Skip to main content

An empirical evaluation of controversial practices in popular adoption manuals

Marjorie Lindner Gunnoe & Emily Helder


Co-sleeping, regressive parenting, and punishment the first year post-adoption were used to predict externalizing, internalizing, attachment disturbances and full-scale IQ at five years post-adoption for 38 children adopted internationally at ages 0-11. Co-sleeping predicted less internalizing. Regressive parenting predicted fewer attachment disturbances and lower IQ in older children (adopted at ages 4-11). Regressive parenting was unrelated to adjustment at the five-year anniversary in younger children (adopted at ages 0-4). Earlier onset of nonphysical punishment predicted less internalizing; earlier onset of physical punishment predicted higher IQ. Analyses are framed by a review/critique of popular “parenting manuals” for adoptive parents that encourage very different degrees of intrusiveness in the promotion of attachment and exercise of parental control. 


Adoptive parents face unique challenges. Awareness of this has prompted a corpus of popular trade books geared toward adoptive parents including the Top 5 Books You Must Read When Adopting from International Adoption or Foster Care (New Beginnings, 2017). These are:

Adopting the Hurt Child (Keck & Kupecky, 1995, revised 2009a)

Toddler Adoption (Hopkins-Best, 1997, revised 2012)

The Connected Child (Purvis, et al., 2007)

Beyond Consequences, Logic and Control (Forbes & Post, 2010)

Twenty Things Adopted Kids Wish their Parents Knew (Eldridge, 1999)

(Because the latter is not so much a manual as a delving into the elements of loss an adopted child might feel, subsequent references to the four “must-read” manuals refer to this list minus Eldridge.)

Some practices discussed in these manuals are controversial. Three such controversial practices are: co-sleeping, regressive parenting, and punishment. As a developmental psychologist who has adopted internationally (first author) and a clinical researcher who works with adoptive families (second author), we meet adoptive parents who are concerned about whether they can, should, or must engage in these practices. As experienced researchers, we presume that broad response tendencies and global parenting style are much more influential than specific practices, but we would still like to be able to offer data-informed opinions on the use of these practices. Unfortunately, empirical evaluation of these specific practices in an adoption context is meager.

The objective of this report was to explore whether the three controversial practices could be categorized as likely helpful or likely harmful in a high-risk adoption context (i.e., children adopted from institutional care and/or after chronic abuse and neglect). We used data from an intensive study of families who had adopted internationally, seeking links between parenting the first year and child adjustment at the five-year anniversary of the adoption. We tested four manifestations of child adjustment: externalizing, internalizing, attachment disturbances, and full scale IQ. Because data about parenting the first year were retrospective for many of our families (and we do not have retrospective child adjustment data), it is important to explicitly state that our analyses do not control for pre-adoption, or even “at adoption,” child adjustment. Accordingly, we do not claim that these practices caused the outcomes with which they are associated. We simply categorize the three early practices as associated (or not) with four manifestations of child adjustment at the five-year mark. 

Areas of agreement in the four must-read adoption manuals

Approaches to parenting in the scholarly literature include traditional Attachment theory, Behaviorism, and Baumrind’s parenting styles. Almost universally, adoption manuals emphasize that these standard approaches are insufficient for adopted children with non-standard life experiences, especially in the realm of attachment. Four premises that undergird the four must-reads are:

1. The adopted child has experienced disrupted attachment and trauma and is now wounded/hurt.

2. Intentional “reparenting” is required to help the child develop a secure attachment.

3. Attachment to the adoptive parents mediates the relationship between early trauma and subsequent            thriving.

4. Reparenting requires a “new toolbox” of parenting practices.

Applied, this means that parenting practices that work with non-adopted, typically-developing children may be inappropriate, or have little effect, in an adoption context. Alternatively, practices that are inappropriate for non-adopted children may be helpful for children who have experienced pre-adoption trauma. Per Keck and Kupecky (2002): “Many of these ideas might sound contrary to good parenting. In fact, they would be if you were parenting a securely attached child” (p. 62, emphasis theirs).

The call for a new toolbox is both contra-indicted and supported in the scholarly research. The call is contra-indicated by studies of adoptive families demonstrating that broad response tendencies (e.g., responsiveness) and global parenting styles (e.g., authoritative) tend to “work” in standard ways in adoptive families (Juffer, et al., 2005; Garvin, et al., 2012; Pitula, et al, 2019; Stams, Juffer & van IJzendoorn, 2002; Tan, et al., 2012, 2015; Santos-Nunes, et al., 2017).

The call for a new toolbox is supported by studies suggesting that child adjustment is less directly predicted by variations in parental support than is typically found in the general parenting literature. Despite similar or above average support in adoptive homes compared to nonadoptive homes, researchers have documented higher rates of family conflict (Reuter, et al., 2009) and parent-reported externalizing (Stams, Juffer, Rispens, et al., 2000). Croft et al. (2001) characterized the emotional valence of the parent-child relationship in adoptive homes as primarily a function of child competence rather than parent-to-child positivity/negativity. Additionally, child adjustment may be more directly predicted by variations in parental control in adoptive (vs. nonadoptive) families. Studies assessing control suggest that adopted children may benefit more than nonadopted children from structure/limit-setting (Lawler, et al., 2017). They may even benefit from strict control when deprivation has been extreme (Audet & Le Mare, 2011).

Adoption specialists are more likely to agree on the appropriateness of broad tendencies than specific practices. While all the adoption authors we have read (scholarly and popular) promote high responsiveness, only some promote co-sleeping and regressive parenting. This is because these practices can be responsive or not, depending on the child (e.g., some children are scared to be alone at night; others fear physical contact). Similarly, punishment (even spanking) can be authoritative or authoritarian, depending on the manner in which it is administered (Baumrind (1996, p. 412). Because knowing the broad dimensions of good parenting may be insufficient to classify the appropriateness of particular practices in particular contexts, research on specific practices is also needed.

Primary controversy across the four must-read adoption manuals

While the authors of the four must-reads agree that reparenting requires a new toolbox, they disagree on some tools that go into the box. Paralleling the general parenting literature, the primary overarching disagreement in adoption manuals is the appropriateness of power-assertive parental control. Representing the two poles are Keck and Kupecky (high power-assertive) and Forbes and Post (low power-assertive).

Keck and Kupecky have been very influenced by Foster Cline who is known for various techniques geared toward compelling attachment in children with attachment disturbances. One such technique is “holding,” wherein a child is restrained across the laps of two therapists and/or parents, while the lead holder forces the child to maintain eye-contact and confront their early trauma (Cline, 1992). This emotionally intrusive experience putatively “jump starts” arrested emotional development (Keck & Kupecky, 1995, p. 174). (For a theoretical refutation of this supposition, see Weir, 2011.) Holding is very controversial and sometimes physically dangerous and has been denounced by the American Professional Society on the Abuse of Children (Chaffin et al., 2006). Cline is also known for the much less controversial book Parenting with Love and Logic (Cline & Fay, 1990). This book is geared toward a general (i.e., non-adoptive) audience and helps parents influence children by orchestrating “natural” consequences for bad behavior.

Keck and Kupecky have two popular manuals, Adopting the Hurt Child (1995, 2009a) and the companion book Parenting the Hurt Child (2002, 2009b). Although each book devotes only a few pages to holding, their underlying premise is that parents must exert strong Cline-style control over a hurting child. In this model, the hurting child copes with early abuse and neglect by developing an exaggerated need for control. They maintain this sense of control by ingratiating themselves to easily manipulated, “stupid” adults (2009a, p. 170; 2002, p. 31), and by making sure that they never do exactly what an authority figure directs them to do. Sometimes the child’s noncompliance is overt. Other times it is so subtle the parent may not even be aware of it (e.g., when directed to pack their lunch and then put on their shoes, the child does the tasks in reverse order to assure self that they are not being controlled). Although terrified of relinquishing control, the hurting child actually wants their parents to be in charge, “in a position to do what parents are supposed to do: provide, protect, love, nurture, and control” (2009, p. 197). The goal then, is to help/force a child to transfer control from self to parents. This transfer is fostered by the techniques emphasized in most adoption books (e.g., responsiveness, fidelity) combined with overt, covert, and sometimes feigned control in the few battles the parents decide to “pick” (e.g., “Jonathon, how did you know that today I didn’t want your dishes in the dishwasher? I want to rinse the dishes in a special way, and you’re making it easier for me to do that. Thanks!” (Keck & Kupecky, 2002, pp. 66-7).)

An explicitly-titled rebuttal to the Keck & Kupecky/Cline approach is the Forbes and Post book: Beyond Consequences, Logic and Control (2010). The main tenets of this book are that: “all negative behavior arises from an unconscious fear-based state of stress” (p. 13), that love must replace fear, and that love is not controlling. “To love a child,” write these authors, “is merely to be present …in harmony with the child’s needs and desires” (p. 14). Adult-attempts to overtly control a child signify that “we feel threatened or fearful about some aspect of ourselves” (p. 24). Forbes and Post refer to their model as The Stress Model, encouraging parents to respond similarly to all negative behaviors: Assume a nonthreatening posture, tell the child that their behavior is the result of stress, assure them they are safe and loved, and encourage them to come to the parent next time they feel stressed. When parents feel that they must attempt to influence, they should give children the physical and emotional space they need to decide that a request is not a threat (e.g., a one-armed hug and walk away) (p. 88). Forbes and Post claim that once a child feels safe, their self-regulatory system can “catch-up” to its developmental potential and negative behaviors will decrease in “two weeks to thirty days” (p. 56).

