vary due to missing data.ECR ¼ Experiences in Close Relationships Scale; BDI - IA ¼ Beck Depression Inventory; DES ¼Dissociative Experiences Scale; RSES ¼ Rosenberg Self-esteem Scale; SSQ ¼ Social Support Ques-tionnaire; PARQ ¼ Parental Acceptance and Rejection Questionnaire: Warmth/Acceptance andHostility/Rejection subscales combined.

Attachment & Human Development 339

snack time and a subsequent 10-minute infant–mother free-play session. For free-play, mothers were provided age-appropriate toys and were told, ‘‘The last part oftoday’s visit is just you and your baby in the playroom for 10 minutes doingwhatever you’d like. All of the toys are still out for you.’’ Finally, mothers completedpost-intervention questionnaires while the infant was supervised by a familiar staffmember.

Circle of Security Perinatal Protocol (COS-PP; Cooper et al., 2003)

Mothers were assigned to therapeutic intervention groups (six mothers per group,informally called ‘‘circle groups’’) that focused on parenting. Once filled, the groupswere closed to new members and remained together for the duration of treatment.During their third trimester, mothers began to meet twice weekly in groups to receivethe COS-PP (Cooper et al., 2003). Treatment groups continued to meet twice weeklyfor 90-minute sessions after the infants were born until the infants’ were 12 monthsold. Each group was co-led by two therapists who remained with the group untilcompletion. All therapists were extensively trained by the protocol developers, whoalso conducted weekly supervision of the groups to ensure quality of theinterventions and fidelity to the intervention protocol. Two of the group leaderswere PhD-level clinicians, and the remaining interveners were Master’s-levelclinicians.

Content of the group sessions was based on the clinical perspective of the originalCircle of Security model (Cooper et al., 2005; Hoffman et al., 2006). According tothis model: (1) secure child–parent attachment is essential for healthy development;(2) lasting security results from a parent’s developing specific relationship capacitiesrather than learning techniques to manage her child’s behaviors; (3) a parent’srelationship capacities are best enhanced if she is operating within a secure-baserelationship with the therapist; and (4) treatment is most effective when focused onthe specific strengths and struggles of the particular parent–child dyad. Thus, groupsessions utilized video clips of mother–infant interactions to facilitate discussion ofcomplex ideas related to attachment theory in user-friendly terms in a safe, groupsetting.

While pregnant, mothers were introduced to the basics of mother–infantinteraction from the perspective of attachment theory by first watching anddiscussing 72 stock footage clips of mothers interacting with their babies. Thesestock footage clips, which included examples of secure, insecure, and disorganizedattachment, sensitive and insensitive parenting, as well as ‘‘before and after’’mother–infant interactions from women who had participated in previous COSgroups, allowed mothers to build observational skills. Starting when their infantsreached two months of age and continuing for the duration of the intervention,mothers took turns being the focus of a session. For each session, the therapistsselected four clips of the target mother interacting with her infant in a variety ofactivities in the residential treatment center (e.g., feeding, free play, and face-to-faceinteractions) to view and discuss during the group session. The first clip was intendedto activate the mother’s caregiving system (e.g., her infant crying); the second clipshowcased the mother responding sensitively in a situation in which she oftenstruggled (e.g., she comforted her distressed infant instead of ignoring him); the thirdclip demonstrated a remaining struggle (e.g., she missed her infant’s signal to beheld); and the fourth clip celebrated the mother–infant dyad and provided

340 J. Cassidy et al.

affirmation of the mother’s parenting by showcasing her competence in an area inwhich she had previously struggled. These individualized sessions enabled mothers toaccomplish the COS goals described earlier: (1) become better at understanding theirinfants’ needs; (2) enhance their observational and inferential skills; (3) recognizetheir emotional responses to their infants’ behavior and understand how theseresponses influence their infant; (4) learn to regulate their emotions; and (5) respondto their infants’ signals appropriately.

Measures

Strange Situation Procedure (Ainsworth, Blehar, Waters, & Wall, 1978). Thiswidely used 20-minute laboratory procedure, in which the parent and a femalestranger alternately leave and return to the infant who remains in the toy-filledplayroom, was used to assess infant attachment to mother. Secure infants showinterest in interaction with or proximity to the parent, with little of the avoidancecharacteristic of the insecure/avoidant infants or the ambivalence characteristic ofthe insecure/ambivalent infants. Infants in a third insecure category demonstrate‘‘disorganized or disoriented’’ behavior (Main & Solomon, 1990). The reliability andvalidity of the Strange Situation have been extensively demonstrated (e.g., Belsky &Cassidy, 1994; Solomon & George, 2008).

Two highly reliable coders coded each videotaped Strange Situation. In order toensure that the principal coder (Susan Paris) remained blind to the nature of thissample, videotapes from this sample were intermixed for coding with videotapesfrom a moderately at-risk sample that she was coding at the same time. Because theassessments were conducted in the same laboratory room, and because participantswere similar in age and racial/ethnic background, this coder was unaware that morethan one sample was being coded. Jude Cassidy served as the second coder.Reliability between the two coders was high; coders agreed on placement across thefour groups for 16 of 20 infants (80%; k ¼ .55) and they agreed on placement insecure vs. insecure groups for 17 of 20 infants (85%; k ¼ .63). Disagreements weresettled by one of two additional coders blind to the nature of the sample, who agreedwith Paris in three of the four cases.

