Research Platform
Issue Date: July 01, 2015
Published Online: February 09, 2016
Updated: January 01, 2020
Play Assessment and Instrument Development: How to Develop a Reliable and Valid Self-Report Measure
Author Affiliations
  • West Coast University
Article Information
Assessment Development and Testing / Pediatric Evaluation and Intervention / Rehabilitation, Participation, and Disability / Assessment/Measurement
Research Platform   |   July 01, 2015
Play Assessment and Instrument Development: How to Develop a Reliable and Valid Self-Report Measure
American Journal of Occupational Therapy, July 2015, Vol. 69, 6911500183.
American Journal of Occupational Therapy, July 2015, Vol. 69, 6911500183.

Date Presented 4/18/2015

Instrument development procedures were used to construct a reliable and valid self-report measure of play in middle childhood. The results increase the understanding of the occupation of play in its many forms and functions.

SIGNIFICANCE: Play, as a central occupation of childhood, has often been identified as an important but elusive construct in occupational therapy and related disciplines. Occupational therapists often evaluate components of the child’s play performance utilizing a variety of data inclusive of observation, interviews, and related assessments. However, few reliable and valid measures exist on the assessment of play in middle childhood from the perspective of the child. Self-report assessments with adequate psychometric properties are needed to develop meaningful goals, document ongoing treatment effectiveness and client-reported outcomes, as well as contribute to evidence-based practice in our profession. Although observation and interviewing with caregivers and teachers often are necessary to validate and succinctly describe the play performance of children, self-report inventories can provide valuable insight into the child’s perceived style of play.
INNOVATION: Recognizing the need for a self-report play instrument, the lead author created a pilot version of a play instrument as part of a master’s thesis. The four phases of instrument development were used to develop the instrument and included the planning and construction phase followed by the quantitative evaluation and validation phase as outlined by Benson and Clark as well as Thorndike and Thorndike-Christ. There are no other known play instruments in occupational therapy or related disciplines that attempt to obtain reliable and valid estimates of self-reported play activities and play styles of children aged 7 to 11 yr. Having assessments that can determine whether a potential play deficit exists or areas of perceived success in play performance can help clinicians and researchers further define and clarify the concept of play as an important occupation in middle childhood from the perspective of the child.
APPROACH: Research questions included the following: Can a two-part, self-report inventory reliably and validly measure the play activities and play styles of children aged 7 to 11 yr? Can a two-part, self-report inventory discriminate between the play of children without disabilities and those children with known learning and developmental disabilities?
Few, if any, reliable and valid measures exist on the assessment of play in middle childhood from the perspective of the child. Self-report assessments with adequate psychometric properties are needed to develop meaningful goals, document ongoing treatment effectiveness and client-reported outcomes, as well as contribute to evidenced-based practice in our profession. Difficulty in play performance has been documented for children who have learning, emotional, and coordination difficulties. The need to provide reliable and valid measures of play for the child with learning and developmental disabilities was supported by the writing of A. Jean Ayres and subsequent occupational therapy researchers.
METHOD: Instrument development procedures outlined by Benson and Clark as well as Thorndike and Thorndike-Christ were utilized to develop the pilot instrument in this study. We conducted two pilot studies with the instrument using a sample of convenience. The first pilot study was completed as part of a master’s degree and consisted of construction of the instrument and administering it to 78 children with and without learning disabilities (partial results listed below). The second pilot test (the purpose of the present study) was conducted to reconfirm the earlier results with an updated inventory based on a larger sample size. The results of the second pilot test are reported below.
Data were collected from students attending middle and upper-middle income elementary schools during recess as well as during a summer camp in the southern California area. In addition, data were collected from children attending occupational therapy private practice clinics over several therapy sessions.
Children (aged 7 to 11 yr) were recruited over the course of a year (2009 to 2010) from a sample of convenience. Participant criteria included children between the ages of 7 and 11 yr from middle and upper-middle socioeconomic backgrounds. Children with known disabilities either had a mild intellectual deficit or average intelligence on the basis of teacher or parent reports; these children included those with learning disabilities, autism spectrum disorder, Down syndrome, and/or mild mental retardation. Site administrators gave written approval for the study, and letters of introduction describing the study with consent forms were sent home. Those children and parents who agreed to be in the study, met the participant criteria, and returned the signed consent forms were included in the study. The study sample consisted of 124 children: 89 with no disabilities and 35 with disabilities (17 children with learning disabilities, 13 children with autism, 3 children with Down syndrome, 1 child with both Down syndrome and autism, and 1 child with bipolar disorder).
