Inclusive and Supportive Education Congress International Special Education Conference Inclusion: Celebrating Diversity? 1st - 4th August 2005. Glasgow, Scotland

home about the conference programme registration accommodation contact

Feasibility and Validity of a Parent-Child Group Model
Of Early Intervention

Presenters: Jeanette McCollum and Tweety Yates
jmccollu@uiuc.edu

"PIWI Replicability," a project funded by the Office of Special Education Programs, was designed to evaluate the replicability of a parent-child group approach to early intervention in two types of settings in which infants and toddlers and their families receive publicly-funded services in the U.S.: Early Intervention (EI, under Part C of the Individuals with Disabilities Education Act) and Early Head Start (EHS). Replicability was defined as including two related perspectives: "validity" (whether staff and family perceptions of various aspects of the model were congruent with PIWI's theoretical foundations) and "feasibility" (whether programs were able to implement the model, given the specific characteristics of their settings). Of the 17 programs that initially began the replication process, 13 (7 EI and 6 EHS) completed it. Of the four sites that did not complete the process, three experienced turnover in the administrator who had made the initial commitment to participate, while the remaining site decided not to continue because it already had a parent-child model that it felt adequately met its needs. A full report of results from the 13 programs is contained in the final project report (Final Report, 2003). This presentation provides a case analysis of four sites, all of which were EI sites serving children with disabilities and their families under Part C: the two with most success in implementing the model and the two with least success, determined on the basis of observed fidelity to the model. Our specific research goal was to explore factors that might contribute to a program's ability to replicate the model.  

Early childhood settings in the U.S. present unique challenges to the change process because of factors such as high staff turnover, uncertainty of funding, and lack of local control due to frequent changes in higher-level policy and procedures (Winton & McCollum, 1997). Literature on systemic change underscores the importance of viewing the change process as one part of a larger ecological system, understandable only when other systems that influence the success of the change effort are taken into account (Squire & Reigeluth, 2000). Characteristics of individuals that may influence change at the individual level include education and experience in the field. Participant beliefs may have special importance in implementing the PIWI model because parent-child relationships are the primary foundation of the model, yet are strongly grounded in each individual's views of appropriate parenting (Greenfield & Suzuki, 1998).

Characteristics of the training process that may influence the ability to adopt and sustain innovation have been described in the personnel training literature. As summarized by Wolfe and Snyder (1997), factors that influence whether learners are able to transfer new skills and knowledge to the work place include the characteristics and experiences of the program participants as well as the design of the training. The organizational context may also be influential in whether what is learned in training can be applied in the work place. For example, these factors may include time, administrative support, and organizational incentives for change.

The current case analysis was undertaken to identify and describe factors that may contribute to a program's ability to provide parent-child groups based on the PIWI model. A maximum variation approach was used to highlight these factors. Our research questions for the current analysis were:

1. What factors (personal, professional, administrative, contextual) appear to influence whether early intervention programs delivering services under Part C are able to implement and sustain parent-child groups based on the PIWI model?

2. Does level of parent-child engagement differ in low vs. high fidelity implementers of the model?

Description of the PIWI Model

PIWI (Parents Interacting with Infants) is a model of service delivery that combines developmental and family-centered perspectives into a relationship-based model in which parent-child interaction is both a focus and a context for understanding children's development and for intervention. The framework for what happens in PIWI is a 4-part philosophy, with each section of the philosophy further explicated through a set of specific implementation guidelines. These guidelines are used as the central organizer for making decisions about practice as parent-child groups (or home visits) are developed, implemented, and evaluated within the context of a specific early childhood program (McCollum, Gooler, et al, 2001). They also form the basis for the Fidelity of Implementation Checklist that is used to monitor program implementation and fidelity to the model.  

PIWI practices are based on the assumption that caregivers' sensitivity and responsiveness to their child's developing dispositions and behavior are dependent on their ability to observe and interpret their child's development from the child's perspective. Each PIWI group session is organized around a developmental observation topic (“DOT”) that supports parents as observers and interpreters within the context of the session and then as carried over to home (McCollum, Yates, et al., 2001). Group discussions with parents at the beginning and end of each session are organized around the developmental topic. To support this discussion, the parent-child interaction portion of the schedule is planned so that particular, observable aspects of development related to the topic will be highlighted, increasing the likelihood that parents will be able to observe their child in relation to the topic. Roles of the PIWI facilitators (service providers) are carefully defined, with a focus on organizing and leading group discussions and supporting parent-child interaction in a way that involves parents and responds to parents' concerns, interests, and ongoing experiences during the group.   Also important to implementing the model is the availability of sufficient team members to interact with all families on a frequent basis during the parent-child interaction portion of the group session (McCollum & Yates, 1994). Individualized goals and strategies can be embedded within the flow of the group.

