How accurately deltas (statistics measuring the difficulty of items) established by pre-test populations reflect deltas obtained from final form populations, and the consequent utility of pre-test deltas for constructing final (operational test) forms to meet developed statistical specifications were studied. Data were examined from five subject…

The third section of this two-volume report examines the utility of the model developed in section I for application in a classroom testing situation. The model was applied in arithmetic classes through a weekly testing program. The teacher specified the generic task being taught, tests were constructed using a table of random numbers to select…

The High School and Beyond (HS&B) Local Labor Market Indicators machine-readable data file (MRDF) provides various measures of the conditions of the local labor market in the geographic areas in which the 1,015 HS&B sample high schools are located. The data are presented in such a way as not to reveal the identities of the participating…

A review of the literature about practices in evaluating art education programs led to the design of the new evaluation model described in this paper. The literature review revealed that a growing number of authors advocate qualitative methods, and that a debate exists regarding the combination of qualitative and quantitative data gathering and…

Three statistical problems related to criterion-referenced testing are investigated: estimation of the likelihood of a false-positive or false-negative decision with a mastery test, estimation of true scores in the Compound Binomial Error Model, and comparison of the examinees to a control. Two methods for estimating the likelihood of…

Five data collection techniques and two sampling methods were examined in terms of costs, rates of return, quality of the data collected, and degree of representativeness in assessing needs and interests of public television viewers in rural areas. Data collection techniques were (1) mail only, (2) mail incentive, (3) telephone only, (4) mail and…

Two sampling designs were compared for the planned 1978 national longitudinal survey of high school seniors with respect to statistical efficiency and cost. The 1972 survey used a stratified two-stage sample of high schools and seniors within schools. In order to minimize interviewer travel costs, an alternate sampling design was proposed,…

A logistic model was used to generate data to serve as a proxy for an immediate retest from item responses to a fourth grade standardized reading comprehension test of 45 items. Assuming that the actual test may be considered a pretest and the proxy data may be considered a retest, the effect of regression was investigated using a percentage of…

A stratified two-stage cluster sample design was chosen for the Texas Assessment Project and is described in this document. The sample was designed so as to achieve a chosen level of precision in item analysis data within each of the 20 geographic regions chosen for analysis. Four sampling frames were developed: 9-year-olds; sixth graders;…

A Monte Carlo simulation was used to ascertain the degree of inflation that can occur in regression estimates when samples contain randomly occurring instances of a pattern among correlations called cooperative suppression. Ten thousand samples of scores on three variables were randomly drawn from a population in which the correlations among the…

Sixteen item statistics recommended for use in the development of criterion-referenced tests were evaluated. There were two major criteria: (1) practicability in terms of ease of computation and interpretation and (2) meaningfulness in the context of the development process. Most of the statistics were based on a comparison of performance changes…

The state of the art in observational studies is discussed, with emphasis on explanatory research--studies of program effectiveness in which an attempt is made to explain the reasons for the results as well as describing the results. In addition to the collection of data and presentation of statistical facts, the explanatory study involves the…

The average design effects for statistics estimated from the base-year National Longitudinal Study data are presented. Attempts to partition the effects into those due to stratification, clustering, and unequal weighting are discussed. The expected increases in subpopulation sample sizes due to oversampling are calculated and compared with the…

Data from the verbal portion of the College Entrance Examination Board Scholastic Aptitude Tests were used in an experimental test of the accuracy of equating for a variety of models in three categories: linear equating, equipercentile equating, and item characteristic curve equating. The models were tested for both mean squared error and bias.…

By means of computer simulation of test responses, the reliability of item analysis data and the accuracy of equating were examined for hypothetical samples of 250, 500, 1000, and 2000 subjects for two tests with 20 equating items plus 60 additional items on the same scale. Birnbaum's three-parameter logistic model was used for the simulation. The…