A developmental screening test : development, validation, and implementation
In 1975 the North Carolina General Assembly appropriated funds to develop and implement a Statewide Prekindergarten Screening Program (SPSP) for the 1975-77 biennium. The major purpose of the program was to identify developmental disabilities of four-year-olds during the year before their entrance into kindergarten. Phase I screening involved the administration of the North Carolina Psychoeducational Test (PET). The PET was the most important factor in determining whether a child should be referred for Phase II screening which involved the administration of the McCarthy Scales of Children's Abilities (MSCA). The PET's ultimate purpose was to identify children with the most serious deficiencies. However, studies indicated that the PET was identifying a large percentage of children as needing Phase II screening when, in fact, they did not. More specifically, approximately 50 percent of the referrals were false positives. A more discriminating diagnosis in Phase I screening would not only save manpower and administrative costs but also would save parents the time required for additional follow-up screening.
The purpose of this study was to develop a reliable and valid screening instrument which, when compared with the North Carolina Psychoeducational Test, would be a more sensitive instrument and would reduce the number of four-year-old children inappropriately referred for Phase II screening. The study required the selection of test items, development of test format, piloting of test, revision of the test, and
iv
statistical analyses based on its administration to a sample of children from northwestern North Carolina.
A comprehensive survey of developmental and preschool tests was conducted to determine which test items might serve as good indicators of developmental functioning for five age groups. Sixty items were chosen to be included on the screening test. These items were selected from twelve preschool developmental and screening tests. The items were divided into six functional areas: : visual memory, auditory memory, gross motor development, visual motor performance, language development, and concept development. In each of these areas there were ten items, two for each of the five age groups.
Forty-six children were tested in the pilot study. The children ranged in ages from two and one-half years to five and one-half years. An attempt was made to include children representative of both sexes, black and white children, and children from a wide range of socioeconomic groups. The percentage of children passing each item was determined. . Items which were not passed by 75 percent of the children were examined and revised or replaced with other items. The final revision of the instrument included sixty items, ten in each of the developmental areas.
Eighty-six children were involved in the norming sample. In each age group the percentage of children who passed each item was calculated. . The level of item difficulty was determined by computing the relationship between each item and the total score using a point biserial. Consideration was given to the percentage of children passing
vi each item according to race, sex, and occupational level of the child's mother and father. The sample was compared to the general population in northwestern North Carolina in regard to sex, race, and occupational groups using a chi-square test.
Determination of the reliability of scoring by the examiners was made using a Fisher Exact Test and two measures of association-the phi coefficient and the contingency coefficient.
The validity of the screening instrument was computed using the and the split-halves method. Thirty-one children were selected for the validity study. They were tested with the screening instrument and the McCarthy Scales of Children's Abilities. A chi-square, a phi coefficient and the contingency coefficient were used to determine the correlation between each of the six functional areas of the screening instrument and the five functional areas of the MSCA. The correlation between the total scores of the two instruments was also computed.
The comparative study involved the comparison of scores on the screening instrument with scores on the PET. The purpose of this comparison was to determine the extent the screening instrument and the PET correlated with the MSCA. A sample of sixteen children was administered the screening instrument, the PET, and the MSCA. Scores were compared to determine which test made a more accurate decision in referring children for further diagnostic testing.
The results of the norming study indicated that nine out of the sixty items were not passed by 75 percent of the children. Nine of
vii these sixty items had negative point biserials. There was no significant difference in the percentage of blacks and whites passing the items except on one item. There was no significant difference in the percentage of males and females passing the items except for one item in one age group. Only minor differences were calculated in the way the children answered the items when categorized according to occupational group of both father and mother.
No relationship between the examiners and their ratings was found, indicating that the ratings of the examiners were consistent. Using the a reliability of .938 was computed. The split halves statistical method computed a R of .873l which when converted to a reliability coefficient was .932.
A chi-square value of 6.09524 was computed when the total score of the screening test was compared with the total score of the MSCA. This was significant at the .Ol36 level.
The results of the comparative study indicated that the screening instrument correlated significantly with the MSCA at the .00l level. The PET correlated significantly at the .0l level. An individual analysis of the sixteen test administrations indicated that the screening test was more sensitive than the PET. Decisions to refer or not to refer were correct in more cases using the screening test rather than the PET.
To summarize, the major conclusions of this study were:
1. The majority (fifty-one out of sixty) of the items on the screening instrument were passed by 75 percent of the population for which they were designed.
2. The screening test had a high interrater reliability.
3. , The screening instrument when compared to a valid and reliable instrument was also found to be a valid instrument.
4. The screening test, when compared with the PET, appeared to be a more sensitive instrument, allowing examiners to make decisions regarding referral for further testing with a greater degree of confidence.
Thesis78b.H342.pdf
9.2 MB
Adobe PDF
6aac11f8d0594d1935a9e2054c5c72e8