An assessment of the North Carolina school-age child care accreditation initiative.

Abstract

The primary purpose of this study was to determine if participation in the North Carolina Quality Enhancement Initiative (NC QEI) improved the overall quality of the 26 participating school-age child care programs in three North Carolina communities. A paired t-test showed a positive and significant increase in the quality of school-age child care environments and teacher/child interactions over the 9-month period, from pre-initiative to post-initiative. Pre-initiative and pre- to post-initiative difference scores on each dependent measure were grouped together using a cluster analysis. Each cluster of programs was compared to determine relationships between these program variables and the program clusters. A state license and a smaller group size were related to higher quality programs at pre-initiative. Director education, teacher salary, state license, and program size were related to greatest program improvement from pre-initiative to post-initiative. In addition, six of the 10 programs were awa rded accreditation by the National School-Age Care Alliance (NSACA). These findings suggest that participation in program improvement initiatives, like the NC QEI, is a viable means of improving the quality of school-age child care programs.

**********

Continuing increases in the number of mothers in the workforce has created a phenomenon called "out-of-school time" care or "school-age child care," which is care for school-age children during the hours when school is not in session. While the term "school-age child care" typically includes children ages 5 to 12, in elementary school, programs for middle-school students (ages 12-14) are also increasing in number (National Institute on Out-of-School Time, 1998). In 1991, a national study found that 1.7 million children, kindergarten through 8th grade, were enrolled in 49,000 formal before- and after-school programs (Seppanen et al., 1993). A more recent study estimated that 39% of kindergartners through 3rd-graders receive some form of non-parental care before and/or after school on a weekly basis (Brimhall & Reaney, 1999). This figure translates to a total of 6.1 million primary grade children, who spend an average of 14 hours per week in out-of-school time child care arrangements. Due to the increasing numb er of children enrolled in schoolage child care programs, policymakers, child related advocates, businesses, and educators are strategizing how to effectively improve the quality of care available to families and children. One professional business organization, the American Business Collaboration of Quality Dependent Care (ABC), funded a pilot Quality Enhancement Initiative (NC QEI) in the state of North Carolina to aid in the implementation of a national system developed by the National School-Age Care Alliance (NSACA) to accredit school-age child care programs.

Prior to the recent concern for the quality of school-age care program, policymakers, educators, and parents were concerned about the quality of infant, toddler, and preschool care and education programs. In the early 1980s, the National Association for the Education of Young Children (NAEYC) began an accreditation program that outlined the criteria necessary to support the physical, cognitive, and socio-emotional development of young children birth to age 8. NAEYC defines accreditation as a process in which a program's director, staff, and parents voluntarily work with representatives of the association to determine whether the program meets nationally recognized criteria for high quality. Programs achieving accreditation have demonstrated a commitment to providing the highest quality care and education.

The NAEYC accreditation system also includes school-age child care classrooms if the majority of children are 8 years old and younger. For example, 31% of all NAEYC-accredited programs also serve school-age children (age 6-8), not just infants, toddlers, and preschoolers. Furthermore, 3% of accredited programs serve school-age children exclusively (Bredekamp & Glowacki, 1996). Programs are awarded accreditation based on their ability to provide developmentally appropriate environments, as defined by NAEYC's position statement on developmentally appropriate practices for children birth to age 8. Since the NAEYC accreditation initiative was created and implemented almost 10 years prior to the NSACA accreditation system, a review of NAEYC accreditation findings provides valuable information in understanding the effectiveness of an accreditation system for child care centers in improving program quality.

Unfortunately, rather limited research has been conducted to measure the level of quality achieved and maintained by NAEYC-accredited programs. Whitebook (1996) argues that there is consensus within the early childhood field, and among policymakers, that NAEYC accreditation standards represent a level of quality that exceeds the licensing standards and current level of care in most states. Because of this consensus, initiatives that help centers achieve NAEYC accreditation attract a wide range of funders (i.e., corporations, foundations, unions, community groups, and governments). While many supporters have committed funds to help programs through the self-study and validation process fundamental to NAEYC accreditation, few have funded research into the success of the NAEYC accreditation system.

Accreditation As a Mechanism To Improve Quality

Although the accreditation process was not their primary focus, two national preschool (birth to age 4) child care studies--the Cost Quality and Child Outcome (CQCO) (Helburn, 1995) and the National Child Care Staffing Study (NCCSS) (Whitebook, Howes, & Phillips, 1989)--included NAEYC-accredited programs in their investigation. Findings from the NCCSS study indicate that accredited programs provided better than average quality of care. The 14 accredited centers in the sample differed from non-accredited centers on all dimensions of quality. In fact, accredited centers paid higher wages, provided better benefits and working conditions, and had lower turnover rates. Furthermore, teachers in accredited programs were better educated and had more early childhood training. The accredited programs also provided more developmentally appropriate activities and had better staff/child ratios. Teachers in accredited programs were rated as providing more appropriate care than teachers in non-accredited programs. Overall, the centers accredited by NAEYC provided high-quality care to children in the NCCSS study.