Maternal sensitivity during play. We used the NICHD mother–infant interactionscales (Owen, 1992) to code maternal responsive behavior during free-play. On afour-point scale ranging from 1 (not at all characteristic) to 4 (highly characteristic),we coded maternal sensitivity to distress, intrusiveness, and positive regard for herinfant (NICHD Early Child Care Research Network, 1999). A composite score,maternal sensitivity during play, was created by summing scores on all scales(intrusiveness was reverse-scored). Two highly trained coders who had previouslyattained reliability with the third author on videotapes from a separate sample codedall videotaped procedures; both coders were blind to the nature of the sample. Thesame procedures used to keep Strange Situation coders blind were followed here.Coders overlapped on a randomly selected 35% of cases, and reliability between thetwo coders (ICC) was .74. Scores of the more senior coder were used as data.

Experiences in Close Relationships scale (ECR; Brennan, Clark, & Shaver,1998). The ECR is a 36-item measure that taps secure versus insecure attachmentstyles. This measure is designed to tap global attachment style, a generalized style

Attachment & Human Development 341

evident across relationships. There are two subscales, each containing 18 items ratedon a seven-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). Theavoidance subscale taps the extent to which a person is uncomfortable with closenessand intimacy, is uncomfortable depending on others, and is uncertain that peoplecan be relied on when needed (e.g., ‘‘I prefer not to show a partner know how I feeldeep down,’’ ‘‘I am very uncomfortable being close to others’’). Higher scores reflectmore avoidant attachment. The anxiety subscale taps the extent to which a person isworried about being rejected, abandoned, or unloved (e.g., ‘‘I worry about beingabandoned,’’ ‘‘I worry that others won’t care about me as much as I care aboutthem’’). Higher scores reflect greater attachment anxiety. At enrollment, internalconsistency (Cronbach’s a) was .75 for avoidance and .88 for anxiety, and at post-intervention it was .94 for avoidance and .86 for anxiety.

Parental Acceptance-Rejection Questionnaire: Warmth/Acceptance and Hostility/Rejection subscales (PARQ; Rohner, 2001). The Warmth/Acceptance andHostility/Rejection subscales of the PARQ tap past experiences of parentalacceptance and rejection (e.g., ‘‘My mother said nice things about me,’’ ‘‘Mymother went out of her way to hurt my feelings’’). Participants respond to 35items by rating how accurately the statements reflect the way they feel about howtheir mother treated them on a four-point scale ranging from 1 (almost alwaystrue) to 4 (almost never true). Items on the acceptance scale were reverse-scoredand summed with items on the rejection scale. As such, possible scores rangefrom 35 to 140, with higher scores indicating greater perceived rejection.Reliability was high at both enrollment (Cronbach’s a ¼ .98). and post-intervention (Cronbach’s a ¼ .99) assessments.

Beck Depression Inventory-IA (BDI-IA; Beck, Rush, Shaw, & Emery, 1979; Beck& Steer, 1993). The BDI-IA is a 21-item self-report measure indicating presence/absence and severity of depressive feelings, behavior within the ‘‘past week,including today.’’ The BDI–IA has received extensive psychometric analysis andempirical validation (for a review, see Beck, Steer, & Carbin, 1988). For thepresent sample, internal consistency (Cronbach’s a) was .71 at enrollment and .89at the post-intervention assessment. Possible total scores range from 0 to 63, withhigher scores indicating greater depressive symptomatology. The widely usedclinical cut-off score of �10 indicates at least mild depressive symptoms (Beck &Steer, 1993).

Dissociative Experiences Scale (DES; Carlson & Putnam, 1993). The DES is a 28-item, widely used self-report measure with good psychometric properties (Carlson &Putnam, 1993). The scale taps several domains of dissociative behavior such asdissociative amnesia, spontaneous trance states, and identity alterations. Participantsrate the degree to which particular dissociative experiences apply to them inpercentages from 0% (never) to 100% (always). Higher scores indicate higher level ofdissociative experiences; a typically used clinical cut-off score is 30 (Carlson, Putnam,Ross, Torem, Coons, Dill, et al., 1993). Reliability was high at both enrollment(Cronbach’s a ¼ .96) and post-intervention (Cronbach’s a ¼ .96) assessments.

Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1979). The RSES has beenreported to be the most widely used self-report self-esteem scale (Robins, Hendin,

342 J. Cassidy et al.

& Trzesniewski, 2001). For each of the 10 items, participants rate their level ofagreement from 1 (agree very much) to 5 (disagree very much). Sample itemsinclude ‘‘At times I think I am no good at all’’ and ‘‘I feel that I have a numberof good qualities.’’ Strong psychometric properties have been reported and thereis extensive validation of its usefulness (Byrne, 1996). Five items are reversescored and then responses to all items are summed to create a total score withhigher scores indicating greater self-esteem (possible range 10–50). Reliability washigh at both enrollment (Cronbach’s a ¼ .85) and post-intervention (Cronbach’sa ¼ .91) assessments.

Social Support Questionnaire (SSQ; Sarason, Levine, Basham, & Sarason,1983). The SSQ is a 27-item measure that taps both the number of perceivedavailable supports as well as satisfaction with the perceived available support. Foreach item, participants list the number of people (up to nine) who provide the type ofsupport described in the item, and then use a six-point scale to rate their satisfactionwith this support. Mean responses are calculated across all 27 items, giving eachparticipant two scale scores: one for quantity, one for satisfaction. Higher scoresindicate more perceived available supports and greater satisfaction with thatsupport. Sarason and his colleagues have reported extensive reliability and validitydata for this measure (for a review, see Sarason, Shearin, Pierce, & Sarason, 1987).In the present study, reliability (Cronbach’s a) for number of supports was .91 atenrollment and .87 at post-intervention; reliability for satisfaction with support was.88 at enrollment and .96 at post-intervention.