In between the poles are the manuals by Hopkins-Best (1997, 2012) and Purvis et al. (2007). Hopkins-Best advocates high control over the attachment process. In her first edition, she credits Cline for various control techniques including “rage-reduction holding therapy” (p. 209; see also pp. 201-2 in Hopkins-Best, 1997). She stresses that: “Recreating attachment is an intrusive[1] process… It involves staying in their faces and imposing the care-giving role upon the (often-resistant) child” (p. 196, 1997). However, Hopkins-Best is not comfortable with Cline-style responses to acting-out. She reports trying two different types of reasoning before eventually restraining her own toddler’s hands when he clawed at her eyes (1997, p. 230; 2012, p. 240).

Conversely, Purvis et al. are more comfortable with behavioral control than intrusive attachment techniques. They place more emphasis on nurturing intimacy than compelling attachment. Their back cover invites parents to “discipline with love.” Parents are instructed to “embody authority” and explicitly “state expectations and consequences” (p. 122). Their first rule for a connected family is that a child “must not dominate the family though tantrums, aggression, back-talk, whining or any other tactic” (p. 136). Although they offer little advice about how to curb these behaviors if a child persists in them, they do direct parents to physically intervene in situations that can be remedied, after giving a child the opportunity to “choose” to remedy the situation themselves (e.g., “It is NOT okay to take other people’s things. You have two choices. You can give it back or I will give it back,” p. 131).

In sum, authors of the four must-reads have very different attitudes toward power-assertive control. Thus, it is not surprising that they disagree on some practices.

Three controversial practices


Parent-child co-sleeping has been variously defined in the scholarly literature as bed-sharing, room-sharing or either of these. Some of the controversy surrounding co-sleeping is a function of this broad definition applied to infants. The American Academy of Pediatrics (APA, 2022) strongly opposes bed-sharing with infants younger than 12 months of age — due the risk of Sudden Infant Death Syndrome (SIDS) — and actively promotes room-sharing with infants 0-6 months of age.

With older children, the debate over co-sleeping focuses on psychological factors. A comprehensive review of 659 papers on the sleep contexts of children under age 18 (Mileva-Seitz, et al, 2017) linked co-sleeping (variously defined) to both desirable and undesirable psychological characteristics. Some desirable characteristics that could be sequelae of co-sleeping included: lower infant cortisol reactivity and higher child cognitive competence, empathy, conscience, and daytime social independence. Some undesirable characteristics that could be sequelae of co-sleeping included: less active sleep at 6 months (important for infants’ neurological development), sleep problems post-infancy, and higher rates of cognitive and behavior problems. Because many studies have been cross-sectional and failed to distinguish whether co-sleeping was intentional vs. reactive (in response to pre-existing child problems), the authors refrained from endorsing or discouraging co-sleeping. They also stressed the importance of cultural context, noting that many of the undesirable associations documented in North America are not found in much of the rest of the world, where co-sleeping is normative, into middle childhood in some countries.

Subsequent to the Mileva-Seitz review, undesirable psychological associations have been reported for two large longitudinal studies in Brazil and China. Santos-Nunes et al. (2017) found higher rates of internalizing and “any psychiatric disorder” at age 6 among persistent co-sleepers, after controlling for excessive infant crying. In one of the most rigorous studies to date, Chen, et al. (2018) documented higher levels of adolescent anxiety, depression and attention problems amongst former co-sleepers (vs. propensity matched controls), after controlling for preschool symptomatology.

While the view of co-sleeping in the empirical literature is more unfavorable than favorable, the reverse is true in the popular adoption literature. Without claiming specific positive outcomes, Hopkins-Best (2012, p. 182) and Keck & Kupecky (1995, p. 110) present co-sleeping as “helpful” and what an internationally-adopted child may be used to. More fervent support can be found online. Several adoption websites promote co-sleeping as a contributor to post-adoption attachment (e.g., Galen, 2017; Phillips, 2021; Robbins, 2016). To back up their recommendation, they appeal to “attachment parenting” (AP; aka “natural parenting”), a popularized spin-off of traditional Attachment theory. The popularized approach (but not the scholarly approach), endorses co-sleeping and “extensive carrying and holding of infants”(Miller & Commons, 2010, p. 1).

Unfortunately for advocates of popularized AP, we could find no direct evidence that co-sleeping promoted secure attachment. This empirical gap was acknowledged by Miller and Commons (p. 8) who wrote that “It is possible, although this has not been shown empirically, that infants of parents who use AP practices … are more likely to become securely attached.” Lack of evidence was also stressed by Mileva-Seitz et al. (2017) who said: “The association between bed-sharing and secure attachment … has not been directly tested” (p. 11).

Intuitively-appealing as they may be, claims that co-sleeping promotes attachment post-adoption are particularly premature. We located only one study linking co-sleeping to any aspect of child adjustment in adoptive families. Tan and colleagues replicated the oft-reported link between co-sleeping and sleep problems in a large international sample of preschool girls adopted from China mostly before the age of 18 months. Because their data were cross-sectional, they used supplemental qualitative data to “cautiously” suggest that co-sleeping was primarily a response to sleep problems rather than reciprocal. They concluded that this relationship needed to be investigated with a design that permitted “stronger causal inference” (Tan, et al., 2009, p. 438). Although our data are insufficient to permit causal inference, they are longitudinal which is an improvement over cross-sectional. Because co-sleeping was associated with negative outcomes in the particularly rigorous study by Chen et al. (2018) discussed prior, and in one study of adoptive families, we hypothesized that early co-sleeping would predict poorer child adjustment.

Regressive parenting

Most parents do not encourage their children to return to a less-developed state, but all four of the must-read adoption manuals encourage parents to encourage regression, to give children a second chance to develop attachment. This is particularly true of Hopkins-Best (1997, p. 138) who wrote:

There is considerable support for the theory that a child who has never experienced healthy symbiosis needs opportunities to regress to that stage before he will be ready to develop healthy attachment. This means that the child needs to go back to the emotional level of a young infant and progress through the developmental processes which he missed.”

So, how do parents orchestrate a return to emotional infancy? Different books suggest different techniques, but all four must-reads recommended bottle-feeding and/or hand-feeding (food placed directly in a child’s mouth). The rationale is that children will eventually transfer the comfort associated with food to the parent. The authors disagree about whether resistant children should be required to acquiescence to regressive feeding techniques. Hopkins-Best wrote: “Do not allow the toddler to bottle-feed himself” (1997, p. 200). Other authors direct parents to propose regressive techniques but (seemingly) permit the child to decline. For example, Forbes and Post (2010, p. 65) said: …”yes, begin bottle-feeding your twelve or even your fourteen year old. If a child needs it, regardless of his age, he will take the bottle.”

Another regressive technique is to keep adopted children in close proximity to the parent rather than encouraging them to interact with a wider circle of contacts and activities. Hopkins-Best (2012, p. 2015) encourages parents to “limit his interaction with a lot of other adults for a while after arrival.” Keck & Kupecky suggest that one parent stay home with the child for several years if possible (2009, p. 119). Forbes and Post encourage cloistering. In response to stealing, these authors direct parents to: “Close the doors to the bedrooms and keep your child in a contained space as close to you as possible all the time” (2010, p. 56).

To the best of our knowledge, the only empirical evaluation of regressive parenting in an adoption context was presented by us. In our report on “school age” children (ages 4-15 at time of adoption), we found that regressive parenting was associated with smaller gains in full-scale IQ and verbal memory from W1 to W3 (i.e., children improved more when parents did not use these techniques). Regressive parenting did not predict attachment disturbances, externalizing or internalizing in our school age children during this particular phase of our project (Helder, et al., 2014).

This left us unsure of what to hypothesize. The adoption manuals promote regressive parenting as beneficial to attachment regardless of child-age at adoption, but non-regressive parenting was associated with greater cognitive gain amongst our older children. Because we had no empirical evidence that regressive parenting was beneficial, we tentatively hypothesized that regressive parenting the first six months post-adoption would be associated with poorer adjustment at children’s five-year anniversary.


The use of punishment is increasingly a matter of controversy in the general parenting literature. The most controversial form of punishment is spanking, generally viewed as the mildest form of “physical” punishment. The National Association of Social Workers (NASW, 2003) explicitly opposed the use of physical punishment in all contexts. Similarly, an article based on a report by the American Psychological Association (APA) Task Force on Physical Punishment of Children recommends that “parents should avoid physical punishment and that psychologists should advise and advocate against it” (Gershoff et al., 2018, p. 626). In making these recommendations, the Task Force emphasized that “Decades of correlational research have linked more physical punishment with more problematic behavior in children (Gershoff, 2002; Gershoff & Grogan-Kaylor, 2016)” (p. 628). However, when methodology moves beyond bivariate associations, some benefits of spanking have been found. Larzelere and Kuhn (2005) found that spanking as a backup to timeout resulted in less defiance or aggression than 10 of 13 alternative tactics to which it was directly compared. The most rigorous meta-analysis of spanking to date used multiple methods to analyze change scores in externalizing and found “a small risk or a small benefit of spanking, depending on the adjustment method used” (Larzelere, et al., 2018, p. 238).