Findings from the CQCO study showed that NAEYC-accredited centers had higher than average quality; as a group, however, the accredited centers (31 out of 401) did not provide the highest quality of care. When the accredited centers were compared with three other types of centers that also provided higher than average quality care (i.e., publicly operated, work-site, and publicly funded), accredited programs did not provide as high a quality of services as some of the other types of programs. However, there was some overlap among types. For example, a higher quality work-site program also may be accredited. The staff/child ratios in accredited programs were similar to those in publicly funded programs, but not as high as those in work-site or publicly operated programs. Accredited programs also employed more teachers with at least a college degree and were more likely to offer health insurance and pay somewhat higher wages for teachers and assistants in comparison to the other centers offering higher quality c are. However, two indicators of quality--staff/child ratios and turnover (called "tenure" in the report)--showed no differences between accredited and non-accredited centers. Therefore, the CQCO study findings suggest that NAEYC-accredited programs are better than average but not necessarily the highest quality centers in a community. While these studies provide valuable information about NAEYC-accredited centers, the NCCSS and CQCO studies did not investigate the impact that participation in the NAEYC accreditation process had on program quality.

Whitebook, Sakai, and Howes (1997) have conducted the only large-scale investigation of the NAEYC accreditation process by assessing 92 child care centers in three California communities. The researchers began tracking centers when they initiated the accreditation process, followed their progress over time, and compared them to other centers in their communities. Interviews with teaching staff and center directors also were conducted. Findings suggest that centers that achieved accreditation were of higher quality when they began the accreditation process and showed greater improvement in overall quality scores, staff/child ratios, and higher staff/child interaction scores than did programs that sought, but did not achieve, accreditation. In fact, centers that began the accreditation self-study but did not complete the process demonstrated no improvement in classroom quality, staff/child ratios, and staff/child interactions. However, almost 40% of the centers were rated as mediocre in quality, in spite of imp rovements they had made while undergoing the accreditation process. A nonprofit status, higher wages, and retention of skilled staff, in combination with NAEYC accreditation, were predictors of high quality care in the participating child care centers.

Given the NAEYC accreditation findings, an accreditation initiative exclusively for school-age child care may be a viable strategy for improving school-age child care programs. The goal of this present study was to determine the effectiveness of the North Carolina initiative. Specifically, the study assessed whether participation in the North Carolina Quality Enhancement Initiative improved the quality of the school-age child care environment and teacher/child interactions, by observing programs before and after participation in the initiative. In addition, information on the relationships between structural and process variables were examined and analyzed as they relate to program quality in school-age child care programs.

Method

Participants

The school-age child care programs in the study were selected by the funders of the initiative--IBM, AT&T, and GE Capital--from among program applicants in three target communities in North Carolina. Participation priority was given to programs that served children of employees at the three funding companies. The goal of the project was to select 10 programs from each community. However, not all of the communities were able to recruit 10 programs; therefore, some communities recruited more than 10, and some did fewer. Twenty-eight out of the 30 selected programs agreed to participate in the evaluation.

Procedures

Pre-initiative observation and program demographic survey data were collected on a total of 28 programs (N=28). Post-initiative data were collected on 26 programs (N=26) in May of 1998: 7 in Greensboro, 8 in Raleigh, and 11 in Charlotte. Over the course of the project, four programs dropped out of the initiative, two of which were included in the evaluation.

The program improvement initiative began with a two-day training event conducted and planned by two NIOST training associates. Prior to the two-day training, program administrators were asked to complete a newly developed questionnaire, Readiness Scale for Program Improvement and Accreditation in School-Age Child Care Programs (O'Connor, 1997), to place each program in one of two groups-First Steps or Team Works. The Team Works programs were on a faster paced track designed to have them ready to apply for NSACA accreditation in the spring of 1998. The First Steps programs were on a more leisurely paced track focusing on targeted program improvements during the first year, with the hope of being ready to apply for NSACA accreditation in 1999. The two groups, First Steps and Team Works, had different training agendas during their two-day training event. Each program was also assigned an adviser, who provided on-site consultation, telephone consultation, and/or resource development for approximately nine months (September 1997 to May 1998). The nine advisers (three from each community) were trained for two days in May of 1997 and were allotted a maximum of 10 hours of technical assistance per program and 10 hours of bi-monthly peer support meetings. In addition, each participating program received two sources of information to assist them in program improvement: the NSACA Pilot Standards (Sisson, 1995) and the Assessing School-Age Quality (ASQ) Kit.