Traumatic Antecedents Questionnaire (TAQ; Herman, Perry, & van der Kolk,1989). The TAQ is a 42-item self-report measure indicating occurrence (yes/no)and frequency (never, rarely, occasionally, often) of traumatic lifetime experiences(e.g., physical abuse, sexual abuse, witnessing violence, domestic chaos, neglect, andloss) during four age periods: birth to six years, 7 to 12 years, 13 to 18 years, andadulthood. For this report, we created variables tapping any occurrence before theage of 18 years of physical abuse (one item), sexual abuse (three items), andwitnessing violence (five items).

Results

Attachment distributions

Fourteen of 20 infants (70%) observed in the Strange Situation were classified assecurely attached to their mothers following intervention. We first compared theobtained frequencies to the frequencies expected based on meta-analytic reports of avariety of high-risk samples (Cyr et al., in press; van IJzendoorn et al., 1999). Theproportion of secure infants was significantly higher than proportions observed insamples of depressed parents (41%; w2 [1, N ¼ 20] ¼ 6.95, p 5 .05), low-SESsamples (48%; w2 [1, N ¼ 20] ¼ 3.88, p 5 .05), samples of substance-abusingmothers (26%; w2 [1, N ¼ 20] ¼ 20.13, p 5 .0001), and samples containingmaltreated infants (5%, p 5 .0001 for the exact binomial test of goodness of fitconducted due to small cell sizes). In addition, the proportion of security for thepresent study was not different from that for typical low-risk samples (i.e., non-clinical, middle-class North American samples [van IJzendoorn et al., 1999]: 62%; w2

[1, N ¼ 20] ¼ .54, ns; see Figure 1).

Attachment & Human Development 343

Four of 20 infants (20%) observed in the Strange Situation were classified asinsecure-disorganized. Because of small cell sizes, exact binomial tests of goodness-of-fit were conducted to determine whether the proportion of disorganizedattachment differed from that expected given meta-analytic reports of other high-risk samples (Cyr et al., in press; van IJzendoorn et al., 1999). Results indicated thatthe proportion of disorganization in the present sample was significantly lower thanthat found in samples of substance-abusing mothers (43%, p 5 .05) andmaltreating mothers (67%, p 5 .001), but not lower than in samples containingdepressed mothers (21%, ns) or in low-SES samples (25%, ns).1 Moreover, theproportion of disorganization in the present sample was not different from theproportion found in non-clinical, middle-class North American samples (15%, ns;see Figure 2).

Maternal sensitivity

To examine maternal sensitivity during play, we compared the maternalbehavior of participants in Tamar’s Children to that of mothers in a separatestudy conducted at the same time (Cassidy et al., 2010). Mothers in thiscomparison group were 33 first-time, economically stressed mothers selected fromthe control group of an existing randomized control study of interventionefficacy.2 Results of an independent samples t-test indicated that the maternalsensitivity of the mothers in Tamar’s Children (M ¼ 8.55, SD ¼ 2.42) did notdiffer from that of mothers in the comparison group (M ¼ 7.33, SD ¼ 2.26;t(51) ¼ 1.85, ns).

Figure 1. Percentage of securely attached infants: Comparison of infants in the Tamar’sChildren program with meta-analytic findings about infants in other samples.

344 J. Cassidy et al.

Maternal psychosocial functioning

We conducted a series of separate paired-samples t-tests to examine the differencesbetween maternal self-reports of psychosocial functioning at pre-intervention(enrollment) and post-intervention time points. Significant differences emerged foronly one of the seven aspects of psychosocial functioning examined. Post-intervention BDI scores tapping maternal depressive symptoms were significantlylower than those obtained at enrollment, t(12) ¼ 2.36, p ¼ .036. In fact, atenrollment, mean maternal BDI scores (M ¼ 12.38, SD ¼ 5.71) were above theclinical cut-off score of 10, whereas mean post-intervention scores (M ¼ 7.34,SD ¼ 8.76) were below this clinical cut-off point. At enrollment, 69% of theparticipants had BDI scores in the clinical range, which was true of only 38% ofmothers at the post-intervention time point, but a McNemar’s test for correlatedproportions failed to reach statistical significance (p ¼ .13). No participant movedfrom the non-clinical to the clinical range over the course of the intervention (seeTable 3).

Discussion

The experiences and characteristics of many incarcerated pregnant women put theirinfants at risk for becoming insecurely attached to them (Belsky et al., 2007; Hanset al., 1999; Hutchinson et al., 2008; Poehlmann, 2005; Putallaz et al., 1998; vanIJzendoorn, 1992). Prior to the studies presented in this volume, there had been noexamination of the attachment status of infants of women involved in the

Figure 2. Percentage of disorganized attachment: Comparison of infants in the Tamar’sChildren program with meta-analytic findings about infants in other samples.

Attachment & Human Development 345

correctional system or of whether, through intervention, this risk could be reduced.In the present study, following intervention in a jail-diversion program, we foundrates of infant attachment security and disorganization equivalent to those typical oflow-risk samples, and found that the mothers in our study were as sensitive asmothers in an existing community sample. These results are striking given the multi-risk nature of our sample.