Some researchers who oppose spanking also oppose the use of nonphysical punishment including timeout and privilege removal. Eschewing all disciplinary consequences, they recommend only “strong positive parenting” techniques such as reasoning and emotion coaching to influence children’s behavior (Durrant, 2016, pp. 247-253; Holden, et al., 2017). This stance has also been challenged by Larzelere and colleagues who compared the documented effectiveness for exclusively positive parenting and for timeout and found that the latter had a much stronger track record of reducing children’s noncompliance (Larzelere, et al., 2020).

Whatever the links between punishment and child adjustment in the broader developmental and clinical literature, we cannot assume that these patterns apply in high-risk adoption contexts. (Remember: New toolbox needed.) Per Keck and Kupecky (2002), a child who has “coped” with food deprivation and the loss of the birth mother may hardly notice being deprived of a toy, and grounding can be counterproductive if it deprives the child of activities that they need to develop healthy peer relationships.

The discipline tool that gets the most attention in the four must-reads is timeout. According to Hopkins-Best (1997), timeouts are unlikely to work because they rely on a child wanting to be in close proximity and please their attachment figure. Moreover, timeouts can be counterproductive in that they: signify rejection, inhibit attachment, reinforce the child’s negative self-concept. They may also prompt the child to disassociate or self-soothe, thereby reinforcing the child’s unhealthy trust in self rather than parents (Keck & Kupecky, 2002; Purvis et al., 2007).

Claims that timeout with traumatized children further traumatizes them have been countered by Dadds and Tully (2019) in a comprehensive analysis of timeout. These researchers emphasize that there is considerable research evaluating the efficacy of parenting programs that include timeout within “child-welfare populations, and this research has found no evidence of adverse effects” (p. 9). For example, Pearl et al. (2012) synthesized results from 15 agencies delivering Parent-Child Interaction Therapy (PCIT) to high-risk families and reported improvement in child trauma symptoms, child dissociative symptoms, child behavior, and caregiver stress. Dadds and Tully speculate that child symptomatology is reduced because parents trained in time-out circumvent escalatory coercive processes and children improve their self-regulatory skills.

The four must-reads say less about physical discipline. Keck and Kupecky (2002, p. 118) echo Cline’s advice that parents should try various methods of discipline until they find one that works. Cline is generally disinclined to spank both psychologically healthy (Cline & Fay, 1990) or at-risk children (1992, p. 144), but sometimes recommends physical discipline as part of a treatment plan focused primarily on attachment. In contrast, Purvis et al. (2007, p. 91) say that “Spare the rod disciplinary techniques are not useful” (which we assume means that they do not approve). Neither Hopkins-Best nor Forbes and Post explicitly address spanking.

We included punishment in this research because the four must-reads say so little about how to compel immediate compliance (when necessary) or how to interrupt patterns of harm to others in the home. A major limitation of most sections of most adoption manuals is that the (literary) adopted child appears without siblings. This permits the luxury of advising mostly positive parenting techniques. For example, Purvis et al. (2007, pp. 114-5) articulate a nicely-sequenced response to an adoptive child intentionally hurting another child at a birthday party: lead the perpetrator to a “think it over” place, emotion-coach, and then lead the perpetrator back to apologize when she is ready. While this is good advice in a birthday party scenario where the parent of the victim can protect their child by not inviting the perpetrator back, it may be insufficient when aggression is chronic and the parent of the perpetrator is also the parent of a victim (i.e., one child within the home poses a chronic risk to another). Of the four must-reads, only Keck and Kupecky (1995, Chap. 13; 2009, Chap. 12) convey sufficient appreciation for the risk that an aggressive child can pose to other family members and to the family system as a whole. They explicitly state that parents may need to use forceful power assertion to protect others in the home.

As with the other two controversial practices, almost no research has investigated the impact of punishment in an adoption context. A PsycInfo title search of the terms adopt* plus any of: discipline, punish*, timeout, time-out, and spank* yielded only two directly relevant studies. One of these was our own report that a measure that combined consistency of consequences with parents’ reported valuing of rule-following was unrelated to externalizing and internalizing two years later (Helder, et al., 2020). The other (more explicit) investigation of punishment was a dissertation. Stillman (2015) surveyed 43 parents of children who had previously experienced a dissolved adoption (i.e., the current adoption was the second adoption for the child). About 80% of the children in the sample were age 11 or under at the time of second adoption and had been living in their new home less than five years.  Stillman asked parents to report on children’s “negative behaviors” and their discipline in two time periods: the first 90 days post-placement and in the most recent 90 days. Concerning the first 90 days, greater decreases in children’s negative behavior from Time 1 to Time 2 were positively associated with: shared-disciplining (rather than one parent primary), talking with the child, yelling/raising voice, and a tally of nine possible types of discipline (some “positive,” some punitive). In contrast, more recent use of timeout and corporal punishment was associated with less improvement in behavior from Time 1 to Time 2. This prompted Stillman to recommend that parents begin disciplining early in the adoption.

There are two particularly commendable aspects of Stillman’s methodology. In contrast to the four must-reads which take a somewhat static approach to punishment (recommending or discouraging specific techniques irrespective of how long the child has been in the family), Stillman carefully attended to the timing of post-adoptive discipline. Stillman also employed dichotomous measures of discipline (e.g., timeout used/not used) rather than measures of frequency. While frequency-based measures may be preferable when researchers have the capacity to partial out pre-existing levels of child problems, this is not an option for researchers studying children whose problems predate adoption. Of course, any measure of administered punishment is inherently confounded with child problems, but non-frequency-based measures are probably less confounded than frequency-based ones. Like Stillman, we were interested in the timing of punishment and recognized the confounding risk inherent in frequency-based measures of punishment. Thus, we collected “onset of punishment” data, asking parents how soon they began using nonphysical and physical punishment post-adoption.

Onset is a decent variable in a less-than-ideal measurement situation, but once again, we had minimal precedent on which to base a hypothesis. On the one hand, we appreciate the very negative emotional reactions that an at-risk child could have in response to even customary punishment. On the other hand, Stillman’s empirical findings were gleaned from a sample quite like our own (high risk, adopted post-infancy). Moreover, many successful parenting programs for at-risk children include a discipline component. Given these empirical findings with similar samples, we hypothesized that earlier onset of punishment would be associated with better adjustment at the five-year anniversary.

Brief summary of the current presentation

In the first wave of a four-wave study, parents were asked to provide retrospective data on parenting practices employed the first year after international adoption. Dichotomous and continuous measure of three controversial practices were used to predict four manifestations of children’s adjustment, five years post-adoption. Because the goal was to categorize practices, rather than “explain-away” association using covariates over which parents have no control (e.g., severity of deprivation), we increased our chances of identifying association by controlling only for age at adoption which was linked to all three practices.

We articulated hypotheses (better viewed as conjectures). Because the academic literature on the use of these practices in the context of adoption is so meager, we included demographic information about parents who engaged in these practices. These descriptions may be useful for therapists and/or subsequent researchers attempting to compare their sample with our own. Associations between the controversial practices were also sought.



Families who had adopted internationally were recruited through agencies that provide adoption and post-adoptive services to a large upper mid-west metro area. Parents were screened via telephone to ascertain that the adoption had occurred within the last five years and that there were no known medical conditions that would independently impact the child’s ability to participate in testing (e.g., Fetal Alcohol Spectrum Disorder, Down Syndrome).

The present report is limited to children who were under age 12 at the time of adoption. There were 44 children in our sample who met this criterion and participated at Wave 1 (W1). Two parents failed to return the take-home packet that included the focal parenting measures, and four left the study after the first wave, reducing the cases potentially available for the longitudinalanalyses to 38. (But not all children were eligible for every longitudinal analysis, as explained in a later section.) Attempts to discern differences between the 38 “completers” and the 6 non-completers yielded two significant differences on 14 demographic indicators tested: Non-completing mothers worked many more hours outside of the home (M for non-completers = 32.0 (SD = 9.8), M for completers = 9.2 (SD = 13.2), F (1,42) = 16.25, p < .001). Two of the non-completers (but none of the completers) were single mothers.

The 38 children available for at least one of our longitudinal analyses were adopted from 10 countries. Age at adoption ranged from 6 – 129 months. All adoptive parents were married, heterosexual, and had at least a high school education. Most mothers did not work outside the home at the time of the adoption and provided the majority of the child’s care. Because the large adoption agencies in our city are faith-based, most parents were religious and ranked religious or humanitarian motivation as their primary reason for adopting (for more on motivation, see Helder et al., 2020). Most children (84%) had siblings (biological to the adoptive parents and/or also adopted, not necessarily biologically related to the target child). Thus, our “prototypical” mother had prior parenting experience and enough financial resources/flexibility to stay at home following the adoption. Although we did not ask whether the mothers in our sample had read the specific adoption manuals used to frame this paper, our qualitative interviews revealed that most parents were involved in adoption-related groups where these kind of books are discussed. See Table 1 for more detailed information on the 38 children in the current study, as well as the parents who specifically employed the three controversial practices (per the dichotomized indicators).