The post-initiative observation visits were conducted in May of 1998, prior to the NSACA accreditation endorser visits. The same age group visited in the fall of 1997 was observed post-initiative. The data collectors conducting the observations were blind to which programs had applied for accreditation.

Measures

Information on school-age child care program structural or regulatory features was collected via a questionnaire, which requested information about the staff/child ratio, group size, director's level of education, and number of children served. Pre-initiative and post-initiative observations were conducted using three observation measures. First, the School-age Care Environment Rating Scale (SACERS) (Harms, Jacobs, & White, 1996) was used to assess the school-age child care environment. The SACERS assesses the developmental appropriateness of the school-age child care program, focusing on 43 items covering six sub-scales: space and furnishings, health and safety, activities, interactions, program structure, and staff development. There is a seventh sub-scale, of six items, for programs that include children with special needs. Each item is rated on a 7-point scale with the score of 1 signifying inadequate, 3 minimal, 5 good, and 7 excellent. An average score on the 43 items is then calculated. The SACERS was chosen for measuring the program environment because it is a comprehensive "best practice" rating scale for school-age child care programs. The SACERS and the NSACA Pilot Standards (used in the accreditation process) assess similar areas: indoor and outdoor environment, health and safety, activities, interactions, and administration.

Reliability and validity of the SACERS has been evaluated in several ways. Reliability assessments include Cronbach's alpha = .95, inter-rater agreement weighted Kappa = .83, and intraclass correlation r = .96. The validity of the SACERS has been established through high agreement between SACERS scores and expert evaluations of quality. In the present study, inter-rater reliability was established at 73%, and maintained at 80% on 20% of the total programs visited (two in each community). SACERS reliability was re-established among all data collectors after 6 months at 79% and maintained at 80% again on 20% of the total programs visited (two in each community). Inter-rater reliabilities were established to a criterion of 75% exact agreements for all observational measures.

The quality of adult-child interactions was measured using the Caregiver Interaction Scale (CIS) (Arnett, 1989) and the "Human Relationships Keys of Quality" observation section from the Pilot Standards for Quality School-age Child Care (Sisson, 1995). The CIS was selected to measure the quality of teacher-child interactions because of its previous use in national child care studies (Helburn, 1995; Whitebook, Sakai, & Howes, 1997). The CIS is a 26-item scale used to rate a single teacher. A score of 1 indicates that a given behavior is "never true," while a score of 4 indicates that the behavior is "often observed." For example, the CIS measure includes statements such as "speaks warmly to the children" and "doesn't supervise the children very closely." A criterion level of 80% agreement between observers was established in the study for which this measure was developed (Arnett, 1989). Reliability was established at 89%, maintained on 20% of the programs at 92%, re-established at 84% for post-initiative, and maintained at 97%.

In addition, teacher-child interactions were assessed using the "Human Relationships Keys of Quality" from the NSACA pilot accreditation standards ASQ program observation (O'Connor, Gannett, Heenen, & Mattenson, 1996). The human relationships (HR) category consists of 9 keys, as well as standards specific to each key, for a total of 36 items. A score of 0 indicates "no evidence" or "not met," while a score of 3 indicates the standard is fully met. "Staff relate to children in positive ways" and "Staff use positive techniques to guide children's behavior" are two examples from the nine human relationship keys. In order for a program to achieve accreditation, NSACA guidelines state that the program must score at least a 10 on each human relationship key and at least a two on each standard. Reliability studies on the entire ASQ Program Observation Instrument (O'Connor, Wheeler, Harms, & Oryer, 1994) indicate an overall instrument intra-class correlation coefficient of .84, a test-retest Kappa co-efficient of .85 , and an overall Cronbach's alpha of .89. Inter-rater reliability was established at 78%, maintained at 93%, re-established at 78%, and maintained at 95% throughout 20% of the total observations.

The data collector arrived about 45 minutes before the children arrived, met with the site director, received a tour of the program, and observed until most of the children were picked up by their parents. The observation lasted approximately 3 hours.

Results

A summary of the program and director demographic information is summarized in Table 1. Tests for normality were computed on each of the three observational scores (CIS, SACERS, and HR), pre-initiative and post-initiative, using the Shapiro-Wilk test for normality. All distributions were normal, except for the HR post-initiative score distribution (p=.0094). This distribution was skewed toward the higher scores (ranging from 4.67 to 10.67, M = 8.24), which was noted. An alpha level of .05 was used to determine significance for all statistical tests.

On the SACERS, pre-initiative scores differed significantly from post-initiative, t (25) = 3.73, p = .0005. In addition, on the HR, pre-initiative scores differed significantly from post-initiative, t (25) = 2.64, p = .0070; on the CIS, pre-initiative scores differed significantly from post-initiative, t (25) = 1.70, p = .0511 (see Table 2). On the SACERS, 21 individual programs improved their scores; on the HR, 17 programs improved; and on the CIS, 18 programs improved. Overall, there was a positive and significant increase on each observational measure pre-initiative to post-initiative.