The lack of a control group precludes the conclusion that it is the services offeredthrough Tamar’s Children that led to the study outcomes and, as such, theexploratory, program-development status of the current study must be emphasized.It is indeed possible that other factors accounted for these outcomes. For instance, itis possible that the mothers who were able to remain in the program and receive fulltreatment have characteristics that would have led their infants to develop secureattachments to them without intervention. This possibility is somewhat reduced,however, given that these mothers were comparable on many important dimensionsboth to mothers who left the program early and to incarcerated mothers moregenerally (US Department of Justice, 2008). Just as it is not possible to assume thatthe intervention led to the study outcomes, it is not possible to know whichintervention components (if any) were responsible. The comprehensive Tamar’sChildren intervention included multiple components designed to address themultiple stumbling blocks to infant attachment security and sensitive parentingthat the mothers faced. Moreover, it is possible that simply keeping mothers andinfants together (rather than sending infants to live in the community withnonmaternal caregivers, which is typically the case) may have been an interventionsufficient to reduce the risk of insecure attachment (see Murray & Farrington, 2008,for a review of the improved child development outcomes associated with alternativesentencing).

One component of the larger intervention that may have been particularlyimportant is the trauma treatment. Interviews that the program’s clinical director(Andrea Karfgin) conducted at various stages of the program with six staff membersand six mothers indicated widespread support for the notion that treatingparticipants for their trauma was crucial in helping mothers develop their capacitiesto set aside their problematic strategies for regulating emotion and to deal with the

Table 3. Comparison of mothers’ pre- and post-intervention psychosocial functioning.

MeasurePre-intervention Post-intervention

t-valueM (SD) M (SD)

ECRAvoidance 3.76 (.84) 3.36 (1.20) t(12) ¼ 1.48Anxiety 3.48 (1.27) 2.92 (.95) t(12) ¼ 1.40

BDI - IA 12.38 (5.71) 7.34 (8.76) t(12) ¼ 2.36*

DES 18.85 (19.67) 19.13 (15.27) t(12) ¼ 70.10RSES 23.56 (7.99) 23.67 (7.70) t(8) ¼ 70.05SSQNo. of supports 3.22 (1.60) 2.74 (1.18) t(10) ¼ 1.07Satisfaction 5.35 (.76) 5.50 (.70) t(11) ¼7.54

Note: Sample sizes vary due to missing data. ECR ¼ Experience in Close Relationships Scale; BDI -IA ¼ Beck Depression Inventory; DES ¼ Dissociative Experiences Scale; RSES ¼ Rosenberg Self-esteemScale; SSQ ¼ Social Support Questionnaire.

*p 5 .05.

346 J. Cassidy et al.

painful emotions that surfaced during other components of treatment. In addition,by focusing on the effect of past trauma on current feelings and behaviors, thetrauma treatment component may have helped the mothers learn to reflect on thecauses and consequences of their behaviors: processes that may have contributed tochange.

Interviews also indicated that both staff and participants regarded the COS-PPas a key factor in helping participants to mother their infants sensitively. Inparticular, mothers experienced ‘‘circle groups’’ as consistent, predictable, andnurturing sessions; qualities that may have allowed mothers to explore difficulttopics safely. By providing mothers with a secure base, these treatment groups mayhave allowed mothers, in turn, to serve as a secure base for their infants.Furthermore, staff and mothers indicated that the video-based nature of the COS-PP intervention provided a foundation for enhancing maternal sensitivity bydeveloping mothers’ observational skills before their infant’s arrival and by servingas a model for sensitive mothering. In addition, all mothers interviewed indicatedthat they considered COS-PP groups to be the best component of the largerintervention and viewed the COS-PP, along with trauma treatment, as fundamentalto the program as a whole.

The wrap-around social support services offered as part of the Tamar’s Childrenprogram may also have been critical to program outcomes given that mothers andinfants moved into the community-living phase of the program when infants were 6-months of age and still forming initial attachments to their mothers (Bowlby, 1982).Assuaging mothers’ stress related to obtaining economic resources such as housing,employment, child care, and health care during this transitional period by providingwrap-around social services may have been crucial to mothers’ capacities to remaindrug-free, behave sensitively, and foster security in their infants. The residentialphase of the program may also have been crucial: removing mothers from their usualexternal cues while they learned how to cope with painful emotions may have betterprepared them for abstaining from drugs once they entered back into thecommunity.

It is possible, in fact, that any one of the services offered through the Tamar’sChildren program was solely responsible for the study results. For example, giventhat low-SES alone is a risk factor for insecure attachment (van IJzendoorn et al.,1999), the wrap-around social services may have alleviated enough economic stressto create a reduced risk of attachment insecurity. It also could be argued that theCOS-PP alone was sufficient to foster secure infant attachment. These arguments,however, assume that the trauma, substance abuse, and mental health problemstypical of this population (each of which is associated with insensitive parenting andinsecure attachment; see Cassidy & Shaver, 2008) do not interfere with the ability ofthese women to rear securely attached children. Although it is possible that treatingonly one of the multiple risks would have reduced the rates of insecure anddisorganized attachment, it is less likely that the resulting rates would parallel soclosely those typical of low-risk, middle-class, North American samples.

We also found a decline in maternal self-reported depressive symptoms over time,a finding that is again difficult to attribute to the intervention because of the lack of acontrol group. Maternal depressive symptomatology was the only one of sevenassessed aspects of psychosocial functioning that changed over time, and thischange could result from chance. Moreover, at least one explanation for thisisolated improvement is unrelated to the intervention: it could be that the timing of

Attachment & Human Development 347

the pre- and post-intervention assessments was particularly linked to events thatmight be related to depressive symptoms. The improvement could simply reflect thefact that at the post-intervention assessment, mothers had been living in thecommunity for 6 months and were no longer subject to the (pre-intervention)circumstances of facing incarceration while being about to give birth to an infantfrom whom they would be separated. It is interesting to note that one reason thatimprovement in other aspects of maternal functioning did not emerge over time maybe that pre-intervention functioning was within the normal range (e.g., this was thecase for maternal dissociation and self-esteem).