The 38 eligible children participated in all four waves of data collection. The first three waves occurred at approximately one-year intervals (mean time from W1 to W2 = 11.7 mos (SD = 1.5); mean time from W2 to W3 = 11.9 mos (SD = 1.5)). W4 occurred approximately three years after W3 (mean time from W3 to W4 = 34.1 mos (SD = 7.1)). At each wave, the adopted child and one parent visited our university. Each visit took about 2.5 – 3 hours (with breaks appropriate for the child’s age and ability). During the visit, the child was tested by a trained member of our research team for a variety of cognitive skills while the parent was interviewed and completed standardized rating scales. Parent consent and child assent were obtained for all forms and tests.

Many of the survey instruments were identical across the four waves of the study. However, the W1 instruments included questions referring to three different time periods in the child’s life. One set of questions asked parents what they knew about the child’s pre-adoption situation. A second set of questions asked parents about their parenting practices the first year the child was in their home. (These are the focal variables for the present report.) A third set of questions asked about “now.”

No financial incentives were given for participation, but parents did receive a copy of their children’s cognitive test scores. The procedure and all data collection instruments were approved by our university’s IRB.

Specification of the five-year anniversary data (“Wave X”)

Although W4 data were  collected approximately five years after W1 data, it is important to state that W4 was not necessarily the child’s five-year anniversary in the home. As stated prior, families could enroll in the study anytime within the first five years post-adoption. At enrollment/W1, actual time since adoption ranged from 1 month to 58 months (M = 19.6, SD = 14.8).  This means that the five-year anniversary occurred at different waves for different children.

To specify five-year anniversary data, we extrapolated a set of “Wave X” scores by identifying which of the four waves was temporally closest to each child’s five-year anniversary. For children who had been in the home less than a year when they began the study, we designated Wave 4 as their Wave X. For one child who was already approaching five years in the home at the time of enrollment, we designated Wave 1 as their Wave X. For children who had been in the home 2-3 years at enrollment, their five-year anniversary corresponded with either Wave 2 or Wave 3. Thus, Wave X could be W1 (n = 1), W2 (n = 2), W3 (n = 15), or W4 (n = 20) for a particular child, but always corresponded to the wave closest in time to that child’s five-year anniversary in the adoptive home.

Readers comparing this report with previous reports by our research team need to distinguish the current approach (prediction of child-specific five-year anniversary data gathered at different waves for different children) from our prior approach (prediction of change in group means as the whole group moved through four waves of the study). Reports based on the prior approach include: Helder et al. (2014, 2016, 2020), Jensen, et al. (2016), and Gorter, et al. (2107).

Two sets of parallel, but distinct, analyses

For each controversial practice, we created a continuous variable for use in four regressions (one per adjustment variable) and a dichotomous variable for use in one MANCOVA (four adjustment variables in the same computation). Dichotomization was based on a meaningful distinction if there was one (e.g., yes/no); otherwise, at the midpoint of the distribution. Conducting somewhat parallel statistical analyses based on slightly different variables permitted us to test for “dosage” effects of the controversial practices. Parallel sets of analyses also afforded a check on significant results based on a small sample. (When a significant finding was obtained using one analytic approach, we expected that we would at least see the same direction of effects in the other approach, even if the inferential statistics were not always of the same magnitude.) Both the regressions and the MANCOVAs controlled for age at adoption.

Different sample sizes for different analyses

There were no missing data for the 38 children included in the five-year anniversary analyses but the sample size varied across the different analyses for two reasons. First, our survey instrument made reference to different lengths of time since adoption. While the analysis of sleeping arrangements “the first month” could include very recently adopted children, comparisons of children punished the first year (vs. not) required that children had been in the home a full year — to ensure an equal amount of time for each group to elicit punishment. Second, some practices are more appropriate for children of specific ages. Eligibility criteria based on both time-in-home and child age-at-adoption were applied to each longitudinal prediction. Criteria and resultant subsample sizes for each analysis are presented in Table 2.


Controversial Practices


Co-sleeping was operationalized as sleeping in the same bed or same room as a parent. Preliminary data examination revealed that the oldest co-sleeper in our sample was adopted at age 6 years and that co-sleeping arrangements did not last long (36% lasted one month; another 36% lasted 1-3 months). To ensure that our comparison of co-sleepers and non-co-sleepers was based on comparably-age children, analyses of co-sleeping were limited to children adopted prior to age 7. The dichotomous measure assessed whether the child co-slept the first full month (yes/no). The continuous measure indicates the total number of months the child co-slept, truncated at 6 months due to one extreme outlier (39 months).

Regressive Parenting

Regressive parenting was measured with a scale created by the authors. It appears in the Appendix of Helder et al. (2014) (as actually formatted). The scale includes five visual 5-point continuums, anchored by two contrasting statements (one regressive, the other non-regressive). For example, Continuum 1 contrasts Did everything possible to make sure child was cared for only by parents vs. Promptly established child in age-appropriate settings supervised by nonparents (e.g., nursery, preschool, school). Each continuum also includes a midpoint specifying “in the middle.” In data entry, the regressive pole in each pair was entered as 5 and the contrast was entered as 1. The other four items in this scale are: Continuum 2 = Did things to emphasize child’s dependence on parents for food (e.g., put child back on a bottle, placed food directly in child’s mouth) vs. Encouraged child to practice age-appropriate table manners; Continuum 3 = Made sure all nice things came only from parents vs. Permitted child to accept sweets, presents, hugs from relatives, friends and teachers; Continuum 4 = Encouraged regression/dependence on parents vs. Encouraged age-appropriate autonomous behavior; Continuum 5 = “Going easy” on child’s misbehavior vs. Confronting child’s misbehavior “head on.”

The items were crafted to apply (as much as possible) to children of diverse ages. The instructions specify that parents should report on their parenting the first six months post-adoption. To create the continuous measure of regressive parenting, scores on the five continuums were summed. Total scores could range from 5 – 25 with higher scores indicating a more regressive approach.[2] The actual range on this scale was 7 – 24 (M = 17.1, SD = 4.9, Cronbach’s α  = .79.) A Shapiro-Wilks test indicated sufficient normality in the distribution with no outliers. To create the dichotomous measure of regressive parenting, we divided cases at the theoretical midpoint of 15 (i.e., not the numerical mean). Parents who scored above the midpoint were assigned to the “regressive” group and parents who scored at or below the midpoint were assigned to the “non-regressive” group.

Earlier Onset of Punishment

To create the continuous indicator of onset of punishment we averaged parents’ response to two items: “How soon did you/spouse use non-physical punishment with your child (e.g., time-out, taking away privileges)?” and “How soon did you/spouse use physical punishment with your child (e.g., spanking, swatting)?” Response options ranged from (after a year = 1) to (immediately/first day or two = 6), so that higher averaged scores on this variable correspond with “sooner” (i.e., earlier) punishment. Parents could also choose (have not done this = 0).  Preliminary data examination indicated that all parents employed nonphysical punishment within the first year and 73% administered physical punishment. Earlier onset of one category did not predict earlier onset of the other (r(17) = -.22, p = .35). Scores on the continuous (combined) punishment variable ranged from 1.5 – 5 (M = 3.1, SD = .90). To create a dichotomized variable of punishment, the continuous variable was split at the numerical midpoint. This distinguished parents who had punished the first month from those who had punished later during the first year.

Child Adjustment

Four aspects of child adjustment were examined: externalizing, internalizing, attachment disturbances, and full-scale IQ. Externalizing and internalizing were selected because they are standard indicators of child adjustment. Attachment disturbances were included because co-sleeping and regressive parenting are specifically promoted as helpful for attachment in popular adoption manuals and blogs. Full scale IQ was included because it has been previously linked to parenting in this sample of children (Helder et al., 2014) and in larger studies of previously institutionalized international adoptees (e.g., Croft, et al., 2001). Inclusion of IQ also serves as a more objective (non-parent reported) assessment of children’s post-adoptive functioning. Prior to analyses, the Shapiro-Wilks test was applied to all adjustment scales to assure normal distributions. Outliers were also identified and recoded as necessary.

Externalizing and Internalizing

At each of the four waves, the adoptive parent completed the parent-report Behavioral Assessment Scale for Children – 2 (BASC-2, Reynolds & Kamphuis, 2004).  This instrument yields an Externalizing Problems Composite comprised of items from the conduct problems, aggression, and hyperactivity subscales of the BASC-2, and an Internalizing Problems Composite comprised of items from the anxiety, depression, and somatization subscales. Standardized scores on this index were calculated using age-based norms for the measure, resulting in T-scores (standardization sample M = 50, SD = 10). Higher T-scores indicate a greater degree of symptoms. When scores for the 38 children eligible for longitudinal analyses were examined, scores for externalizing met the criteria for a normal distribution and no outliers were identified. Externalizing scores ranged from 37 – 84 (M = 54.5, SD =12.6). Scores for internalizing also met the criteria for a normal distribution, but barely (p = .051). Because one (high) outlier was specified, this score was recoded to the value of the next highest score. After recoding, internalizing scores ranged from 32 – 63 (M = 47.3, SD = 8.4).

