Table of Contents | Previous | Next |
CHILD CARE QUALITY: PROCEDURAL QUALITY
NATIONAL CHILD CARE STAFFING STUDY
Measure: Howes and Stewart (1987) scale of adult play with child, as used in the National Child Care Staffing Study
Note: The National Child Care Staffing Study (NCCSS) is a longitudinal study of child care centers conducted in 1988, 1992, and 1997. This description of the Howes and Stewart (1987) adult play scale derives from its use in the Atlanta sample in the original (1988) NCCSS study.
Source
The 1988 NCCSS was coordinated by the Child Care Employee Project staff and funded by a consortium of foundations, including the Carnegie Corporation of New York, Ford Foundation, the Foundation for Child Development, the A.L. Mailman Family Foundation, and the Spunk Fund, Inc. (Whitebook, Howes, & Phillips, 1990, p. ii). Marcy Whitebook, Carollee Howes, and Deborah Phillips, the principal investigators of the NCCSS, worked (at the time of the 1988 study) at the Child Care Employee Project, the University of California at Los Angeles, and the University of Virginia, respectively. The sponsor of the study, the Child Care Employee Project, changed its name to the Center for the Child Care Workforce (CCW) in 1997. (CCW was known as the National Center for the Early Childhood Work Force between 1994 and 1997.) In November 2002, CCW became a program within the American Federation of Teachers Educational Foundation.
One of the authors of the 1988 NCCSS study, Carollee Howes, created—along with Phyllis Stewart—the adult play scale. The scale was first used in the Howes and Stewart (1987) study of the relationship between 1) children’s play with adults, toys, and peers and 2) child care quality and family characteristics. “This five-point scale has predicted children’s developmental outcomes (Howes & Stewart, 1987)” (Whitebook et al., 1990, p. 26). In discussing the creation of the adult play scale, Howes and Stewart observed that “the adult play scale was developed for [their 1987 study] on the basis of individually coded measures used by Rubenstein and Howes (1979) to describe child-care caregiver behavior” (Howes & Stewart, 1987, p. 425). For more information on the original creation of the scale of adult play with child, see Howes & Stewart, 1987.
According to Dr. Howes (personal communication, August 14, 2003), the Howes and Stewart adult play scale was used in FACES observations. It has also appeared in a number of other major studies, including the Cost Quality and Outcome Study and (in a modified version) the Multi-State Study.
Population Assessed
The original study sample consisted of 227 child care centers in five metropolitan areas; within these 227 centers, researchers observed 643 classrooms and interviewed 1,309 teaching staff (including both teachers and assistant teachers). The NCCSS only focused on center-based programs that served children up through 5 years old, operated at least 11 months a year for a minimum of 6 hours a day, served a minimum of 15 children, and employed no less than six staff members. “In summary, there is some potential for bias in the sample given the higher participation rates for non-profit than for-profit centers, centers serving low-income families, and centers that may offer somewhat higher quality care than is typical in the Study sites [metropolitan areas]. However, as a result of the stratified, replacement sampling strategy, the final sample of centers closely matches the distribution of centers across Census tracts and urban and suburban residential areas” (Whitebook et al., 1990, p. 19).
Although the study looked at five metropolitan areas, the adult play scale (along with various assessments of children’s developmental outcomes) was used only in the Atlanta sample. In Atlanta as a whole, Blacks were the largest minority group. About one-third of teaching staff in the 1988 sample (across all five metropolitan areas) belonged to racial/ethnic minorities, and in all metropolitan areas the percentage of members of minority populations was larger in the teaching staff than in the area as a whole. “In Atlanta, two children, preferably a girl and a boy, were randomly selected from each target classroom to be assessed. Two hundred and fifty-five children constituted the child sample: 92 infants, 57 toddlers, and 106 preschoolers” (Whitebook et al., 1990, p. 20).
The demographic characteristics of the Atlanta teachers whose play with children was assessed were not reported in the main report for the original (1988) study. See pages 32–33 of Whitebook et al., 1990, for demographic information on the sample of teachers as a whole.
Periodicity
Data were collected between February and August 1988 for the original study as a whole. The 1992 and 1997 follow-ups did not include the adult play scale.