Conclusions and future directions

In sum, the present study indicates that it is possible for pregnant women in a jail-diversion program with a history of substance abuse to rear infants who are securelyattached to them. Given the multi-risk nature of our sample, this finding should notbe taken lightly. The study’s sample size is small, yet not substantially smaller thanthat of Ainsworth’s original observational study of infant attachment (N ¼ 23;Ainsworth et al., 1978). The findings suggest that future interventions hold promisefor improving the outcomes of mothers involved in the criminal justice system andtheir infants. The lack of a control group in this study, however, makes it impossibleto determine whether the outcomes for the intervention group were accounted for bytreatment and, if so, which components of the treatment were crucial. At the veryleast, the Tamar’s Children program permitted mothers and infants, who wouldotherwise have been separated immediately following the infant’s birth, to remaintogether and begin the initial bonding process.

The strongest experimental design for future intervention research would involvethe inclusion of a randomly assigned group of mothers who received existing services(‘‘business as usual’’). Several practical problems, however, are inherent in a designusing this ‘‘business as usual’’ control group. In US prisons, it is the norm to separatemothers and children for the duration of the mother’s sentence (Carlson, 2001);comparing intervention and control groups in terms of infant attachment security,therefore, would be clouded by the control infants’ having been prevented fromforming an attachment to their mothers. An alternate approach is to create andcompare different treatment groups, thereby shedding light on which treatmentcomponents are necessary and sufficient (i.e., comparing [a] a group receiving all ofthe Tamar’s Children treatment components, [b] a group receiving only the COS-PPintervention without additional support services, and [c] a group receiving onlysupport services without COS-PP treatment). Future research should also examinewhether intervention is effective in settings other than the one examined here, such asa prison nursery or a community-based setting. Given that an estimated 74% ofincarcerated women are single parents of minor children (Morton & Williams, 1998)and 25% are pregnant or have recently given birth when facing incarceration (Fogel,1995), it is important to examine the efficacy of intervening not only with pregnantwomen, but also with mothers who recently gave birth and mothers of olderchildren. Moreover, given that the children of incarcerated women are often caredfor by their grandmothers (Engstrom, 2008; Poehlmann, Park, Bouffiou, Abrahams,Shlafer, & Hahn, 2008), it will be important to examine whether includinggrandmothers in the intervention is a useful strategy for supporting both themothers and their children.

348 J. Cassidy et al.

Other research designs that would shed light on the success of intervention effortsare longitudinal designs that examine the children of individuals involved with thecriminal justice system by measuring a broad spectrum of child developmentoutcomes repeatedly beyond the first year (Dallaire, 2007). Comprehensivelongitudinal follow-up will allow us to investigate how pervasive and persistentintervention effects are in a variety of domains of functioning beyond infantattachment, including the child’s emotion-regulation capacities, school performance,internalizing and externalizing behaviors, and peer group competence. Similarly,repeated comprehensive follow-ups of mothers would add to our knowledge of theeffectiveness of intervention in areas of mothers’ functioning including recidivism,relapse, psychopathology, and relationships with their other children. (See Berlin,2005, for discussion of the ways in which enhancing sensitive mothering and themother–child relationship might contribute to improved maternal functioning in avariety of ways.) In addition, longitudinal follow-up of these dyads will answerquestions about how long intervention effects endure. For instance, measuringattachment security again at 18 months after six months without intervention wouldprovide information about stability of attachment-related outcomes. It will beimportant to know, for instance, whether infants secure at 12 months remain secureeven when their mothers are no longer receiving support services, or whetheradditional intervention is needed to maintain infant secure attachment.

Finally, studies with larger samples, more comprehensive data acquisition, andfollow-up assessments will allow tests of more complex models of mediation andmoderation. The intergenerational transmission of attachment patterns is onemediated model that remains untested for children with mothers involved in thecriminal justice system. In particular, attachment theory posits that a mother’srepresentations of attachment drive her responsiveness to her infant, which in turnshapes her infant’s attachment organization. This model has received both empiricaland meta-analytic support (see de Wolff & van IJzendoorn, 1997) but has yet to betested with incarcerated populations. Moderated models may be particularly usefulfor investigating differential treatment effects. For example, whether mothers’motivation to change moderates intervention effectiveness is an empirical questionthat remains to be addressed. Other potential moderators of interventioneffectiveness that could be examined include mothers’ early loss experiences, trauma,self-efficacy, and dissociative symptomatology. Given extensive evidence that infantsof opiod-dependent mothers show dysregulation of the central and autonomicnervous systems that can make parenting difficult (Velez & Jansson, 2008),examination of the moderating role of infant characteristics (e.g., neonatalabstinence syndrome) will also be important.

Acknowledgements

The research reported here was funded by a grant from the Substance Abuse and MentalHealth Services Administration to the Mayor and City Council for Baltimore City. We aregrateful to Danielle Dallaire and Marinus H. van IJzendoorn who provided helpful commentson an earlier draft of this paper. Portions of these data were presented at the 2007 meetings ofthe Society for Research in Child Development, Boston.