Disturbance of Attachment (DAI)

The presence of attachment disturbances from the parent’s perspective was assessed with the Disturbances of Attachment Interview (Smyke & Zeanah, 1999). This 13-item semi-structured interview addresses a range of attachment-related behaviors relevant to internationally adopted populations.  This includes questions addressing whether the child seeks comfort when distressed, is overly friendly with strangers, and displays hypervigilant attachment behaviors.  Each item is rated by the interviewer on a scale of 0 – 2 based on parent response to interview probes. Then, a total score is formed by summing (possible range 0 – 26). The interview was administered and scored by the second author, who had training in the use of the interview and in attachment formation in the context of early deprivation. Higher scores indicate a greater disturbance in attachment relationships. Internal reliability for each wave was acceptable, given the heterogeneity inherent in a measure that assesses both disinhibited social engagement and reactive attachment disorder (α = .64, .81, .68, .77). Preliminary data exploration indicated one outlier. After recoding the highest score to the next highest score, DAI scores ranged from 0 – 8 (M = 2.61, SD = 2.68). Unfortunately, scores on the DAI were still not normally distributed. Results involving the DAI must be viewed more tentatively than results involving the other adjustment variables.

Full-scale IQ

Intelligence was operationalized as the full-scale intelligence score on the Wechsler Intelligence Scale for Children-IV (WISC-IV, Wechsler, 2003) or the Wechsler Preschool and Primary Scale of Intelligence-III (WPPSI-III, 2002), depending on the age of the child.  Scores were standardized based on age-based norms (M=100, SD = 15).  Preliminary data exploration indicated that full-scale IQ scores were normally distributed even though one score was identified as an outlier. This (low) score was recoded to the next lowest score.  After recoding, full-scale IQ scores ranged from 65 – 122 (M = 95.5, SD = 14.3). 

Demographic measures

To supplement the demographic profiles derived from the dichotomous indicators of the controversial practices (Table 1), we sought demographic associations with the continuous indicators of the three practices. Significant correlations are reported in the last column of Table 2.


Results from MANCOVAs based on dichotomous variables are presented in Table 3. For all three MANCOVAs (co-sleeping, regressive parenting, punishment) and all four dependent variables (externalizing, internalizing, attachment disturbances, IQ), Levene’s test indicated equal variances across groups. Results from the regression analyses examining the continuous variables are presented in Table 4.


Contrary to our hypothesis that co-sleeping would be associated with adjustment problems, results from the MANCOVA based on the dichotomous measure indicated that children who co-slept the first month exhibited less internalizing at 5 years post-adoption than non-co-sleepers. Additionally, a nonsignificant trend (p = .08) suggested that co-sleepers had fewer attachment disturbances than their non-co-sleeping counterparts. The link between co-sleeping and internalizing was replicated in the regression analyses; more months of co-sleeping predicted less internalizing.

Regressive parenting

No differences between children whose parents scored above vs. below the theoretical midpoint on the regressive parenting scale were indicated with the MANCOVA. Some association between regressive parenting and attachment disturbances wassuggested (p = .06) in the regressions; more regressive parenting was associated with fewer attachment disturbances.

As with co-sleeping, these results ran counter to our hypothesis. This was surprising because the hypothesis was based on our own prior finding (from the same project) that non-regressive parenting predicted greater post-adoptive IQ gains across waves for “school age” children ages 4 – 15 at the time of adoption (Helder et al., 2014). To discern the reason for this discrepancy, we divided thecurrent sample into two age groups: the children who were also included in the school age paper (ages 4-11, n = 14) and the children too young for the school age paper (adopted prior to age 4, n = 15). We also identified the central item in the scale (the one associated with the lowest “alpha if item deleted” from the Cronbach’s alpha output). The item was Continuum 3: Made sure all nice things came only from parents vs. Permitted child to accept sweets, presents, hugs from relatives, friends and teachers.

We then ran regressions for attachment disturbances and for IQ by age group, first using the scale score, and second using the nice things item. For the younger children, neither the scale score nor the nice things item predicted adjustment at the five-year anniversary. For the older children, both the regression scale score (β = -.71, p = .04) and the nice things item (β = -.51, p = .01) were associated with attachment disturbances in a desirable direction (more regression with less disturbance). However, a nonsignificant trend for nice things and IQ was suggested in an undesirable direction (i.e., more regression was associated with lower IQ) (β = .64, p = .08). The latter assured us that our prior analyses including older children (through age 15) were sound.


In the planned MANCOVA distinguishing children who had been punished the first month vs. not, we obtained a nonsignificant trend (p = .07) suggesting that earlier punishment was associated with less internalizing at the five-year anniversary. This trend was replicated in the planned regressions (p = .10). Additionally, we obtained a significant link between earlier punishment and IQ in the regressions, indicating that children who had experienced earlier punishment had higher full-scale IQ scores at the five-year mark.

We then repeated the analyses, analyzing the two categories of punishment separately. Although our primary goal was to discern the relationship between child adjustment and onset of punishment broadly defined, the nonsignificant correlation between the two onsets obtained in our preliminary analyses suggested that they may be conceptually distinct. Additionally, researchers conducting meta-analyses on punishment often require separate effect sizes.

Separating the two types of punishment prompted a change in the dichotomization strategy for the onset of physical punishment. For analyses of nonphysical punishment and the two types combined, the most comparable cell sizes are created by contrasting children punished the first month with those punished in 2-12 month range. (Recall that all parents reported nonphysical punishment the first year). In contrast, the most sensible dichotomization for physical punishment contrasted children who had experienced physical punishment the first year with those whose parents reported “have not done this” (n = 5). (Regardless of how we dichotomized, one cell was going to be small, and “never-spanked” children constitute a theoretically meaningful category in the broader spanking debate (Gunnoe, 2013).)

Distinguishing the two types of punishment clarified and strengthened associations obtained with the combined punishment variable.  Earlier nonphysical punishment was associated with lower internalizing in regression analyses (β = -.51, p = .03, R2 = .31, p = .05) but not MANCOVA. Onset of nonphysical punishment was unrelated to IQ. In contrast, earlier onset of physical punishment was associated with higher IQ in regression analyses (β = .60, p = .03, R2 = .27, p = .08), and use of physical (vs. “have not used”) was associated with higher IQ in MANCOVA (Multivariate F = 4.22, p = .02, Univariate F = 15.44, p = .001). Onset of physical punishment was unrelated to internalizing.

Associations between controversial practices

We also sought associations between the controversial practices. We paired dichotomous variables with continuous variables in round-robin rotation, and continuous variables with continuous variables. Sample sizes were small because we applied the age-criteria and time-in-home criteria from both practices (which only partially overlapped for most pairings). We found no suggestion of association; use of one practice did not predict use of another practice.


We documented associations between parents’ use of three controversial parenting practices and four manifestations of child adjustment at the child’s five-year anniversary in the adoptive home. All of the associations we obtained were “good” — in the sense that each practice predicted either fewer problems (internalizing or attachment disturbances) or higher full-scale IQ.

We preface our discussion of these associations by reiterating the caution we stressed in the opening paragraphs of our report: We were unable to control for children’s adjustment at the time they were receiving/eliciting these practices. The obtained associations cannot be viewed as causal. The most we can claim is that each of the practices was part of a total parenting package that predicted better adjustment on at least one commonly assessed manifestation of child adjustment, at the five-year anniversary of the adoption. Additionally, our hypotheses were tentative and general. They were tentative because there is almost no academic literature on these practices. They were general in that we hypothesized “poorer” or “better” adjustment, without explicating why a practice might be linked to specific manifestations of child adjustment. We had so little to go on that we did not want to take the reader down speculative dead-ends. We simply reported that some authors of popular adoption manuals promote co-sleeping and regressive parenting as beneficial for attachment, and some authors disparage punishment as limited in utility and/or counter-productive. Our task now, is to offer plausible explanations for why the associations we obtained in these exploratory analyses might be causal, or at least why they were associated with child adjustment at the five-year mark.

Possible reasons for the obtained associations


Co-sleeping the first month predicted lower internalizing at five years post-adoption. Total months of co-sleeping also predicted lower internalizing, and potentially fewer attachment disturbances. These findings run counter to our hypothesis. These findings are in accord with two of the must-read adoption manuals, various blogs, and the popularized “Attachment parenting” initiative presented in the Introduction.

Initial co-sleeping was likely associated with positive outcomes for the reasons cited by its promoters. In the case of international adoption, the very scary transition from a familiar environment to an unfamiliar one is probably eased by nighttime proximity to the only constant in the extended adoption/travel process. One specific mechanism influencing both internalizing and attachment disturbances may be cortisol regulation. Beijers et al. (2013) exposed infants to the Strange Situation at 12 months and found better cortisol regulation by those who had co-slept the first six months of life. Several research teams have linked cortisol regulation to internalizing in at-risk children (e.g., van der Vegt et al., 2010, Badanes et al., 2011). Marceau et al., (2015) found direct and indirect links (via cortisol) between parenting and internalizing in a large sample of 6-year-old adoptees.      

In reckoning the value of co-sleeping for the children in our sample, it is important to remember that 72% of our co-sleeping children ceased co-sleeping by the end of the third month. In other words, co-sleeping was a transitional, rather than persistent, arrangement. This is likely the reason that co-sleeping was associated with desirable, rather than undesirable, manifestations of child adjustment. (Our unsupported hypothesis was based on studies of persistent co-sleeping, which many researchers believe to be child-driven, and most of our families were not persistent.) Per these results, we support co-sleeping (broadly defined) as a transitional practice for recent adoptees ages 1-7 (and room-sharing with infants).  