The training process used in this project was based on principles derived from the personnel preparation literature (McCollum & Catlett, 1997; Wolfe & Snyder, 1997). Administrators were asked to commit to an extended (10-12 months) interaction with project staff. They also agreed to send a team including themselves and at least two future group facilitators to a 2-day training at the PIWI project site, and to accommodate and support a minimum of two 1-day training visits by project staff to their program once the PIWI groups had been implemented. Initial training at the PIWI site was offered to 1-2 programs at a time, with the first day focusing on the PIWI model and components and the second day on observation of a PIWI parent-child group and the development of next-step action plans. Self-assessment, goal setting and problem solving were included in the initial training and in follow-up site visits.   Staff from each program developed a team plan for initiating PIWI groups. After approximately 2-3 group sessions had been held at the site, PIWI staff conducted a 2 nd training visit. Team members shared experiences from implementation of their action plans and presented successes and dilemmas. Additional training was also provided on the more difficult aspects of the model. Telephone consultation was available throughout the project, and each site was contacted monthly to discuss implementation progress and issues. Thus, every effort was made, within the constraints of the budget and of staff time, to ensure that the training and follow-up process was responsive to the needs of the participants and individualized to what they were experiencing in their own sites.

Selection of Case Study Programs

Selection of the four programs described in this case analysis was determined from scores obtained on the Fidelity of Implementation Checklist based on the PIWI implementation guidelines. Specifically, the four programs selected were the two with the highest and the two with the lowest scores. This checklist was completed by project staff subsequent to each training visit, using information from observations and from discussions with staff. Four categories of items were included (child, family, facilitator team, setting). For each of the 15 items on the checklist, observers used a 1-5 rating scale ("not observed" to "frequently observed"). Two staff members together conducted over a third of the site visits, and reliability on the checklist was maintained at over 85%. Consensus on ratings was then established between the two raters. Item ratings were summarized into an average rating for each category, as well as overall. Ratings from the final visit (approximately 8-10 months after initial training) were used for the purposes of selecting the case study programs for the current analysis.

The two sets of programs selected differed from one another on all four categories of the checklist as well as on the overall average. In each program, the PIWI groups included children with a variety of disabilities and their parents (primarily mothers). All groups met weekly for approximately 1 1/2 hours.

Results

To facilitate cross-case analysis of factors that may have contributed to these programs being able to implement the PIWI model, a variety of types of data from the study were examined to identify themes that appeared to differentiate the two sets. Four clusters of features emerged, and were used to develop a cross-case summary. Findings in each category are summarized below. As shown in Table 1, Programs A and B are low fidelity programs, whereas C and D are high fidelity programs.

Program Context. Of the two programs with the highest fidelity to the PIWI model, and similarly of the two with the lowest fidelity, one was urban and one was rural. Administrative units of the four projects were varied, with a public school district represented in each set. Thus, neither location nor administrative unit alone distinguished between programs with higher and lower levels of fidelity. In addition, while one might expect that funding would be an issue, 3 of the 4 programs were able to support the parent-child groups through existing sources, whereas the fourth was able to offer the parent-child groups because it applied for and received a grant for this purpose. Thus, differences in availability of funding also did not differentiate the programs.

Another resource that one might expect would influence fidelity to the model is availability of an adequate space for holding the parent-child groups. Each PIWI session is planned around a different developmental topic, and the environment is planned to support that topic. Hence, easy access to particular materials and equipment to use for this purpose is critical. However, of the two programs that had to use non-dedicated space, there was one in each set of programs. One of these, a high fidelity program, had to change locations twice during the course of the project. In comparing these four programs, however, space, while representing an inconvenience, did not emerge as a feature that influenced fidelity to the model.