Notes

1. The van IJzendoorn et al. (1999) meta-analysis reported that low-SES samples andsamples of depressed mothers did not differ from non-clinical, middle-class NorthAmerican samples in the proportion of disorganized infants.

Attachment & Human Development 349

2. Because the intervention study from which the comparison participants were selectedfocused on irritable infants, the 33 control group participants whose infants did not meetcriteria for high irritability were used (i.e., for the 33 infants, the mean of two NeonatalBehavioral Assessment Scale scores [Brazelton, 1973] within the first month was 56).Comparison group mothers did not differ from Tamar’s Children mothers on race/ethnicity, but did differ on mean age such that Tamar’s Children mothers (M ¼ 33.3,SD ¼ 5.4) were significantly older than the comparison group mothers (M ¼ 23.9,SD ¼ 5.2), t(46) ¼ 5.76, p 5 .001.

References

Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment. Hillsdale,NJ: Erlbaum.

Baradon, T., Fonagy, P., Bland, K., Lenard, K., & Sleed, M. (2008). NewBeginnings: An experienced-based programme addressing the attachment relationshipbetween mothers and their babies in prisons. Journal of Child Psychotherapy, 34, 240–258.

Barkauskas, V.H., Low, L.K., & Pimlott, S. (2002). Health outcomes of a community basedprogram for pregnant women offenders and their infants. Journal of Midwifery andWomen’s Health, 47, 371–379.

Beck, A.T., Rush, A.J., Shaw, B.F., & Emery, G. (1979). Cognitive therapy of depression. NewYork: Guilford Press.

Beck, A.T., & Steer, R.A. (1993). Manual for the Beck Depression Inventory. San Antonio: ThePsychological Corporation.

Beck, A.T., Steer, R.A., & Carbin, M.G. (1988). Psychometric properties of the BeckDepression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8,77–100.

Belsky, J., Bell, B., Bradley, R., Stallard, N., & Stewart-Brown, S. (2007). Socioeconomic risk,parenting during the preschool years, and child health age 6 years. European Journal ofPublic Health, 17, 508–513.

Belsky, J., & Cassidy, J. (1994). Attachment theory and evidence. In M. Rutter & D.F. Hay(Eds.), Development through life: A handbook for clinicians (pp. 373–402). Oxford:Blackwell.

Belsky, J., & Fearon, R.M.P. (2008). Precursors of attachment security. In J. Cassidy & P.R.Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (2nded., pp. 295–316). New York: Guilford Press.

Belsky, J., & Jaffee, S.R. (2006). The multiple determinants of parenting. Developmentalpsychopathology, Vol. 3: Risk, disorder, and adaptation (2nd ed., pp. 38–85). Hoboken, NJ:Wiley.

Berlin, L.J. (2005). Interventions to enhance early attachments: The state of the field today. InL.J. Berlin, Y. Ziv, L. Amaya-Jackson, & M.T. Greenberg (Eds.), Enhancing earlyattachments: Theory, research, intervention, and policy (pp. 3–33). New York: GuilfordPress.

Bloom, B. (1995). Imprisoned mothers. In K. Gabel & D. Johnston (Eds.), Children ofincarcerated parents (pp. 21–30). New York: Lexington Books.

Bowlby, J. (1982). Attachment and loss. Vol. 1: Attachment. New York: Basic Books.Brazelton, T.B. (1973). Neonatal Behavioral Assessment Scale. Clinics in developmental

medicine (2nd ed). Philadelphia: J.P. Lippincott.Brennan, K.A., Clark, C.L., & Shaver, P.R. (1998). Self-report measurement of adult

romantic attachment: An integrative overview. In J.A. Simpson & W.S. Rholes (Eds.),Attachment theory and close relationships (pp. 46–76). New York: Guilford Press.

Bruns, D.A. (2006). Promoting mother–child relationships for incarcerated women and theirchildren. Infants and Young Children, 19, 308–322.

Byrne, B.M. (1996). Measuring self-concept across the life span: Issues and instrumentation.Washington, DC: American Psychological Association.

Byrne, M.W., Goshin, L.S., & Joestl, S.S. (2010). Intergenerational transmission of attachmentfor infants raised in a prison nursery. Attachment & Human Development, 12, 375–393.

Carlson, E. (1998). A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69, 1107–1128.

350 J. Cassidy et al.

Carlson, E.B., & Putnam, F.W. (1993). An update on the Dissociative Experiences Scale.Dissociation: Progress in the Dissociative Disorders, 6, 16–27.

Carlson, E.B., Putnam, F.W., Ross, C.A., Torem, M., Coons, P., Dill, D.L., et al. (1993).Validity of the Dissociative Experiences Scale in screening for multiple personalitydisorder: A multicenter study. American Journal of Psychiatry, 150, 1030–1036.

Carlson, J.R., Jr. (2001). Prison nursery 2000: A five-year review of the prison nursery at theNebraska Correctional Center for Women. Journal of Offender Rehabilitation, 33, 75–97.

Carlson, J.R., Jr. (2009). Prison nurseries: A pathway to crime-free futures. CorrectionsCompendium, 34, 17–22.

Cassidy, J., & Shaver, P.R. (Eds.) (2008). Handbook of attachment: Theory, research, andclinical applications (2nd ed). New York: Guilford Press.

Cassidy, J., Woodhouse, S.S., Sherman, L.J., Stupica, B., & Lejuez, C.W. (2010). Enhancinginfant attachment security: An examination of treatment efficiency and differentialsusceptibility. Manuscript submitted for publication.