Regressive parenting

We obtained less evidence to support intentionally regressive parenting, although this approach was associated with fewer attachment disturbances (as a trend) in the planned regressions. This specific link was in keeping with the claims of those who advocate regressive parenting.

This pattern (one “good” trend, but mostly lack of association) ran contrary to the hypothesis derived from our prior work, prompting additional analysis to reconcile the results of this study with the results for “school age” children presented by Helder et al. (2014). Our post-hoc synthesis suggested that intentionally regressive parenting the first six months was unrelated to the adjustment of children adopted prior to age 4 and both beneficial and harmful for children adopted after age 4. In the current study, regressive parenting predicted fewer attachment disturbances and a lower IQ (trend) at the five year anniversary for children ages 4-11 at the time of adoption. In our 2014 study, regressive parenting predicted less cognitive gain over a two year period for children ages 4-15. (Controlling for IQ at Wave 1 in predicting IQ at Wave 3 afforded a greater degree of causal inference than the current analyses permit). Because regressive parenting was associated with fewer attachment disturbances in our older children but seems to have inhibited their cognitive development, we are reluctant to either encourage or discourage intentionally regressive parenting.

Going forward, we intend to reframe. When we composed the items for this scale many years ago, we were responding to Hopkins-Best’s (1997) assertions that “Recreating attachment is an intrusive process”(p. 196) and that a toddler must not be permitted to hold his own bottle (p. 197). We questioned this stance because it struck us as unnecessarily demeaning (and maybe unfeasible) to try and push a child back developmentally. Subsequently, Purvis et al. (2007, p. 166) suggested that parents and children hand-feed each other. We like this suggestion because it accomplishes some of the same things that uni-directional hand-feeding promotes without creating a power-struggle over an activity that is age-appropriate for the child (i.e., feeding oneself). Reframing regressive techniques as intimacy techniques may help parents balance the various catch-up processes in which their children need to engage.

The most influential item in our regressive parenting scale involved limiting who the child could receive nice things from. Again, we like Purvis et al.’s suggestion that parents develop an attachment ritual for a caregiver-handoff wherein a parent reiterates their primary status and explicitly states that the alternate caregiver will be “your boss” and give you hugs in the place of me, while I am away (p. 190). Adopted children of all ages need intimacy with primary caregivers(s) but this should not come at the expense of the “village” that helps school-age children develop their cognitive potential.

Earlier punishment

Earlier onset of punishment was associated with better child adjustment at the five-year anniversary. Better adjustment was hypothesized based on prior research and is in keeping with two of the four must-read adoption manuals. Purvis et al. endorse nonphysical disciplinary consequences but are short on specifics. Keck & Kupecky (2002) encourage parents to try many different types of punishment, warning that specific techniques may not work with specific children.

Specifically, earlier onset of punishment was associated with higher IQ. This association complements the negative association between regressive parenting and IQ for school-age children. (Recall: Continuum 5 in the regressive parenting scale contrasts “going easy” vs. confronting children’s misbehavior head on.) Additionally, a relationship between earlier punishment and lower internalizing was suggested (trend). Because a desirable link between punishment and child adjustment may be disconcerting to some readers, we devote more discussion to this controversial practice than the other two, readily acknowledging that the associations could be practice driven and/or child driven.

If the association between punishment and desirable child outcomes is practice driven, the most likely explanation is that children low in self-regulation need their parents to provide external regulatory structure. Often, the need for external regulation is presented as a function of age.  Baumrind (1996, p. 408) wrote that for toddlers and preschoolers, “compliance is most effective when the adult briefly explains the rule and provides a consequence if the child persists in disobeying.” These external consequences help children develop habits that permit parents to relax control during adolescence. However, adopted children’s development may be more a function of deprivation than chronological age. In the same way that children who have experienced early deprivation benefit from concerted intimacy experiences to catch up in the emotional domain, they may benefit from concerted regulatory experiences to catch up in the behavioral domain.

Studies assessing parents’ regulatory attempts in laboratory play sessions with toddlers and preschoolers have suggested that these attempts are beneficial for post-institutionalized children’s adjustment. Applying a cross-lag panel design, Lawler et al. (2017) found that parental structure and limit-setting predicted fewer child regulation difficulties for internationally adopted children but not for a control group. Similarly, Pitula et al. (2019) reported that a combined measure of responsiveness and limit-setting predicted better peer relationships in kindergarten, but only for previously-institutionalized adopted children. Limit-setting combined with responsiveness is the definition of authoritative parenting. Tan et al. (2012, 2015) reported that authoritative parenting predicted less externalizing and internalizing in adopted Chinese girls, while permissive parenting and authoritarian parenting predicted more.

As stated prior, children who have experienced severe deprivation may even benefit from atypically-strict control. Audet and Le Mare examined parents’ use of authoritarian parenting with three groups: non-adopted Canadian children, post-institutionalized Romanian children adopted before 4 months of age, and post-institutionalized Romanian children adopted after 8 months. In contrast to the first two groups (for whom authoritarian parenting “worked” as anticipated), higher levels of authoritarian parenting with the more extensively deprived group predicted lower inattention/overactivity at age 10.5 (after controlling for I/O at age 4.5). The authors explain the association like this:

Children with extensive deprivation experience are at considerable risk of I/O difficulties. In behaving in an authoritarian way, parents of children with lengthy institutional experience (and who are high in I/O) may actually be displaying sensitivity to their child’s needs. That is, such children may need an environment that is controlling, structured, and strict (which to a certain extent, characterizes authoritarian parenting). These parenting practices likely provide the external supports that enable high I/O children to focus their attention and regulate their activity (Audet & Mare, 2011, p. 113).

Interestingly, we did not obtain an association between punishment and externalizing, the aspect of child adjustment we viewed as the best candidate for external regulatory support. Our inability to predict externalizing is becoming a pattern (c.f., Helder et al., 2014, 2020) and may have something to do with the ages of our children. In a study of domestically adopted Korean children, internalizing was stable over time, whereas externalizing increased from ages 5-7 and then decreased significantly thereafter (Ahn, et al., 2017). Unfortunately, we do not have enough children older than 7 to analyze them separately or test for a curvilinear relationship. 

Although external regulatory support in the form of punishment did not predict externalizing, it may be the mechanism linking punishment to cognitive gains. Prior to Audet and Le Mare’s (2011) proposal that external control may focus attention, Baumrind (1996) asserted that externally supported focus facilitates increased proficiency and self-efficacy that eventually becomes “increased interest in sustained effort” (1996, p. 409). Again, externally supported focus may be particularly helpful for children who struggle with inattention/overactivity, as about of a third of our children do (see Helder et al., 2016). Evidence that externally supported focus can be part of a parenting package associated with cognitive competence in more representative families comes from the finding that teenagers who recalled age-delimited spanking (not past age 12) reported higher academic achievement than teens who reported that they had never been spanked, after accounting for fathers’ parenting styles (Gunnoe, 2013).

Alternatively, the association between punishment and IQ could be child-driven. Perhaps adoptive parents are less likely to punish children with lower IQ scores. This seems plausible, but this pattern would be contrary to Croft et al.’s (2001) finding that children’s cognitive ability at the time of adoption was a strong predictor of the emotional valence of the parent-child relationship, and low cognitive ability predicted more negative interactions, not less.

A practice-driven relationship between punishment and children’s growing self-regulation may also explain the suggested association between punishment and internalizing (p < .10 in the analysis with the combined variable, significant in the analyses focused specifically on nonphysical punishment). In contrast to the four must-reads which warn that time-out may further traumatize children, Dadds and Tully (2019) assert that traumatized children need discipline that is calm, controlled, brief, and predictable to practice emotional and behavioral regulation. According to Eldridge (the adult adoptee who wrote the 5th book on the must-read list), children also need the message that they possess the capacity for self-control. Eldridge recounted being home alone and using her mother’s broach to carve “I love you Mommy” into her parents’ dresser.  Upon return, her mother’s only response was to choke out “We love you too, honey.” Eldridge’s adult assessment of the situation is that her mother meant well, but she “denied the reality of the destruction I’d caused and failed to follow up with age-appropriate consequences” (1999, p. 74). Included in Eldridge’s list of what adoptive parents need to know is this: “If I don’t learn that I am accountable for my behavior, I will cling to a victim mind-set and act it out forever” (p. 198).

The experience of non-abusive discipline may also facilitate children’s sense of security in parents and the broader social environment (Quetsch, et al., 2017). When parents discipline in a safe way they communicate: competent management, clear behavioral standards, and a predictable world where violations of standards are met with consequences. In the Keck and Kupecky model, hurting children are terrified of relinquishing control but actually want their parents to be in charge (1995, p. 188). Safe discipline may also communicate parental investment. Per Eldridge, lack of discipline produces a child who feels like an illegitimate son or daughter, not truly belonging (p. 194).