Subscales/Components
The adult play scale contains no subscales or components. As the description of the scale below should make clear, the measure contains only one item, a continuum on which the nature of adult’s play with children is recorded at regular intervals. In the NCCSS, the adult play scale was used as a rating of teacher-child interaction, itself a construct that measures classroom quality. Other measures of teacher-child interaction used in the NCCSS were the appropriate caregiving subscale of the ECERS and ITERS and the Arnett scale of teacher sensitivity. The adult play scale is referred in the NCCSS report as “the Howes and Stewart (1987) measure of the level of adult involvement with children” (Whitebook, Howes, & Phillips, 1990, pp. 25–26).
Procedures for Administration
In the 1988 NCCSS, trained and experienced “research assistants spent a total of at least two hours in each classroom assessing quality [via multiple measures]. In most cases, each classroom was visited on more than one day; in all cases, the time a classroom was observed covered both morning and afternoon activities” (Whitebook et al., 1990, p. 23). Classroom observations took place before teacher interviews. “In Atlanta, one research assistant additionally observed each target child’s interaction with his or her teaching staff for six five-minute blocks evenly distributed over a two-hour period. Interactions were rated every 20 seconds using the Howes and Stewart (1987) measure of the level of adult involvement with children” (Whitebook et al., 1990, p. 25-26). A different research assistant was used to administer the Arnett scale of teacher sensitivity. For more information on the use of observation protocols of child care quality in the NCCSS, see page 21 and 23 to 26 of Whitebook et al., 1990.
Carollee Howes stressed that the adult play scale is “an observational instrument that requires considerable training” (personal communication, August 14, 2003).
Psychometrics/Data Quality
In the 1988 NCCSS, “inter-rater reliabilities were established to a criterion of 80 percent agreement for all observational measures prior to data collection…At mid-point, within-site reliabilities (based on 5% of the center sample) exceeded 90 percent” (Whitebook et al., 1990, p. 21).
“Kappa inter-observer reliability scores for the [Howes & Stewart] adult involvement measure were .92” (Whitebook, Howes et al., 1990, p. 26).
The adult play scale has been used in a number of other major studies, including the Cost Quality and Outcome Study and FACES; a modified version was used in the Multi-State Study. It was created for the Howes & Stewart study, in which it was used to assess adult play with 55 toddlers (30 female, 25 male, all between 11 and 30 months old), each child attending a different family day-care home full time. For the Howes & Stewart (1987) study, “the test-retest reliability of the adult play scale was assessed by observing four children for 1 week after their initial observations. Test-retest reliability was .85” (Howes & Stewart, 1987, p. 425).
Languages Available
Other than English, information about the languages in which this measure is available is not readily available.
Items Included
The following description of the adult play scale comes from its original source, Howes & Stewart (1987):
“Adult play with child. The caregiver’s behavior with the child was coded; the child’s response or dyadic involvement was not coded. Adult play was rated as ignores (0) if the adult ignored the child; as routine (1) if the caregiver touched the child for changing or other routine care giving but made no verbal responses to child; as minimal (2) if the care giver touched the child only for necessary discipline, to move a child away from another, to answers direct requests for help, or to give verbal directives with no reply encouraged; as simple (3) if the caregiver answered the child’s verbal bids but did not elaborate or used some unnecessary positive physical contact; as elaborated (4) if the caregiver engaged in some positive physical gestures, maintained a close proximity to the child, acknowledged the child’s statements and responded but did not restate the child’s statement, sat with the child during play, or suggested materials; as intense (5) if the caregiver hugged or held the child, restated the child’s statements, engaged the child in conversation, played interactively with the child, or sat and ate with the child and provided a social atmosphere. The mean level of play with adults was calculated by weighting the frequency of play at each level by the scale point, summing the weighted frequencies, and dividing by the total frequency of play” (Howes & Stewart, 1987, p. 425).
References and Source Documents
Howes, C., & Stewart, P. (1987). Child’s play with adults, toys, and peers: An examination of family and child-care influences. Developmental Psychology, 23, 423-430.
Rubenstein, J., & Howes, C. (1979). Caregiving and infant behavior in day care and homes. Developmental Psychology, 15, 1-24.
Whitebook, M., Howes, C., & Phillips, D. (1990). Who cares? Child care teachers and the quality of care in America. Final report: National Child Care Staffing Study. Oakland, CA: Child Care Employee Project.
| Table of Contents | Previous | Next |