The primary contextual variable that did appear to distinguish between lower and higher fidelity programs was the target population for the groups, which was driven by the source of funding. The higher fidelity parent-child groups were in programs providing exclusively Part C services to children with disabilities and their families, whereas the lower fidelity groups were in programs providing not only Part C services but also state funded block grant services to low income children/families. These two types of programs were also in communities that differed economically. Consequently, although all programs contained families of a range of income levels, and while ethnic diversity did not differentiate between lower and higher fidelity groups, the overall income and educational levels of the participating families did differ between the two sets of programs.

Service Delivery. Previous experience with parent-child groups did not distinguish between lower and higher fidelity programs. Three of the four programs had some previous experience with parent-child groups. In both lower fidelity programs, the previous groups had been relatively unstructured and did not emphasize parent-child interaction. In the one higher fidelity program with a previous parent-child group, there had been a more direct focus on parent-child interaction.

As might be expected from the fidelity ratings, following the PIWI model as intended clearly distinguished between lower and higher fidelity programs. Staff in the higher fidelity programs not only followed the PIWI schedule, but even at the first site visit were attempting to use all of the components of the model. These included the developmental topic plans provided by project staff, as well as the triadic strategies recommended for interacting with and supporting competence and confidence in parent-child dyads. In the lower fidelity programs, neither of these latter two components was used; instead, the primary focus was the environment.

Another important service delivery feature that clearly distinguished between the two sets of programs was that the higher fidelity programs organized their groups into sets of a limited number of sessions, and also limited the number of dyads in each. This resulted in (a) consistency in parent and child participants across the group meetings in each series, and (b) a sufficient number of facilitators to ensure adequate supportive interactions with each parent-child dyad. Higher fidelity programs also used a team process for planning and implementing their groups, whereas lower fidelity programs tended to have a lead person and an assistant who did not function as an active team member either in planning or implementing the groups.

In general, staff in the higher fidelity programs were more stable and had more team members. In addition, these members represented different professional disciplines. Time was spent in team planning, and during the groups team members shared roles (e.g., introducing the developmental topic) and moved equally among all dyads present, irrespective of discipline. In contrast, in the lower fidelity programs, not only was there turnover in the staff, but responsibility for planning, implementing, and supporting parent-child dyads was not shared.

Staff Characteristics. There were enormous differences between lower and higher fidelity programs not only in the number of individuals on each team and in style of team interaction, but also in the educational backgrounds and professional qualifications of staff. In higher fidelity programs, in contrast to lower, the average education of group facilitators was higher, with all staff having a bachelor's degree or higher. In addition, most staff in higher fidelity programs had professional licenses in a discipline that provides EI services. In both lower fidelity programs, the primary staff person was someone without professional training in birth-3, and also with a bachelor's degree or less. In higher fidelity programs, but not the lower fidelity programs, the group facilitators were the same staff who also provided other EI services. Administrators in the two higher fidelity programs also had more years of experience with birth-3 year olds, and in both higher-fidelity programs, this administrator was part of the PIWI team.

Beliefs about parent-child interaction were coded from transcribed interviews with each staff member prior to beginning the training, using a protocol developed for use in cross-cultural research (Chen & McCollum, 2001). There were no differences between the two sets of programs in beliefs about how children benefit from parent-child interaction (e.g., emotional benefits, cognitive benefits) or in parents' roles in facilitating these benefits (e.g., teacher, caregiver).

One other staff variable of particular interest in this case analysis was the level of understanding of the PIWI model among program staff. Understanding of the model emerged during our initial analyses of post-program interviews as a potentially important contributor to a program's ability to achieve a high level of fidelity. To assist in comparing level of understanding across programs, we developed a process for categorizing participants' responses to the exit interview question, "What contributed to achieving your goals for participating in this project?" This question was asked separately in relation to both (a) achievement of child and family goals, and (b) achievement of staff and program goals. For each of these categories of goals, responses were placed into one of the eight components of the PIWI model by first listing responses of all staff within each program, and then assigning each response to one of the 8 components, using an independent review/consensus process (McCollum & Yates, 2003). These were then tallied across components into two sets: (a) those that fell within the four components viewed by the PIWI developers as representing the model's theoretical underpinnings (i.e., philosophy, dyadic strategies, triadic strategies, and developmental topics), and (b) those that fell into the components representing the 4 more logistical features of the model (i.e., environment, planning, individualizing, and team interaction). The percent of statements that fell into the theoretical components was used as a score that characterized each program's understanding of the model.