If Eldridge is correct that punishment helps children feel legitimate, it may be particularly important in families where children have siblings who are the biological children of the adoptive parents (71% of our children). Early sibling relationships impact adopted children’s adjustment later (Tan, 2008) and many parents understand this. One adoptive parent who participated in the qualitative component of Stillman’s (2015) dissertation reported that he and his spouse strove to discipline “with equality” because they “wanted their adopted children to feel as much a part of their home as their nonadopted children” (p. 89). Another parent reported frustration on the part of his non-adopted children when discipline standards were perceived as different.

As with punishment and IQ, the association between punishment and internalizing could be child-driven. Perhaps parents were slower to employ power-assertion with children who were exhibiting higher internalizing at the time of the adoption (and were still exhibiting higher internalizing five years later, despite a later onset of punishment). This would be in keeping with the finding that parents use less corporal punishment with children high in internalizing symptomatology (Grogan-Kaylor & Otis, 2007).

Although our punishment results confirmed our hypothesis, we acknowledge that there are questions generated by our results that we cannot answer. First and foremost: Why did the two types of punishment predict different desirable outcomes? Being unsure of the mechanisms whereby the two types were differentially associated, we limit our endorsement of earlier punishment to punishment in general, rather than a specific category of punishment. And we do this with a qualifier: Parents should punish “early” only if it seems appropriate for their specific child. As stressed in the Introduction, controversial practices can only be deemed appropriate (or not) with respect to a specific child. Co-sleeping or not co-sleeping, regressing or not, punishing earlier or punishing later, may all indicate that our (well educated, agency-vetted, mostly experienced) parents were demonstrating sensitivity to the needs of their particular child. Finally, we reiterate that our measures of punishment assessed onset, not frequency. Our data can be used to support earlier punishment but cannot inform a discussion about how much punishment is appropriate with at-risk children.

Correspondence between the four must-read adoption manuals and our findings

Because we used the four must-read adoption manuals to scaffold this investigation of controversial practices, it is fitting that we now assess these books on the basis of our findings (and some personal inclinations). All the books recommend high parental responsiveness. The primary difference is the degree to which they promote overt parental control over the attachment process and children’s misbehavior.

Keck and Kupecky promote co-sleeping, regressive parenting, and punishment. Their books permit us to check the most boxes, but we are uncomfortable with some of their specific techniques (intentionally keeping children disoriented (2002, p. 68), dishonest feigning of control, and “holding”). Purvis et al. do not address co-sleeping but offer the most authoritative model, encouraging parents to nurture honestly and exert authority promptly. Hopkins-Best promotes co-sleeping but her approach strikes us as over-controlling with respect to attachment and under-controlling with respect to misbehavior. Forbes and Post make under-control their thesis. Thus, the books by Purvis et al. and by Keck and Kupecky are the most compatible with our data. The book by Forbes and Post is the least compatible.

Empirical affinity with Purvis et al.’s manual makes sense. Of the four must-reads, theirs is the most closely connected with a program of academic research. Purvis’ Trust-Based Relational Intervention (TBRI) based on “empowering, connecting and correcting” has been linked to documented positive child outcomes in post-adoptive outpatient therapy (Howard, et al., 2014) and school-based residential treatment (Parris, et al., 2015). That said, we recognize that one book does not fit all, and the other books also have elements to recommend them. We appreciate Keck and Kupecky’s poignant depictions of life in the trenches. These authors have a wealth of experience with the most troubled children that will resonate with many struggling parents. We appreciate Hopkins-Best’s intermittent attention to a sibling. We appreciate Forbes and Post’s attention to the physiological stress process. We think Forbes and Post’s posture may be somewhat more appropriate for older children than younger ones. Although we could not find a target age-range, their book reads as though it is for parents of teens.

Strengths and Limitations

The primary strength of the present report is our applied emphasis. Given the plentitude of controversy over certain practices and the dearth of empirical address within an adoption context, our work will be of interest to adoption professionals who, like us, have been asked: Can I, should I, must I engage in these practices? Other strengths include our longitudinal design, our spectrum of adjustment variables including researcher-tested IQ, our complete (non-missing) data for all variables included in the longitudinal analyses, and the heterogeneity of birth countries represented.

Limitations include our inability to control for baseline child adjustment and the exploratory nature of our analyses. We were also limited by the number of children who were both age-eligible and time-in-home eligible for two of the three controversial practices. Our attachment measure was helpful for identifying clinical disturbances of attachment, but the capacity to distinguish secure vs. insecure attachment might have resulted in stronger associations between attachment and the controversial practices. The generalizability of our findings may be impacted by our convenience sample, heterogeneity of child age, and homogeneity of several family demographics including parents’ marital status (married/heterosexual), adoption motivation (religious/ humanitarian), attendance at religious services (high), and family size (multiple siblings).


Desirable manifestations of child adjustment at the five-year anniversary of an international adoption were predicted by initial co-sleeping with children under the age of 7 and earlier onset of punishment for children ages 2-11. A synthesis of the current findings with our 2014 findings suggests that intentionally regressive parenting may be inconsequential for children adopted prior to age 4 and both beneficial (fewer attachment disturbances) and detrimental (lower IQ) for children ages 4-11. Associations may not be causal and may not generalize to families with a different demographic profile. Future research should attempt to discern whether the associations between these controversial practices and desirable child outcomes is still present after accounting for baseline child adjustment and if so, to identify the mechanisms contributing to these associations.  


Ahn, J., Byun, M.-h, & Kwon, J.-s. (2017). Trajectory of problem behaviours of Korean adopted children: using piecewise hierarchical linear growth modelling. Child and Family Social Work, 22, 461-471.

American Academy of Pediatrics (2022). Sleep-related infant deaths: Updated 2022 recommendations for reducing infant deaths in the sleep environment. Pediatrics, 150(1).

Audet, K. & Le Mare, L. (2011). Mitigating effects of the adoptive caregiving environment on inattention/overactivity in children adopted from Romanian orphanages. International Journal of Behavioral Development, 35(2), 107-115.

Badanes, L.S., Watamura, S.E., & Hankin, B.L. (2011). Hypocortisolism as a potential marker of allostatic load in children: Associations with family risk and internalizing disorders. Developmental Psychology, 23(3), 881-896.

Baumrind, D. (1996). The discipline controversy revisited. Family Relations: An Interdisciplinary Journal of Applied Family Studies, 45(4), 405-414.

Beijers, R., Riksen-Walraven, M., & de Weerth, C. (2013). Cortisol regulation in 12-month-old human infants: Associations with the infants’ early history of breastfeeding and co-sleeping. Stress, 16, 267-277.

Chaffin, M., Hanson, R., Saunders, B.E., Nichols, T., Barnett, D., Zeanah, C., Berliner, L., Egeland, B., Newman, E., Lyon, T., LeTourneau, E., & Miller-Perrin, C. (2006). Report of the APSAC task force on attachment therapy, reactive attachment disorder, and attachment problems. Child Maltreatment, 11(1), 76-89.

Chen, Z., Liu, X., & Liu, J. (2018). Early childhood co-sleeping is associated with preadolescent behavioral problems: A cohort study.  Sleep, 41, A305.

Cline, F.W. (1992). Hope for high risk and rage filled children. Evergreen, CO: EC Publications.

Cline, F. & Fay, J. (1990). Parenting with love and logic: Teaching children responsibility. Colorado Springs, CO: Piñon Press.

Croft, C., O’Connor, T.G., Keaveney, L., Groothues, C., Rutter, M., and the English and Romanian Adoption Study Team. (2001). Longitudinal change in parenting associated with developmental delay and catch-up. Journal of Child Psychiatry and Psychology, 42(5), 649-659.

Dadds, M.R., & Tully, L.A. (2019). What is it to discipline a child: What should it be? A reanalysis of time-out from the perspective of child mental health, attachment, and trauma. American Psychologist, 74(7), 1-15.

Durrant, J.E. (2016). Positive discipline in everyday parenting, 4th Edition.Stockholm: Save the Children.

Eldridge, S. (1999). Twenty things adopted kids wish their adoptive parents knew. Dell.

Forbes, H.T. & Post, B.B. (2010). Beyond consequences, logic, and control: A love based approach to helping children with severe behaviors. Boulder, CO: Beyond Consequences Institute.

Galen, J. (2017, February, 12) Should I co-sleep with my adopted child?: The science may surprise you!

Garvin, M.C., Tarullo, A.R., Van Ryzin, M., & Gunnar, M.R. (2012).  Postadoption parenting and socioemotional development in postinstitutionalized children. Development and Psychopathology, 24, 35-48.

Gershoff, E.T. (2002). Corporal punishment by parents and associated child behaviors and experiences: A meta-analytic and theoretical review. Psychological Bulletin, 128, 539-579.                                  

Gershoff, E.T., & Grogan-Kaylor, A. (2016). Spanking and child outcomes: Old controversies and new meta-analyses. Journal of Family Psychology, 30(4), 453-469.

Gershoff, E.T., Goodman, G.S., Miller-Perrin, C.L., Holden, G.W., Jackson, Y., & Kazdin, A.E. (2018). The strength of the causal evidence against physical punishment of children and its implications for parents, psychologists, and policymakers. American Psychological Association, 73(5), 626-638.

Grogan-Kaylor, A., & Otis, M.D. (2007). The predictors of parental use of corporal punishment. Family Relations, 56(1), 80.

