This article reviews the evidence on the extent to which experts' perceptions of item difficulties, captured using comparative judgement, can predict empirical item difficulties. This evidence is drawn from existing published studies on this topic and also from statistical analysis of data held by Cambridge Assessment. Having reviewed the…

Although violence risk assessment knowledge and practice has advanced over the past few decades, it remains practically difficult to decide which measures clinicians should use to assess and make decisions about the violence potential of individuals on an ongoing basis, particularly in the short to medium term. Within this context, this study…

This study reports on the validation of the language domain subtest of a developmental assessment scale for Cantonese Chinese preschool children. Three hundred and seventy eight multi-stage randomly selected children between 3;4 and 6;3 years of age were tested on the 104-item subtest. Fifty-four of these children, spreading across three age…

First-year students are still failing at an alarming rate. This is an international issue that universities face and there is currently no clear indication of the cause of the problem as universities move from being elite to providing mass education. This article examines the possible correlation between students' high school performance and…

Background: The clinical diagnosis of ADHD is time-consuming and error-prone. Secondary care referral results in long waiting times, but primary care staff may not provide reliable diagnoses. The Development And Well-Being Assessment (DAWBA) is a standardised assessment for common child mental health problems, including attention…

The Texas Projection Measure (TPM) has grown out of the state's need to meet the requirements of No Child Left Behind (NCLB). An examination of the state's method of predicting 8th grade mathematics scores reveals that several factors have been ignored in the process of developing the model, including assumptions in its underlying statistical…

Eight indexes for item analyses of small samples of test scores are described. Each index and a corresponding statistical significance test is defined, and tables of critical values of the index and associated right-tail probabilities are given. A computer program for calculating all indexes or any subsets is described. (JD)

How the validity of gain scores varies with the standard deviations of pretest and posttest scores and the correlation between the two are analyzed. Earlier findings that under realistic testing conditions difference scores can have excellent predictive value are supported. Conditions under which gain scores have optimum validity are specified.…

Where do those in schools start sorting the wheat from the chaff, genuine growth potions offering long-term improvement from the elixirs, short-term opportunism and/or unrealistic expectations? The current and growing emphasis on evidence informed policy and practice is as good a place as any. The purpose of this article is to take up the issues…

A new statistical model based for identifying potential major donors was developed from a portion (n=53,276) of the Northwestern University (Illinois) alumni database. The model, utilizing logit analysis, found that the most critical measure was individuals' past giving records. Results indicate prospects with a low past giving level will rarely…

An evaluation instrument, DataPac, was developed and pilot-tested to assess the efficacy of corporate health promotion programs. This article reports results of a study to evaluate the DataPac questionnaire for construct validity, criterion validity, and stability reliability. Implications for worksite health programs are discussed. (IAH)

Data from the Canadian Test of English for Scholars and Trainees (CanTEST) are compared to data from the Test of English as a Foreign Language (TOEFL) to establish CanTEST as a valid admissions tool for English-as-a-Second Language college applicants. Data are taken from four groups of examinees who took both tests. (eight references) (LR)