Profile Variables Related to Pre-Initiative Scores

Because of the small sample of programs and even smaller cell sizes of profile variables, any further tests of statistical significance tests were not appropriate. Rather, it was deemed more appropriate to cluster the dependent variable scores by program and then study the cluster relationships by program characteristics, or by the profile variables, such as group size. A cluster analysis is a multivariate technique that groups programs into clusters so that the programs in the same cluster are more similar to one another than they are to programs in the other clusters by some predetermined selection criteria. The intent is to maximize the homogeneity of programs within the clusters while also maximizing the heterogeneity among the groups (Hair, Anderson, Tatham, & Black, 1998). The cluster predetermined selection criteria for this cluster analysis were the mean scores on the three dependent measures.

The pre-initiative scores on all three dependent measures were cluster analyzed using the average method of hierarchical clustering, resulting in three clusters of programs. The means and standard deviations of all three measures by cluster are summarized in Table 3. Interpreting what types of programs these clusters represent involves using the selection criterion variables to name or assign a label accurately describing the clusters (Hair et al., 1998). Cluster One (n=7) had the programs with the lowest mean scores, or below the average of all 30 programs on all measures except on the CIS (see Table 2). Cluster Two (n=5) had the programs at about the overall mean scores of participating programs on all measures, as reported in Table 2. Cluster Three (n=14) had the programs with the higher than average mean scores on all three measures. Clustering the programs by all three dependent measures indicates that about half (12) of the programs were about average or below average, and about half (14) were above ave rage. These clusters were compared with profile variables associated with high-quality child care programs.

To create the profile variables (i.e., director education, director salary, program size, auspice, state license, and staff/child ratio) used in this analysis, the frequencies and percentages were grouped to reduce the data (or cell sizes) to no more than three groups for each profile variable. Director education was divided into high school education (n=2), some college (n=5), and four years of college or more (n=19). The director salary variable was split into two fairly even groups: $10,000 to $20,000 per year (n=10) and $20,000 or higher (n=11). Program size was grouped by 1 to 30 children (n=7), 31 to 70 children (n=14), and 71 plus children (n=5). By using classifications made by the National Study of Before- and After-School Programs (Seppanen et al., 1993), the programs were divided into small programs, medium programs, and large programs. The program auspice was categorized as for-profit (n=9) or non-profit (n=17). Participating programs either had a state license (n=19) or they did not (n=6). The st aff/child ratios were split into three groups: 1:8-1:12 (n=10), 1:14-1:15 (n=11), and 1:18-1:24 (n=5). These ratio groupings represent low teacher/child ratios, average teacher/child ratios, and high teacher/child ratios. For accreditation, NSACA requires 1:8-1:12 for children age 6 and below and 1:10-1:15 for children age 6 and above (low teacher/child ratios). North Carolina's "A" License ratios are 1:20 for age 5 and below and 1:25 for age 5 and older (high teacher/child ratios).

To examine the relationship between the profile variables (i.e., director education, director salary, staff/child ratios, program size, auspice, and state license) and program clusters--all categorical variables--a measure of association for a contingency table was used. Due to the small sizes of many cells, interpretation of the chi square test statistic would be suspect. Therefore, Pearson's measure of association (or Pearson's P) for a contingency table was computed. The computed associations between the clusters of programs and the profile variables are as follows: staff/child ratios P = .24, director education P = .31, director salary P = .14, program auspice P = .08, program size P = .38, and state license P = .58. The P coefficient can be interpreted like a correlation coefficient; therefore, the associations of .35 or higher warranted closer examination (Bishop, Feinberg, & Holland, 1975). In this case, those variables were program size and state license. The cross-tabs contingency table (see Table 4) reports the distribution of profile variables by cluster.

Cluster Three, the highest quality cluster, had all (100%) licensed centers and a slightly larger percentage (38%) of smaller center sizes. Cluster One, the lowest quality cluster, had a higher percentage (72%) of medium size programs and 71% of the programs were not licensed. Overall, a state license and smaller program size were related to a higher quality program clusters pre-initiative, but lower staff/child ratios, higher director education, higher director salary, and non-profit program status were not associated with the higher quality program clusters pre-initiative.

Profile Variables Related to Pre-Initiative to Post-Initiative Scores

The pre-initiative to post-initiative difference scores on all three dependent measures were clustered using Ward's method of hierarchical clustering, resulting in three clusters of programs. The means and standard deviations of all three difference scores by cluster are summarized in Table 5. Cluster One (n=9) had programs with negative change scores, while Cluster Two (n=9) had the programs with small, positive changes in scores. Cluster Three (n=8) had the programs with large, positive change scores.