The first part of the instrument measures the types and frequencies of play activities in four domains (i.e., gross motor, fine motor, social group, and solitary play activities) and consists of 40 items. Its second part measures the types and frequencies of play behaviors in four domains (i.e., cooperation, physical coordination, social participation, peer acceptance) and likewise consists of 40 items. It also includes neutral and “lie” items to deter children from answering according to a response set. A Likert-scale format was used to detect frequency of time use for the play activity inventory and perceived performance of play behaviors. For example, “I play ball games” had the following response options: never, about once or twice a year, about once or twice a month, about once or twice a week, and almost every day. An example of a play style inventory item is, “I play by the rules of the game.” The response options were never, hardly ever, sometimes, a lot, and always. Items were written by the primary investigator in this study on the basis of a review of the play literature, field observations, and feedback from children representative of the study sample as well as occupational therapists who served as content experts in the construction phase of the instrument.
For typically developing children, the instrument was administered primarily by the investigator on site at the camp or school. Typically, the inventory was administered in a small group of 2 to 4 children outdoors in a quiet location or in a classroom setting. However, in some cases, up to 30 students were seen when they could complete the inventory with minimal assistance. Children with disabilities recruited from the clinic were administered the instrument by the treating therapist in a 1:1 clinic setting with assistance as needed to read the items. Parents who participated in the study (for an estimate of concurrent validity) received a similar version of the instrument mailed home after the investigator had received the child’s completed survey. Directions were given to answer the survey specific to how they perceive their child’s frequency of play activities and style of play. All of the completed inventories were reviewed for completeness and were scored by the principal investigator in preparation for data analysis.
Item analyses and intercorrelations for each item, subscale, and total test score for both the play activity and the play behavior inventory were calculated with either Spearman rank or Pearson correlations. Internal consistency was measured with Cronbach’s alpha. Test–retest reliability was also measured with Pearson correlation. Parent–child agreement (concurrent validity) was measured with Pearson correlation; parent–child differences and construct validity were assessed with paired-sample t tests. All data were analyzed with SAS Version 9.2.
RESULTS: The results of both pilot studies support the theoretical and empirical research that suggests children with disabilities have a different style of play than children without disabilities. Specifically, on the first pilot test, statistically lower scores were received by children with learning disabilities on the first pilot test of the instrument’s total Play Style Inventory as well as on the Physical Coordination subscale (p = .003) and the Cooperation subscale (p = .017). These data suggest that children with learning disabilities perceived themselves as having significant difficulty in both the physical coordination aspects of the play activity as well as in perceived cooperation during play activities. On the second pilot test, analysis of the data on 124 children indicates that each inventory (Play Activity Inventory and Play Style Inventory) achieved acceptable values of .84 and .79, respectively. These findings are similar to the first pilot test (.87 and .83) and provide an estimate of internal consistency or how well the items on the scale and subscales measure the same content domain.
Test–retest reliability correlations on the second pilot test were likewise significant for both parts of the instrument (.69 and .82, respectively), suggesting that the instrument is a fairly accurate measurement of the child’s reported play performance over time.
No statistically significant differences were found between genders or between the two groups (with or without disabilities) on the self-reported Play Activity Inventory but significant differences were detected on the self-reported Play Style Inventory. That is, both groups perceived engagement at the same frequency for each of the four play activity categories. However, qualitative differences were reported in the performance of these activities. Specifically, statistically significant lower scores were found for the group with disabilities on the total inventory score of the instrument’s Play Style Inventory (p = .002) and on the following subscale scores: Cooperation (p = .01), Peer Acceptance (p = .002), Social Participation (p = .003), and Physical Coordination (p = .003). These results are consistent with the first pilot test and the literature review that identifies many children with learning and developmental disabilities as having difficulty in social and physical interactions during play. Additional separate analyses on the parent–child correlations indicate much stronger correlations for the total Play Style Inventory (r = .96) and on the majority of the subscales, suggesting parents whose child had a disability were, in general, more likely to agree with their child’s rating about perceived performance in play, with the exception of the Cooperation subscale (r = −.08). However, caution is needed in the overall interpretation of these findings due to the small sample size of parent–child surveys.
CONCLUSION: The data from the second pilot test of the instrument reconfirms the results of the first pilot test: that it is a valid and reliable tool for the evaluation of a child’s perceived performance in play during middle childhood. The domains of play activities and play styles assessed by the instrument provide another key facet of information to contribute to the multidimensional assessment of play for the child with a suspected play deficit. Limitations of this research include a small sample size and lack of generalization to children from lower socioeconomic backgrounds or other geographic areas. In addition, although cultural diversity was present in this sample, data were not analyzed to determine whether there were any significant differences.