Cooper, G., Hoffman, K.T., & Powell, B. (2003). The Circle of Security Perinatal Protocol.Unpublished manuscript,Marycliff Institute.

Cooper, G., Hoffman, K.T., Powell, B., & Marvin, R. (2005). The Circle of Securityintervention: Differential diagnosis and differential treatment. In L.J. Berlin, Y. Ziv, L.M.Amaya-Jackson, &M.T. Greenberg (Eds.), Enhancing early attachments: Theory, research,intervention, and policy (pp. 127–151). New York: Guilford Press.

Cordero, L., Hines, S., Shibley, K., & Landon, M. (1992). Perinatal outcome and women inprison. Journal of Perinatology, 10, 205–209.

Cyr, C., Euser, E.V., Bakermans-Kranenburg, M.J., & van IJzendoorn, M.H. (in press).Attachment security and disorganization in maltreating and high-risk families: A series ofmeta-analyses. Development and Psychopathology.

Dallaire, D. (2007). Children with incarcerated mothers: Developmental outcomes, specialchallenges and recommendations. Journal of Applied Developmental Psychology, 28, 15–24.

De Wolff, M.S., & van IJzendoorn, M.H. (1997). Sensitivity and attachment: A meta-analysison parental antecedents of infant attachment. Child Development, 68, 571–591.

DeKlyen, M., & Greenberg, M.T. (2008). Attachment and psychopathology in childhood. InJ. Cassidy & P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinicalapplications (2nd ed., pp. 637–665). New York: Guilford Press.

Dozier, M., Stovall-McClough, K.C., & Albus, K.E. (2008). Attachment and psychopathol-ogy in adulthood. In J. Cassidy & P.R. Shaver (Eds.), Handbook of attachment: Theory,research, and clinical applications (2nd ed., pp. 718–744). New York: Guilford Press.

Elicker, J., Englund, M., & Sroufe, L.A. (1992). Predicting peer competence and peerrelationships in childhood from early parent–child relationships. In R.D. Parke & G. Ladd(Eds.), Family-peer relationships: Models of linkage (pp. 77–106). Hillsdale, NJ: Erlbaum.

Engstrom, M. (2008). Involving caregiving grandmothers in family interventions whenmothers with substance use problems are incarcerated. Family Process, 47, 357–371.

Fogel, C.I. (1995). Pregnant prisoners: Impact of incarceration on health and health care.Journal of Correctional Health Care, 2, 169–190.

Fogel, C.I., & Belyea, M. (2001). Psychological risk factors in pregnant inmates: A challengefor nursing. American Journal of Maternal Child Nursing, 26, 10–16.

Goldberg, S. (1997). Attachment and childhood behavior problems in normal, at-risk, andclinical samples. In L. Atkinson & K.J. Zucker (Eds.), Attachment and psychopathology(pp. 171–195). New York: Guilford Press.

Hairston, C.F., Bates, R.E., & Lawrence-Wills, S. (2003). Serving incarcerated mothers and theirbabies in community-based residences (Research Brief: Children, Families, and the CriminalJustice System), University of Illinois at Chicago, Jane Addams College of Social Work.

Hans, S., Bernstein, V., & Henson, L. (1999). The role of psychopathology in the parenting ofdrug-dependent women. Development and Psychopathology, 11, 957–977.

Herman, J.L., Perry, J.C., & van der Kolk, B.A. (1989). Childhood trauma in borderlinepersonality disorder. American Journal of Psychiatry, 146, 490–495.

Hertsgaard, L., Gunnar, M., Erikson, M.F., & Nachmias, M. (1995). Adrenocortical responseto the strange situation in infants with disorganized/disoriented attachment relationships.Child Development, 66, 1100–1106.

Attachment & Human Development 351

Hoffman, K.T., Marvin, R.S., Cooper, G., & Powell, B. (2006). Changing toddlers’ andpreschoolers’ attachment classifications: The Circle of Security intervention. Journal ofConsulting and Clinical Psychology, 74, 1017–1026.

Hutchinson, K.C., Moore, G.A., Propper, C.A., & Mariaskin, A.R. (2008). Incarceratedwomen’s functioning during pregnancy. Psychology of Women Quarterly, 3, 440–453.

Kestenbaum, R., Farber, E.A., & Sroufe, L.A. (1989). Individual differences in empathyamong preschoolers: Relation to attachment history. New Directions for Child Develop-ment, 44, 51–64.

Kubiak, S., Young, A., Siefert, K., & Stewart, A. (2004). Pregnant, substance-abusing, andincarcerated: Exploratory study of a comprehensive approach to treatment. Families inSociety, 85, 177–186.

Lyons-Ruth, K., Easterbrooks, M.A., & Cibelli, C. (1997). Infant attachment strategies, infantmental lag, and maternal depressive symptoms: Predictors of internalizing andexternalizing problems at age 7. Developmental Psychology, 33, 681–692.

Lyons-Ruth, K., & Jacobvitz, D. (2008). Attachment disorganization: Unresolved loss,relational violence, and lapses in behavioral and attentional strategies. In J. Cassidy &P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications(2nd ed., pp. 666–697). New York: Guilford Press.

Main, M., & Solomon, J. (1990). Procedures for identifying infants as disorganized/disoriented during the Ainsworth strange situation. In M.T. Greenberg, D. Cicchetti, &E.M. Cummings (Eds.), Attachment in the preschool years: Theory, research, andintervention (pp. 121–160). Chicago: University of Chicago Press.