As shown in Table 2, differences between lower and higher fidelity programs were strikingly apparent in both sets of comments, demonstrating greater understanding of the model in the higher fidelity programs. In addition, these differences are entirely consistent with the program descriptions above: the higher fidelity programs stressed all components of the PIWI model whereas the lower fidelity programs tended to emphasize primarily logistic components; further, the percent of comments related to the theoretical components was consistently higher in the higher fidelity programs. It is also interesting that it was only in the higher fidelity programs that staff viewed the theoretical components of the model (e.g., changes in parents' understanding of their children, changes in their own roles with parent-child dyads) as contributing to their own growth as professionals and to changes in program quality.

Program Culture . A fourth set of potential program influences on achievement of fidelity included factors that we call "program culture," including reasons for participating in the project, commitment to the PIWI model, and relationships among staff team members. In both higher fidelity programs the team made the decision to participate in the project; in the lower fidelity programs, the administrator made this decision. In all four programs, administrators came with staff to the initial training. However, in the two higher fidelity programs, administrators then took an active role on the teams that implemented the groups, whereas in the two lower fidelity programs administrators took much less active roles. In one, the administrator rarely even looked in on the groups. In the other, the administrator handed responsibility for the groups to a new facilitator without encouraging her to attend the PIWI training. Staff in the higher fidelity programs also sought out additional support from the PIWI staff, beyond the minimum initial training and site visits. For one program, this extra support was essential to their completing the process.

Only in the higher fidelity programs was there a primary focus on supporting parent-child relationships as a reason for selecting the model. In contrast, in the lower fidelity programs, the primary reason for participating in the project was as a way to achieve a consistent schedule and structure for parent-child groups and to achieve smooth transitions between activities and parts of the schedule.   Administrators in these lower fidelity programs saw the PIWI project as a way to help their staff achieve higher quality parent-child groups, whereas the teams in the higher fidelity programs saw the project as a way to support parent-child relationships. Further, only in the higher fidelity programs did the staff seek to use the model in new ways and to go beyond the developmental topic plans provided to them.

Parent-Child Engagement . Parent-child engagement within the context of the play group sessions was used as a measure of whether use of the PIWI model resulted in one of its primary intended outcome, based on the assumption that engagement sets the stage for other aspects of development. PIWI staff observed engagement during the parent-child play portion of the groups during each of their visits. High levels of engagement were found in all programs; over 70% of the parent-child play time, parents and children were interacting with one another. Thus, use of the model appeared to put into place the environmental conditions and support necessary to promote parent-child interaction. However, parent-child dyads in both lower fidelity programs demonstrated relatively higher levels of "neither engaged" and of "child alone" than in the higher fidelity programs, whereas the higher fidelity programs both had higher levels of "both engaged" and "parent available."  

Discussion

The purpose of the current analysis was to explore factors that may contribute to a program's ability to implement the PIWI model for parent-child groups, and in the process to gain insight into factors that may contribute to a program's ability to adopt any new model. We used maximum variation sampling to select four programs from among the 13 in our study, including the two demonstrating the least fidelity to the PIWI model and the two demonstrating the most. A case study approach was then used to describe each program and to examine5 different clusters of factors across the two sets of programs.

As noted in the systemic change literature (Squire & Riegeluth, 2000), extracting any particular features that explains a system's ability to change may be difficult.   In the current case study, features that distinguished lower and higher fidelity programs represented a complex interplay of interdependent factors related to context, service implementation, staff characteristics, and program culture. Thus, among other features, the two programs in which the future group facilitators had made the decision to participate in the project also served children and families exclusively under Part C, had more highly educated staff who worked with the same families in other capacities, had extensive knowledge in their own disciplines, used a team-based approach to planning and implementing the groups, had a highly involved administrator, understood the PIWI model more fully, and sought out assistance from the PIWI staff. In contrast, in the two programs that made the least progress, participation had been initiated by an administrator who subsequently was relatively uninvolved in the project. The population served represented a mixture of families served under Part C and under a state block grant for low income families, staff were fewer in number and less well educated, a team process was not used, and there was less evidence that staff had a full understanding of all of the components of the PIWI model. Staff in these programs rarely sought extra assistance from PIWI staff, although in one program the primary facilitator worked very hard with PIWI staff during follow-up visits to put new practices in place.