To examine the relationship between the profile variables (i.e., director education, director salary, staff/child ratios, program size, auspice, and state license) and program clusters--all categorical variables--a Pearson P measure of association for a contingency table was calculated. The computed associations between the clusters of programs and the profile variables are as follows: staff/child ratios P = .26, director education P = .43, director salary P = .24, program auspice P = .04, program size P = .47, and state license P = .39. Associations that warranted closer examination were those greater than .35, which included prograin size, state license, and director education. The cross-tabs contingency table (see Table 5) summarizes the distribution of program size, state license, and director education variables by cluster.

Cluster Three, the programs that increased the most, had more directors with a high school education (25%) and the lowest percentage of directors with four years of education or more. Also, Cluster Three had an equal number of non-state licensed programs (50%) as well as licensed programs (50%). Cluster One, programs that declined in scores, had a larger percentage (44%) of large-size centers (with 71+ children), yet were predominantly licensed. Cluster Two, programs that improved slightly, were all licensed centers except for one, and had a larger percentage of medium-size programs (66%) and no large programs. Cluster Two also had a higher percentage (89%) of directors with four years or more of education. Programs that had directors with a higher level of education, were larger, and were licensed by the state showed the least improvement. The programs that showed the greatest improvement had directors with a lower level of education and fewer licensed programs. Large-size programs were more likely to show a decline in scores. Overall, a state license, smaller program size, and director education were related to program improvement pre-initiative to post-initiative, but lower staff/child ratios, teacher turnover, higher director salary, and non-profit program status were not associated with program quality improvement from pre-initiative to post-initiative.

Discussion

Participation in the North Carolina Quality Enhancement Initiative (NC QEI) improved the quality of the participating programs' school-age child care environment and teacher/child interactions. Overall, there was a positive and significant increase on each measure from pre-initiative to post-initiative. Twenty-one of the 26 programs improved the overall quality of their school-age child care environments, as assessed by the SACERS. Likewise, 18 programs had teachers who improved the quality of their interactions with children, as assessed by the Caregiver Interaction Scale, and 17 programs improved interactions, as assessed by the Human Relationship section of the accreditation observation. Furthermore, six of the 10 programs that applied to be accredited by the National School-Age Care Alliance (NSACA) were awarded accreditation. Four were deferred, but only one applied for a deferral accreditation visit and was subsequently awarded NSACA accreditation in the fall of 1998. Since then, another NC QEI program was accredited in January of 2001. In the absence of a comparison group, it is impossible to conclude that participation in the NC QEI was the sole cause of improvement in the quality of the programs. Nonetheless, programs did make significant gains in overall program quality and teacher/child interactions during the nine months they participated in the NC QEI, suggesting that the involvement of school-age child care programs in program improvement initiatives is a viable means of improving the quality of school-age child care programs.

The NC QEI findings are consistent with the results of the evaluation of the first pilot of the NSACA National Improvement and Accreditation System for school-age child care programs, which reported a positive mean score change on all but one of the 21 program observation keys (Miller, 1997). The initial pilot of the NSACA accreditation system reported summary information on 75 programs across the United States. Pre- and post-data were collected on 26 of the 32 ABC program sites in California, Colorado, Georgia, and New Jersey, using the Pilot Standards as the observation measure. The evaluator of the initial pilot acknowledges as confounds the small sample size, as well as biases in the observation reliabilities (the ASQ advisers were the pre- and post-test observers). In the NC QEI evaluation, observers had high reliabilities and were blind to program groupings and accreditation applicants. Furthermore, in the present study the program participants were not aware of the evaluation measure being used to asse ss their improvement. For these reasons, the more rigorous design of the NC QEI evaluation provides stronger empirical evidence in support of the NSACA accreditation self-study process than the first pilot did.

It is important to note that the programs awarded NSACA accreditation were above minimal (3.0) in quality, but were not considered of good (5.0) quality according to the SACERS instrument. The overall mean SACERS score for accredited programs was 4.87 at post-initiative, indicating that many of these programs did not quite achieve the level of good quality. At pre-initiative, the overall mean score for programs that later achieved accreditation was 3.81 (minimal-3.0 to good-5.0 quality). These findings are consistent with NAEYC accreditation study findings (Howes & Galinsky, 1996; Whitebook, Sakai, & Howes, 1997; Zellman & Johansen, 1996), indicating that the accreditation process does improve the quality of preschool child care programs. Specifically, accredited programs in the Whitebook, Sakai, and Howes (1997) study had a mean score on the ECERS (Harms & Clifford, 1980) of 4.58 at pre-test and 5.22 at post-test. The programs in the NC QEI study showed greater improvement (+1.06) than the programs in the NA EYC study (+.64) but were of lower quality (4.87 compared to 5.22) at post-test, as measured by the ECERS and SACERS. The NAEYC accreditation study (Whitebook, Sakai, & Howes, 1997) also found that 39% of NAEYC-accredited programs were rated as mediocre in quality. In the NC QEI study, the scores of the six accredited programs on the SACERS ranged from 4.16 to 5.61. Due to these mediocre scores, concern has been raised about whether accredited centers truly reflect high quality, and whether these centers can sustain high quality. The level of quality and sustainability of quality are issues that need to be addressed by NSACA as more programs around the country apply for accreditation.