Marvin, R.S., Cooper, G., Hoffman, K., & Powell, B. (2002). The Circle of Security project:Attachment-based intervention with caregiver–preschool child dyads. Attachment &Human Development, 4, 107–124.

Morton, J.B., & Williams, D.M. (1998). Mother/child bonding: Incarcerated womenstruggle to maintain meaningful relationships with their children. Corrections Today, 60,98–104.

Murray, J., & Farrington, D.P. (2008). The effects of parental imprisonment on children. InM. Tonry (Ed.), Crime and justice: A review of research (vol 37, pp. 133–206). Chicago:University of Chicago Press.

Myers, B., Smarsh, T., Amlund-Hagen, K., & Kennon, S. (1999). Children of incarceratedmothers. Journal of Child and Family Studies, 8, 11–25.

NICHD Early Child Care Research Network (1999). Child care and mother–child interactionin the first 3 years of life. Developmental Psychology, 35, 1399–1413.

Owen, M.T. (1992). The NICHD study of early child care mother–infant interaction scales.Unpublished manuscript, Timberlawn Psychiatric Research Foundation, Dallas, TX.

Poehlmann, J. (2005). Representations of attachment relationships in children of incarceratedmothers. Child Development, 76, 679–696.

Poehlmann, J., Park, J., Bouffiou, L., Abrahams, J., Shlafer, R., & Hahn, E. (2008).Representations of family relationships in children living with custodial grandparents.Attachment & Human Development, 10, 165–188.

Putallaz, M., Costanzo, P.R., Grimes, C.L., & Sherman, D.M. (1998). Intergenerationalcontinuities and their influences on children’s social development. Social Development, 7,389–427.

Radke-Yarrow, M., McCann, K., DeMulder, E., & Belmont, B. (1995). Attachment in thecontext of high-risk conditions. Development and Psychopathology, 7, 247–265.

Robins, R.W., Hendin, H.M., & Trzesniewski, K.H. (2001). Measuring global self-esteem:Construct validation of a single-item measure and the Rosenberg Self-Esteem Scale.Personality and Social Psychology Bulletin, 27, 151–161.

Rohner, R.P. (2001). Handbook for the study of parental acceptance and rejection (rev. ed.).Storrs, CT: Rohner Research.

Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.Sable, M., Fieberg, J., Martin, S., & Kupper, L. (1999). Violence victimization experiences of

pregnant prisoners. American Journal of Orthopsychiatry, 69, 392–397.Sarason, B.R., Shearin, E.N., Pierce, G.R., & Sarason, I.G. (1987). Interrelations of social

support measure: Theoretical and practical implications. Journal of Personality and SocialPsychology, 52, 813–832.

352 J. Cassidy et al.

Sarason, I.G., Levine,H.M., Basham,R.B.,&Sarason,B.R. (1983).Assessing social support: TheSocial Support Questionnaire. Journal of Personality and Social Psychology, 44, 127–139.

Schroeder, C., & Bell, J. (2005). Labor support for incarcerated pregnant women: The DoulaProject. The Prison Journal, 85, 311–328.

Shaw, D., Owens, E., Vondra, J., & Keenan, K. (1996). Early risk factors and pathways in thedevelopment of early disruptive behavior problems. Development and Psychopathology, 8,679–699.

Shelton, B.J., & Gill, D.G. (1989). Childbearing in prison: A behavioral analysis. Journal ofObstetric, Gynecologic and Neonatal Nursing, 18, 301–308.

Solomon, J., & George, C. (2008). The measurement of attachment security and relatedconstructs in infancy and early childhood. In J. Cassidy & P.R. Shaver (Eds.), Handbookof attachment: Theory, research, and clinical applications (2nd ed., pp. 383–416). NewYork: Guilford Press.

Spangler, G., & Grossmann, K.E. (1993). Biobehavioral organization in securely andinsecurely attached infants. Child Development, 64, 1439–1450.

Sroufe, L.A. (1983). Infant–caregiver attachment and patterns of adaptation in preschool: Theroots of maladaptation and competence. In M. Perlmutter (Ed.), The Minnesota Symposiaon Child Psychology: Vol. 16. Development and policy concerning children with special needs(pp. 41–83). Hillsdale, NJ: Erlbaum.

US Department of Justice. (1998). Women offenders: Programming needs and promisingapproaches (NCJRS Publication No. NCJ 171668). Rockville, MD: Author.

US Department of Justice. (2002). Services for families of prison inmates: Special issues incorrections (No. 017272). Longmont, CO: Author.

US Department of Justice. (2008). Parents in prison and their minor children (NCJRSPublication No. NCJ 222984). Rockville, MD: Author.

van IJzendoorn, M.H. (1992). Intergenerational transmission of parenting: A review of studiesin nonclinical populations. Developmental Review, 12, 76–99.

van IJzendoorn, M.H., Schuengel, C., & Bakermans-Kranenburg, M.J. (1999). Disorganizedattachment in early childhood: Meta-analysis of precursors, concomitants, and sequelae.Development and Psychopathology, 11, 225–249.

Velez, M., & Jansson, L.M. (2008). The opioid dependent mother and newborn dyad:Nonpharmacologic care. Journal of Addiction Medicine, 2, 113–120.

Wiewel, B., & Mosley, T. (2006). Family Foundations: A new program for pregnant andparenting women offenders with substance abuse histories. Journal of OffenderRehabilitation, 43, 65–91.

Williams, L., & Schulte-Day, S. (2006). Pregnant in prison—The incarcerated woman’sexperience: A preliminary descriptive study. Journal of Correctional Health Care, 12,78–88.

Attachment & Human Development 353