James S. Roberts

A General Item Response Theory Model for Unfolding Unidimensional Polytomous Responses

The generalized graded unfolding model (GGUM) is developed. This model allows for either binary or graded responses and generalizes previous item response models for unfolding in two useful ways. First, it implements a discrimination parameter that varies across items, which allows items to discriminate among respondents in different ways. Second, the GGUM permits response category threshold parameters to vary across items. Amarginal maximum likelihood algorithm is implemented to estimate GGUM item parameters, whereas person parameters are derived from an expected a posteriori technique. The applicability of the GGUM to common attitude testing situations is illustrated with real data on student attitudes toward abortion.

VALIDITY ISSUES IN THE LIKERT AND THURSTONE APPROACHES TO ATTITUDE MEASUREMENT

This article highlights the theoretical differences between the Likert and Thurstone approaches to attitude measurement and demonstrates how such differences can lead to discrepant attitude estimates for individuals with the most extreme opinions. Both simulated data and real data on attitude toward abortion are used to demonstrate this discrepancy. The results suggest that attitude researchers should, at the very least, devote more attention to the empirical response characteristics of items on a Likert attitude questionnaire. At most, these results suggest that other methods, such as the Thurstone technique or one of its recently developed item response theory counterparts, should be used to derive attitude estimates from disagree-agree responses.

A Unidimensional Item Response Model for Unfolding Responses from a Graded Disagree-Agree Response Scale.

Binary or graded disagree-agree responses to atti tude items are often collected for the purpose of attitude measurement. Although such data are sometimes ana lyzed with cumulative measurement models, recent studies suggest that unfolding models are more appro priate (Roberts, 1995; van Schuur & Kiers, 1994). Ad vances in item response theory (IRT) have led to the development of several parametric unfolding models for binary data (Andrich, 1988; Andrich & Luo, 1993; Hoijtink, 1991); however, IRT models for unfolding graded responses have not been proposed. A parametric IRT model for unfolding either binary or graded re sponses is developed here. The graded unfolding model (GUM) is a generalization of Andrich & Luos hyperbolic cosine model for binary data. A joint maximum likeli hood procedure was implemented to estimate GUM pa rameters, and a subsequent recovery simulation showed that reasonably accurate estimates could be obtained with minimal data demands (e.g., as few as 100 respon dents and 15 to 20 six-category items). The applicability of the GUM to common attitude testing situations is illus trated with real data on student attitudes toward capital punishment. Index terms: attitude measurement, graded unfolding model, hyperbolic cosine model, ideal point process, item response theory, Likert scale, Thur stone scale, unfolding model, unidimensional scaling.

Posttreatment results of combining naltrexone with cognitive-behavior therapy for the treatment of alcoholism.

Naltrexone, an opiate antagonist medication, has been reported to be efficacious in the treatment of alcohol dependence when added to psychosocial treatments. Although the within-treatment efficacy of naltrexone has received primary attention, there has been little published on the outcome of individuals once the medication is discontinued. Animal studies have led to concern regarding a quick rebound to heavy drinking. This report extends the data previously reported by evaluating the outcome in alcoholic subjects during the 14 weeks after a 12-week treatment with naltrexone or placebo in conjunction with cognitive behavioral therapy. Of the 131 subjects evaluated during the treatment phase, 124 (95%) had up to 14 weeks of posttreatment drinking data available for analysis. Measures of craving and blood markers of heavy drinking were also evaluated. By the end of treatment, naltrexone demonstrated significantly greater efficacy than placebo. However, once the medication was discontinued, there was a gradual increase in relapse rates, heavy drinking days, and drinks per drinking day, and fewer days of abstinence were reported. By the end of the 14-week follow-up period, although naltrexone-treated subjects were, on average, still doing better than control subjects, the effectiveness of naltrexone was no longer statistically significant. There was no evidence that naltrexone subjects had an immediate return to heavy alcohol use as suggested in animals. These data suggest that, for a number of alcoholic subjects, continued treatment with naltrexone, or perhaps psychosocial intervention, for longer than 3 months is indicated. Future research should identify which alcohol-dependent individuals may need prolonged treatment to improve treatment success in the long term.