Gorter, B.A., Helder, E.J., Oh, Y.I., & Gunnoe, M.L. (2017). Are disinhibited social behaviors among internationally adopted children mediated by attachment environment or by children’s difficulties with inhibitory control? Adoption Quarterly, 20(4), 291-308.

Gunnoe, M.L. (2013).Associations between parenting style, physical discipline, and adjustment in adolescent reports. Psychological Reports, 112(3), 933-975.

Helder, E.J., Brooker, B., Kapitula, L.R., Goalen, B. & Gunnoe, M.L. (2016). Predictors and correlates of inattentive/overactive behaviors in internationally adopted children, Applied Neuropsychology: Child, 5(4), pp. 237-251.

Helder, E.J., Mulder, E., & Gunnoe, M.L. (2016). A longitudinal investigation of children adopted internationally at school age. Child Neuropsychology, 22 (1), 39-64.

Helder, E., Gunnoe, M.L., & Timmermans, H. (2020). Religious motivation to adopt as a predictor of adoptive family structure, parental discipline, and outcomes. Adoption Quarterly.

Holden, G.W., Grogan-Kaylor, A., Durrant, J.E., & Gershoff, E.T. (2017). Researchers deserve a better critique: Response to Larzelere, Gunnoe, Roberts, and Ferguson (2017). Marriage & Family Review, 53, 465-490.

Hopkins-Best, M. (1997). Toddler adoption: The weaver’s craft. Perspectives Press.

Hopkins-Best, M. (2012). Toddler adoption: The weaver’s craft. (Revised edition). Jessica Kingsley Publishers.

Howard, A.R., Parris, S.R., Nielsen, L.E., Lusk, R., Bush, K, Purvis, K.B., & Cross, D.R. (2014). Trust-based relational intervention (TBRI) for adopted children receiving therapy in an outpatient setting. Child Welfare, 93(5), 47-64.

Juffer, F., Bakermans-Kranenburg, M.J., & van IJzendoorn, M.H. (2005). The importance of parenting in the development of disorganized attachment: Evidence from a preventive intervention study in adoptive families. Journal of Child Psychology and Psychiatry, 46(3), 263-274.

Jensen, R.J., Helder, E.M., & Gunnoe, M.L. (2016). Attachment disturbances delay language acquisition in internationally adopted children. Adoption Quarterly, 19 (3), 210 – 223.

Keck, G.C. & Kupecky, R.M. (1995).  Adopting the hurt child: Hope for families with special-needs kids. Colorado Springs, CO: Piñon Press. (Revised 2009; Colorado Springs, CO: NavPress)

Keck, G.C. & Kupecky, R.M. (2002). Parenting the hurt child: Helping adoptive families heal and grow. Colorado Springs, CO: Piñon Press. (Revised 2009; Colorado Springs, CO: NavPress)

Larzelere, R.E., Gunnoe, M.L., & Ferguson, C.J. (2018). Improving causal inferences in meta-analyses of longitudinal studies: Spanking as an illustration. Child Development. 89(6), 2038-2050.

Larzelere, R.E., Gunnoe, M.L., Roberts, M.W., Lin, H., & Ferguson, C.J. (2020). Causal evidence for exclusively positive parenting and for timeout: Rejoinder to Holden, Grogan-Kaylor, Durrant, and Gershoff (2017). Marriage & Family Review, 56, 287-319.

Larzelere R.E., & Kuhn B.R. (2005). Comparing child outcomes of physical punishment and alternative disciplinary tactics: a meta-analysis. Clinical Child and Family Psychology Review, 8, 1-37.

Lawler, J.M., Koss, K.J., & Gunnar, M.R. (2017). Bidirectional effects of parenting and child behavior in internationally adopting families. Journal of Family Psychology, 31(5), 563-573.

Marceau, K., Heidemarie, K.L, Neiderhiser, J.M., Reiss, D., Shaw, D.S., Natsuaki, M.N., Fisher, P.A., & Leve, L.D. (2015). Combined influences of genes, prenatal environment, cortisol, and parenting on the development of children’s internalizing versus externalizing problems. Behavior Genetics, 45, 268-282.

Mileva-Seitz, V.R, Bakermans-Kranenburg, M.J., Battaini, C. & Luijk, M.P.C.M. (2017). Parent-child bed-sharing: The good, the bad, and the burden of evidence. Sleep Medicine Reviews, 32, 4-27.

Miller, P.M., & Commons, M.L. (2010). The benefits of attachment parenting for infants and children: A behavioral developmental view. Behavioral Development Bulletin, 16(1), 1-14.

National Association of Social Workers. (2003). Social work speaks: National Association of Social Workers policy statements, 2003-2006. NASW Press.

New Beginnings (2017, August 29). Top 5 books you must read when adopting from international adoption or foster care.

Parris, S.R., Dozier, M., Purvis, K.B., Whitney, C., Grisham, A., Cross, D.R. (2015). Implementing Trust-based relational intervention in a charter school at a residential facility for at-risk youth. Contemporary School Psychology, 19(3), 157-164.

Pearl, E., Thieken, L., Olafson, E., Boat, B., Connelly, L., Barnes, J., & Putnam, F. (2012). Effectiveness of community dissemination of parent-child interaction therapy. Psychological Trauma: Theory, Research, Practice, and Policy, 4(2), 204-213.

Phillips, L. (2021, August, 10). To have and to hold. Adoptive Families: The How-to-Adopt and Adoption Parenting Network.

Pitula, C.E., DePasquale, C.E., Mliner, S.B., & Gunnar, M.R. (2019). Peer problems among post institutionalized, internationally adopted children: Relations to hypercortisolism, parenting quality and ADHD symptoms. Child Development, 90(3), 339-355.

Purvis, K.B., Cross, D.R., & Sunshine, W.L. (2007). The connected child. Bring hope and healing to your adoptive family. McGraw Hill.

Quetsch, L.B., Lieneman, C., & McNeil, C.B. (2017, May). The role of time-out in trauma-informed treatment for young children. Society for Psychotherapy.

Reynolds, C.R. & Kamphaus, R.W.  (2004).  Behavior Assessment System for Children (2e). Pearson.

Robbins, J. (2016, December 1). Co-sleeping and adoption, why we did it.

Rueter, M.A., Keyes, M.A., Iacono, W.G., & McGue, M. (2009). Family interactions in adoptive compared to nonadoptive families. Journal of Family Psychology, 23(1), 58-66.

Santos-Nunes, M., Narciso, I., Vieiro-Santos, S., Roberto, M.S. (2017). Parenting and emotional well-being of adoptive school-aged children: The mediating role of attachment. Children and Youth Services Review, 81, 390-399.

Stams, G.-J., J.M, Juffer, F., Rispens, J., & Hoksbergen, R.A.C. (2000). The development and adjustment of 7-year-old children adopted in infancy. Journal of Child Psychology and Psychiatry, 41(8), 1025-1037.

Stams, G.-J., J.M, Juffer, F., & van IJzendoorn, M.H. (2002). Maternal sensitivity, infant attachment, and temperament in early childhood predict adjustment in middle childhood: The case of of adopted children and their biologically unrelated parents. Developmental Psychology, 38(5), 806-821.

Stillman, S. A. (2015). Parental practices of discipline with re-adopted children: A mixed methods study. The Southern Baptist Theological Seminary, ProQuest Dissertations Publishing.

Smyke, A.T. & Zeanah, C.H.  (1999).  Disturbances of attachment interview.  Unpublished manuscript.

Tan, T.X. (2008). Impact of biological children’s adjustment on their siblings who were adopted from China. Adoption Quarterly, 11(4), 278-295.

Tan, T.X., Camras, L.A., Deng, H., Zhang, M., Lu, Z. (2012). Family stress, parenting styles, and behavioral adjustment in preschool-aged adopted Chinese girls. Early Childhoood Research Quarterly, 27, 128-136. 

Tan, T.X., Gelley, C.D., & Dedrick, R.F. (2015). Non-child-related family stress, parenting styles, and behavior problems in school-aged girls adopted from China. Journal of Child and Family Studies, 24(10), 2881-2891.

Tax, T.X., Marfo, K., & Dedrick, R.F. (2009). Preschool-age adopted Chinese children’s sleep problems and family sleep arrangements. Infant and Child Development, 18(5), 422-440.

van der Vegt, E.J.M, van der Ende, J., Verhulst, A.C., & Tiemeier, H. (2010). Childhood adversity modifies the relationship between anxiety disorders and cortisol secretion. Biological Psychiatry, 68(11), 1048-1054.

Wechsler, D. (2003). Wechsler intelligence scale for children, 4th edition (WISC-IV). Pearson.

Weir, K. (2011). Playing for Keeps: Integrating family and play therapy to treat reactive attachment disorder. In A.A. Drewes, S.C. Bratton & C.E. Schaefer (Eds.) Integrative Play Therapy (pp. 343-264). Wiley.

[1] In her revised edition, Hopkins-Best omits discussion of holding, substitutes the word “assertive” for “intrusive,” and drops the directive for parents to stay “in their faces” (2012, p. 207).

[2] Readers attempting to reconcile the results of this study with the results reported for “Parenting approach” in Helder et al. (2014) should note that this scale was renamed Regressive parenting and reverse coded for the present study.

Leave a Reply