Although the two pilot NSACA accreditation initiatives have been successful in improving quality, many school-age child care programs still may not seek accreditation. Bredekamp (1999) estimates that without sufficient incentives or mandates, only 5% of programs will voluntarily seek accreditation, and that 5% are probably already providing higher quality care without accreditation. Another issue related to the accreditation process is that parents may not be able to afford the higher cost of quality required by the NSACA standards; therefore, the cost may need to be subsidized. Thus, the school-age child care accreditation system is not a panacea for improving child care quality.

Profile Variables Related to Pre-Initiative Scores

A state license and smaller program size were related to higher quality programs clusters at pre-initiative. This finding is consistent with the preschool child care research studies that have found licensure and group size to be related to program quality (Helburn, 1995; NICHD, 1996; Whitebook et al., 1989). In addition, school-age child care studies have found the number of children enrolled (Rosenthal & Vandell, 1996) and licensing to be associated with higher quality programs (Miller, 1997). Intuitively, it is not surprising that a state license would be related to higher quality. However, it is important to consider the stringency of the licensing standards in a state as it relates to the overall quality of school-age child care programs. Certainly, states with more stringent standards would ensure higher quality programs. Less than adequate standards, as in North Carolina, would not ensure higher quality. It is once again important to note that although a state license was related to higher quality mean scores on the SACERS, the scores still reflect less than good quality.

A smaller program size was also found to be related to the cluster of higher quality school-age child care programs. Rosenthal and Vandell (1996) also found that larger program size in school-age programs is related to poorer quality. Smaller group sizes have been shown to be related to higher quality centers in the preschool child care research (Helburn, 1995; NICHD, 1996; Whitebook et al., 1989). Preschool research is based on group size, which is similar to program size in school-age programs where children are encouraged to group themselves in activities by individual choices and interests, not by age group or classroom groupings. The relationship between program size and class size is similar, in that if teachers have smaller numbers of children with whom to work, they are more likely to know the children as individuals and engage in more frequent staff/child interactions, which would result in more positive staff/child relationships.

Profile Variables Related to Improvement Scores

A state license, smaller program size, and director education were related to program improvement from pre-initiative to post-initiative. Programs with a state license and that were smaller in size showed less improvement than the group with larger programs and no state license. One obvious reason for this finding may be that programs with a license already are meeting a minimal level of quality and therefore had fewer program changes to make than programs without a license. In addition, based on the findings of this study, programs with smaller enrollments were also higher quality to begin with and therefore had fewer changes to make than larger-size programs. It also may be that changes are easier to make in smaller programs because there are fewer environmental changes needed and fewer staff with whom to work.

The programs that showed the least amount of improvement had directors with higher levels of education, while programs with directors with a minimal education showed the most improvement. This NC QEI finding seems contradictory to the preschool child care research, but it may be that directors with more education were already facilitating higher quality programs and therefore had fewer changes to make. Likewise, the programs with directors with less formal education may have shown more improvement because they had more to learn and, therefore, could gain more knowledge by participating in the training and consultation offered by the NC QEI. This finding suggests that staff with less formal education, as well as those with more, can make program improvements if they are given tools and assisted with the process of program improvement.

Barriers to Program Improvement

Of some concern is the fact that a cluster of nine programs in the initiative actually declined in quality over the course of the project. Hypothesizing about what may have accounted for the decline in program quality, when the programs were actually focused on improving overall quality, is important in understanding the process. According to program directors in the first pilot, the greatest barriers to program improvement were staff turnover, finding time to coordinate the improvement effort, getting parents on the ASQ team, and getting questionnaires returned. ASQ advisers in the first NSACA pilot also mentioned having problems building a relationship with some programs, which affected their ability to provide assistance. The quality of the relationships between the ASQ adviser and the program staff was a key factor in the effectiveness of the technical assistance provided and utilized by a program. Staff turnover also was a major barrier to program improvement, because programs that had turnover at the si te director or leadership level, or that had high levels of teacher turnover during the course of the pilot, were less likely to improve in quality (Miller, 1997).

Another barrier to programs achieving accreditation may have been the tremendous variability in program quality at pre-initiative. For example, SACERS scores at pre-initiative ranged from 2.21 to 4.53. Some programs were prepared to attempt the full accreditation process, while others may have been better off focusing on a few selected areas of improvement. The findings from the first pilot (Miller, 1997) also illustrate that not all programs were ready for the accreditation process that is aimed at programs with a "measure of stability, quality and internal leadership" (p. 10).