Multiple previous detoxifications are associated with less responsive treatment and heavier drinking during an index outpatient detoxification

Investigators have found a relationship between the number of previous alcohol withdrawals (AWs) and severity of withdrawal. We evaluated patients with multiple previous AWs, as compared to those with 0-1 previous withdrawals, in an outpatient detoxification trial comparing lorazepam (LZ) to carbamazepine (CBZ). A mixed model analysis of covariance was used to analyze Clinical Institute Withdrawal Assessment for Alcohol-Revised (CIWA-Ar) scores as a function of detoxification history (0-1 vs. 2 or more), drug group (CBZ vs. LZ), assessment day, and hours since last drink. The mixed model analysis of covariance (ANCOVA) indicated a significant detoxification history by assessment day interaction (P< or =.03). Least square means associated with this interaction suggested that the CIWA-Ar scores for the multiple detox patients declined more slowly than those with 0-1 previous detoxifications. Patients with multiple detoxes were 150% more likely to experience a heavy drinking day during treatment (P< or =.03). The multiple detox group drank more each drinking day (P=.001) and a greater proportion of this group had early heavy drinking (P=.0002). In the present study, intensity of AW symptoms and early heavy drinking were independent of treatment medications and were more common in patients who had previously undergone multiple treatments for AW.

Time Required for Blood Lead Levels to Decline in Nonchelated Children

Objective: To determine the time for a decline in blood lead to less than 10 μg/ dL in nonchelated children who are enrolled in case management. Study Design: Retrospective analysis of venous blood lead data of lead-poisoned children followed in a case management program designed to decrease lead exposure. Children were excluded if their blood lead had not yet declined to less than 10 μg/dL, if they received chelation therapy, or if they had not received follow-up for more than 15 months. We calculated the time between peak elevation of lead and decline to less than 10 μg/dL. Data were categorized based on the childs peak blood lead and season in which their peak blood lead occurred. Data were analyzed using ANOVA and linear regression. Kaplan-Meier survival analysis was used to describe data in population form. Results: 579 patients were included in the analysis. Blood leads of 25–29, 20–24, 15–19, and 10–14 μg/dL required 24.0, 20.9, 14.3, and 9.2 months, respectively, to decline to less than 10 μg/dL. For continuous data, a linear relationship was described by the following equation: Time (# of months required to achieve a blood lead less than 10 μg/dL) = 0.845 × peak lead; p < 0.0001. Kaplan-Meier curves complement the findings in a population-based fashion. Conclusions: The mean time for blood lead to decline was linearly related to the peak in blood lead. The time for 50% of the blood lead to decline to less than 10 μg/dL was not linear and varied with peak lead.

GGUM2004: A Windows-Based Program to Estimate Parameters in the Generalized Graded Unfolding Model:

The GGUM2004 computer program estimates parameters for a family of unidimensional unfolding item response theory (IRT) models. These unfolding IRT models predict higher item scores to the extent that a respondent is located close to an item on an underlying latent continuum. This prediction is often consistent with responses to traditional Thurstone or Likert attitude questionnaires. Unfolding IRT models can also be used to measure individual differences in preference and certain developmental processes in which behaviors/cognitions occur in unique stages. The GGUM2004 software estimates parameters of the generalized graded unfolding model (GGUM), developed by Roberts, Donoghue, and Laughlin (2000). In addition, the program can estimate parameters in seven other models that are derived by constraining GGUM item parameters in various ways. GGUM2004 estimates item parameters using a marginal maximum likelihood technique, and person parameters are estimated using an expected a posteriori method. The software can analyze responses to a maximum of 100 items, in which each item has up to 10 graded response categories. The program will allow for a maximum of 2,000 respondents. The GGUM2004 program extends the capabilities of its predecessor (GGUM2000; Roberts, 2001) in many useful ways:

A double‐blind, placebo‐controlled trial of sertraline in depressed adolescent alcoholics: a pilot study