A stable staff and a high level of administrative support appeared to be essential for fully implementing the PIWI model; both were present in the higher fidelity programs, and absent in the lower fidelity programs. The higher fidelity programs also began the process with a clear vision for what would be accomplished; upon implementation, they were then able to make adaptations that better fit their situation. Importantly, this initial vision centered on supporting the parent-child relationship, which is also the central tenet of the PIWI model.

The lower fidelity programs did not begin the process with this same vision; rather, the primary concern in these programs was to provide more structure and focus to their existing parent-child groups. These programs were able to accomplish   this by implementing the logistical components of the PIWI model. All of the programs regarded themselves as implementing the PIWI model. It is likely that each program was in fact implementing the model as they understood it and as it addressed their primary reasons for participating. Thus, the two programs that undertook the process in order to focus on parent-child relationships also demonstrated a more complete understanding of the theoretical components of the model, whereas the two that began the process to provide structure to their parent-child groups understood and implemented primarily the logistical components of the model. Despite these differing levels of implementation of the PIWI model, staff in both sets of programs regarded the model as valid and useful for their populations. However, these perceptions of validity were closely linked to the level at which they understood the model.

Even given different levels of fidelity to the model, all programs were able to use PIWI practices to support parent-child interaction and engagement. Nevertheless, the current case analysis reveals differences between higher and lower fidelity programs in each of the areas of context, program, staff, and climate. In addition, certain features tended to cohere across these areas within each set of programs. The similarity between the features that emerged from this case analysis and those identified in previous studies indicate that our findings are not unique to the PIWI model. Therefore, these features provide a set of considerations that should be explored prior to setting out to adopt or promote any particular model, and that also can be expected and planned for as potential supports or barriers.

References

Chen, Y., & McCollum, J.A. (2001). Taiwanese Mothers’ Perceptions of Parent-Infant Interaction with Children with Down Syndrome.   Journal of Early Intervention, 24 (4), 252-265.

Greenfield, P. M., & Suzuki, L. K. (1998). Culture and human development: Implications for parenting, education, pediatrics, and mental health. In W. Damon & R. M. Lerner (Eds.), Handbook of child psychology: Vol. 4 (5th ed., pp. 1059-1109). New York: Wiley.

McCollum, J. A., & Catlett, C. (1997). Designing effective personnel preparation for early intervention. In P. J. Winton, J. A. McCollum, & C. Catlett (Eds.), Reforming personnel preparation in early intervention: Issues, models, and practical strategies (pp. 105-125). Baltimore, MD: Paul H. Brookes.

McCollum, J. A., Gooler, F., Yates, T. J., & Appl, D. (2001). PIWI: Enhancing parent-child interaction as a foundation for early intervention. Infants & Young Children, 14, 34-45.

McCollum, J. A., & Yates, T. (1994).   Dyad as focus, triad as means:   A family-centered approach to supporting parent-child interactions. Infants and Young Children, 6 (4), 54-63.

McCollum, J. A., Yates, T. J., Gooler, F., & Bruns, D. (2001). Using "DOTs" to support parents as developmental observers during parent-child groups. Young Exceptional Children Monographs, 3, 1-12.

McCollum, J.A., & Yates, T.J. (2003) Final Report: Replicability of a parent-child model of early intervention across participants and settings. Washington, DC: Office of Special Education and Rehabilitative Services, U.S. Department of Education.

Squire, K. D., & Reigeluth, C. M. (2000). The many faces of systemic change. Educational Horizons, Spring, 143-152.

Winton, P., & McCollum, J. A. (1997). Ecosystem perspectives on personnel preparation: Rationale, framework, & guidelines for change. In P. J. Winton, J. A. McCollum, & C. Catlett (Eds.), Reforming personnel preparation in early intervention: Issues, models, and practical strategies (pp. 3-25). Baltimore, MD: Paul H. Brookes.

Wolfe, B. L., & Snyder, P. (1997). Follow-up strategies: Ensuring that instruction makes a difference. In P. J. Winton, J. A. McCollum, & C. Catlett (Eds.), Reforming personnel preparation in early intervention: Issues, models, and practical strategies (pp. 173-190). Baltimore, MD: Paul H. Brookes.

Table 1

Fidelity Ratings* for Four Programs with Lowest and Highest Fidelity Ratings

Note. *5-point scale.

Table 2

Percent of Staff Comments Related to Theoretical Components of PIWI Model

home . about the conference . programme . registration . accommodation . contact