Overall, the positive results from the evaluation of the NC QEI project indicate that working toward NSACA accreditation may be one mechanism that can improve the quality of school-age child care programs. However, it is not a panacea for improving school-age child care quality. The NC QEI-accredited programs that achieved NSACA accreditation status were still not of high quality, as measured by the SAGERS. Also, the accreditation process is an ambitious undertaking requiring significant motivation and hard work on the part of the site director as well as members of the ASQ team. Many school-age child care programs will not voluntarily seek NSACA accreditation; therefore, licensing standards may continue to be the only standards guiding school-age child care programs. Clearly, the relationships identified in this study need further examination with a larger and more nationally representative sample. This information will not only provide additional information about high quality school-age child care programs , but also aid in improving the NSACA accreditation system.

Table 1

Program and Director Demographic Information Summary

Characteristic             n   f   P         M     SD   Range

Total Enrollment           28             54.5   27.5   10-109
State License              25
  Yes                          19  76%
  No                            6  24%
Auspice                    28
  Community College             1   4%
  YMCA                          5  18%
  YWCA                          1   4%
  Church                        5  18%
  Private/Profit                9  32%
  Public School                 7  25%
Teacher/Child Ratios       28
  1:8 - 1:12                   11  40%
  1:14 - 1:15                  12  42%
  1:18 - 1:25                   5  18%
Director Education         28
  HS                            4  14%
  Two Years' Higher Ed.         4  14%
  Three Years' Higher Ed.       1   4%
  Bachelor's                   15  54%
  Master's                      4  14%
Major                      24
  Education                     9  38%
  Child Development             4  17%
  Psychology                    4  17%
  Other (unrelated)             7  28%
Yearly Salary (dollars)    23           22,473  7,856  5-39,000
Benefits                   28
  Health Insurance             22  79%
  Retirement                   17  61%
  Annual Leave                 23  82%
  Sick Leave                   24  86%

Table 2

Mean Scores on Observation Measures, Pre-Initiative to Post-Initiative

                   Pre-Initiative                    Pre-Initiative
Measure     M        SD       Range         M        SD       Range

SACERS     3.41     .72       2.21-4.53    4.09     .90       2.09-5.61
HR         6.58     3.3       0-12         8.24     1.82      4.67-10.68
CIS        2.92     .40       2.04-3.50    3.09     .48       2.12-3.81

          Pre-Initiati
               ve
Measure   t (25)

SACERS    3.73 ***
HR        2.64 **
CIS       1.70 *

Note: * p < .05

** p < .01

*** p < .001.

Table 3

Pre-Intiative Cluster Relationships

          Cluster One             Cluster Two           Cluster Three
          (below average)         (average)             (above average)
          n = 7                   n = 5                 n = 14
          M           SD        M         SD        M           SD

SACERS    2.87         .57      3.23       .32      4.01         .38
HR        3.57        2.07      5.20      2.59      8.57        2.65
CIS       2.85         .19      2.25       .16      3.18         .21

Table 4

License and Program Size by Cluster Contigency Tables

                 ClusterOne (Low           Cluster Two (Medium
                 Quality)Count Col%        Quality) Count Col%

Not Licensed     5           71%         1             20%
Licensed         2           29%         4             80%
Small Programs   1           14%         1             20%
Medium Programs  5           72%         4             80%
Large Programs   1           14%         0              0%

                 Cluster Three (High
                 Quality) Count Col%

Not Licensed      0              0%
Licensed         13            100%
Small Programs    5             38%
Medium Programs   5             36%
Large Programs    4             29%

Table 5

Pre-Initiative to Post-Initiative Cluster Relationships

                               Cluster One
                               (Negative Change) n = 9
                                 M                   SD

Difference Scores Pre to Post

SACERS                         -0.08                 .53
HR                             -1.14                1.73
CIS                             -.21                 .43

                               Cluster Two
                               (Small positive) n = 9
                                M                   SD

Difference Scores Pre to Post

SACERS                          .58                 .48
HR                             1.8                 3.07
CIS                             .14                 .41

                               Cluster Three
                               (Large positive) n = 8
                                M                   SD

Difference Scores Pre to Post

SACERS                         1.18                 .53
HR                             4.64                1.44
CIS                             .65                 .38

Table 6

License, Program Size, and Director Education by Cluster Contingency
Tables

                             Cluster One                 Cluster Two
                          (Least Change)           (Moderate Change)
                             Count Col%                   Count Col%
Variable

Not Licensed            1     12%             1           11%
Licensed                7     83%             8           89%
Small Programs          1     11%             3           33%
Medium Programs         4     44%             6           66%
Large Programs          4     44%             0            0%
HS Education            0      0%             0            0%
Some Higher Ed          2     22%             1           11%
4 Years or More         7     78%             8           89%

                             Cluster Three
                         (Greatest Change)
                                Count Col%
Variable

Not Licensed           4         50%
Licensed               4         50%
Small Programs         3         38%
Medium Programs        4         55%
Large Programs         1         12%
HS Education           2         25%
Some Higher Ed         2         25%
4 Years or More        4         50%

References

Arnett, J. (1989). Caregivers in day-care centers: Does training matter? Journal of Applied Developmental Psychology, 10, 541-552.