In order to preliminarily evaluate the efficacy, safety and tolerability of the serotonin reuptake inhibitor, sertraline, in the treatment of adolescents with a primary depressive disorder and a comorbid alcohol use disorder, a 12‐week double‐blind, placebo‐controlled trial of sertraline plus cognitive behavior group therapy was conducted. Subjects were 10 outpatient treatment‐seeking adolescents. Baseline assessment included the K‐SADS, HAM‐D, SCID, and the Time‐Line Follow‐Back. The HAM‐D and the Time‐Line Follow‐Back were performed weekly thereafter. Both groups showed a significant reduction in depression scores with an average reduction between baseline and endpoint HAM‐D score of −9·8 (F(1,8)=26·14, p⩽0·001), although there were no significant group differences. There was an overall reduction in Percent Days Drinking (PDD); (F(1,8)=8·90, p<0·02) and in Drinks Per Drinking Day (DDD); (F(1,8)=20·48, p<0·002), however, there were no group differences. Depression responders tended to have higher baseline PDD than non‐responders (F(1,8)=3·9, p=0·08) and change in HAM‐D scores tended to correlate with change in PDD (r=0·57, p=0·09). Our data support that sertraline is safe and well tolerated in the treatment of adolescents with depression and alcohol dependence. Small sample size and cognitive behavior group therapy given to all subjects may limit the lack of group differences. Copyright

Incorporating Environmental Health into Pediatric Medical and Nursing Education

Pediatric medical and nursing education currently lacks the environmental health content necessary to appropriately prepare pediatric health care professionals to prevent, recognize, manage, and treat environmental-exposure–related disease. Leading health institutions have recognized the need for improvements in health professionals’ environmental health education. Parents are seeking answers about the impact of environmental toxicants on their children. Given the biologic, psychological, and social differences between children and adults, there is a need for environmental health education specific to children. The National Environmental Education and Training Foundation, in partnership with the Children’s Environmental Health Network, created two working groups, one with expertise in medical education and one with expertise in nursing education. The working groups reviewed the transition from undergraduate student to professional to assess where in those processes pediatric environmental health could be emphasized. The medical education working group recommended increasing education about children’s environmental health in the medical school curricula, in residency training, and in continuing medical education. The group also recommended the expansion of fellowship training in children’s environmental health. Similarly, the nursing working group recommended increasing children’s environmental health content at the undergraduate, graduate, and continuing nursing education levels. Working groups also identified the key medical and nursing organizations that would be important in leveraging these changes. A concerted effort to prioritize pediatric environmental health by governmental organizations and foundations is essential in providing the resources and expertise to set policy and provide the tools for teaching pediatric environmental health to health care providers.

Characteristics of MML/EAP Parameter Estimates in the Generalized Graded Unfolding Model

The generalized graded unfolding model (GGUM) is a very general parametric, unidimensional item response theory model for unfolding either binary or polytomous responses to test items. Roberts, Donoghue, and Laughlin have described a marginal maximum likelihood (MML) approach to estimate item parameters in the GGUM along with an expected a posteriori (EAP) method to estimate person parameters. This article examines the data demands required to produce accurate parameter estimates using these techniques under ideal conditions. It also examines the robustness of parameter estimates under nonideal conditions, in which there are inconsistencies between the prior distribution of person parameters that must be speci.ed when using either the MML or EAP approaches and the true distribution of person parameters. Results from two simulation studies show that accurate item parameter estimates can generally be obtained with approximately 750 to 1,000 respondents. Similarly, responses to approximately 15 to 20, equally spaced, six-category items can yield accurate EAP estimates of person parameters under static testing conditions. The results also suggest that MML item parameter estimates are quite robust to discrepancies between the prior and true distributions of person parameters. EAP parameter estimates are also fairly robust as long as the item response patterns in question are not too extreme. Finally, 20 quadrature points are generally sufficient to integrate over the prior distribution in both the MML and EAP methods when test and sample characteristics are like those simulated. Thus, the MML/EAP approach to parameter estimation in the general graded unfolding model can produce accurate estimates in an ef.cient manner even when there is uncertainty about the true distribution of person parameters.