Bishop, Y M. M., Feinberg, S. E., & Holland, P. W (1975). Discrete multivariate analysis: Theory and practice. Cambridge, MA: MIT Press.

Bredekamp, S. (1999). When new solutions create new problems: Lessons learned from NAEYC accreditation. Young Children, 54(1), 58-63.

Bredekamp, S., & Glowacki, S. (1996). The first decade of NAEYC Accreditation: Growth and impact on the field. In S. Bredekamp & B. Willer (Eds.), NAEYC accreditation: A decade of learning and the years ahead (pp. 110). Washington, DC: National Association for the Education of Young Children.

Brimhall, D. W., & Reaney, L. M. (1999). Participation of kindergartners through third-graders in before- and after-school care. Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement, National Center for Education Statistics.

Hair, J. E., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivari ate data analysis (5th ed.). Englewood Cliffs, NJ: Prentice Hall.

Harms, T., & Clifford, R. M. (1980). Early childhood environment rating scale. New York: Teachers College Press.

Harms, T., Jacobs, E. V., & White, D. R. (1996). School-age care environment rating scale. New York: Teachers College Press.

Helburn, S. W. (Ed.). (1995). Cost, quality, and child outcomes in child care centers, technical report. Denver, CO: University of Colorado.

Howes, C., & Galinsky, E. (1996). Accreditation of Johnson and Johnson's child development center. In S. Bredekamp & B. Willer (Eds.), NAEYC accreditation: A decade of learning and the years ahead (pp. 47-60). Washington, DC: National Association for the Education of Young Children.

Miller, B. M. (1997). Final report and evaluation: Pilot of the national improvement and accreditation system. Wellesley, MA: School-age Child Care Project.

National Institute of Child Health and Human Development (NICHD) Early Childhood Research Network. (1996). Characteristics of infant child care: Factors contributing to positive caregiving. Early Childhood Research Quarterly, 12, 281-303.

National Institute on Out-of-School Time. (1998). Fact sheet on school-age children. Retrieved from www.wellesley.edu/WCW/CRW/SAC/factsht/html

O'Connor, S. (1997). Readiness scale for program improvement and accreditation in school-age child care programs. Wellesley, MA: School-Age Child Care Project.

O'Connor, S., Gannett, E., Heenen, C., & Mattenson, P. T. (1996). Assessing school-age child care quality. Wellesley, MA: School-Age Child Care Project.

O'Connor, S., Wheeler, K., Harms, T., & Cryer, D. (1994). The revision and redevelopment of the ASQ program observation. Wellesley, MA: School-Age Child Care Project.

Rosenthal, R., & Vandell, D. L. (1996). Quality of care at school-aged child-care programs: Regulatable features, observed experiences, child perspectives, and parent perspectives. Child Development, 67, 2434-2445.

Seppanen, P. S., Love, J. M., DeVries, D. K., Bernstein, L., Seligson, M., Marx, F., & Kisker, E. E. (1993). National study of before and after school programs. Final Report. Washington, DC: U.S. Department of Education Office of Policy and Planning.

Sisson, L. (1995). Pilot standards for quality school-age child care. Wellesley, MA: National School-Age Care Alliance.

Whitebook, M. (1996). NAEYC accreditation as an indicator of program quality: What research tells us. In S. Bredekamp & B. Willer (Eds.), NAEYC accreditation: A decade of learning and the years ahead (pp. 31-46). Washington, DC: National Association for the Education of Young Children.

Whitebook, M., Howes, C., & Phillips, D. (1989). Who cares? Child care teachers and the quality of care in America (final report of the National Day Care Staffing Study). Oakland, CA: Child Care Employee Project.

Whitebook, M., Sakai, L., & Howes, C. (1997). NAEYC accreditation as a strategy for improving child care quality: An assessment. Washington, DC: National Center for the Early Childhood Work Force.

Zellman, G. L., & Johansen, A. S. (1996). The effects of accreditation on care in military child development centers. In S. Bredekamp & B. Willer (Eds.), NAEYC accreditation: A decade of learning and the years ahead (pp. 25-30). Washington, DC: National Association for the Education of Young Children.