Two-Group Classification Using Neural Networks*

Artificial neural networks are new methods for classification. We investigate two important issues in building neural network models; network architecture and size of training samples. Experiments were designed and carried out on two-group classification problems to find answers to these model building questions. The first experiment deals with selection of architecture and sample size for different classification problems. Results show that choice of architecture and choice of sample size depend on the objective: to maximize the classification rate of training samples, or to maximize the generalizability of neural networks. The second experiment compares neural network models with classical models such as linear discriminant analysis and quadratic discriminant analysis, and nonparametric methods such as k-nearest-neighbor and linear programming. Results show that neural networks are comparable to, if not better than, these other methods in terms of classification rates in the training samples but not in the test samples.     

The Older Worker in the Low Countries

Abstract. The paper deals with present, demographically-based problems in highly-developed labour markets, with special reference to the Belgian and Dutch experience. Both sluggish inflow by young cohorts and early retirement by senior workers are analysed. The first of these phenomena can be explained by the falling birth rate and increasing educational levels; the second phenomenon is the result of technological changes in the labour market and a social security system which offers attractive opportunities for early outflow. For companies, this trend translates into a loss of experience and a loss of corporate culture. On top of this, the social security system is coming under heavy pressure and scarcity symptoms are likely to develop on the supply side of the labour market.     

The Moderating Effect of Living with a Child Before Incarceration on Postrelease Outcomes Related to a Prison-Based Parent Management Training Program

ABSTRACT

With the rapid growth in incarceration in the United States over the past few decades came dramatic growth in the number of the incarcerated parents with at least one minor child. Parental incarceration places extra stresses and strains on families and children. Almost all of those incarcerated will eventually be released. However, the majority of those released from prison will be rearrested within a year. Finding interventions that can decrease the likelihood of returning to crime and to incarceration are of utmost importance. Using a social bond theory framework, the authors examine the moderating effect of living with a child before incarceration on program outcomes related to a prison-based parent management training program. A significant effect was found. Implications for these findings are discussed for future research and practice.     

Variations in the Life Histories of Incarcerated Parents by Race and Ethnicity: Implications for Service Provision

ABSTRACT

Incarcerated parents have complex life histories that often remain unresolved during incarceration, can continue to create barriers to prosocial success on release, and present similar intergenerational challenges for their children. This study examines the life histories of incarcerated fathers and mothers from the Pacific Northwest and how their experiences vary based on race and ethnicity. Five areas examined were exposure to trauma, child welfare involvement, mental health and substance abuse problems, juvenile justice and adult criminal justice involvement, and intergenerational criminal justice involvement. The sample comprised 359 incarcerated parents, and their racial/ethnic composition was 59% White, 14% African American, 11% multiracial, 8% Native American, and 7% Latino. Few differences were found across racial and ethnic groups. Mothers appeared more similar to each other across groups than fathers. Results illustrated similarities yet some surprising differences with national trends on key study variables. Implications for future research and intervention and prevention are discussed.     

‘Keeping families and children in mind’: an evaluation of a web‐based workforce resource

This study outlines pilot evaluation data of the web‐based training resource ‘Keeping Families and Children in Mind’, designed for clinicians who work with families where a parent has a mental illness. The resource was developed from scoping existing workforce packages and in consultation with consumers, carers, researchers and mental‐health clinicians. Preliminary evaluation data were collected from an urban and a rural site in Australia via focus group interviews and pre‐ and post‐training questionnaires to ascertain the experiences of those who participated in the training. Additionally, training facilitators were invited to maintain journals in order to identify planning and implementation issues when using the resource. Post‐training, participants emphasized the need to work collaboratively with others, as well as the importance of acknowledging and working with the family members of consumers, especially children. Also, participants reported positive changes in knowledge, skill and confidence when working with families affected by parental mental illness. Facilitators highlighted technology issues and the need to work interactively with participants when using the resource. Recommendations regarding policy and future research conclude this paper.     

Temporary Work,Underemployment and Workplace Accommodations: Relationship to Well‐being for Workers with Disabilities

This study examines whether employment status and workplace accommodations are associated with perceived well‐being among workers with disabilities. Data from the 2006 Participation and Activity Limitation Survey conducted by Statistics Canada were used to test the relationship between employment status, receipt of workplace accommodations and well‐being. Findings indicated that fully utilized permanent employees showed greater life satisfaction and less perceived disability‐related discrimination than either temporary workers or permanent workers who were underemployed. These findings support the theory that inadequate employment is associated with deleterious effects on employee well‐being due to inferior need fulfilment and reduced social status. Workplace accommodations were associated with higher levels of well‐being for all workers with disabilities and helped to mitigate the negative effects of temporary status and underemployment. These findings supported the theoretical extension of main effect and buffering models of workplace stress to the prediction of perceived workplace discrimination.     

Determinants of Migrant Career Success: A Study of Recent Skilled Migrants in Australia

Australia has been aggressively pursuing skilled migrants to sustain its population and foster economic growth. However, many skilled migrants experience a downward career move upon migration to Australia. Based on a survey of recent skilled migrants, this study investigates how individual (age, years of settlement, qualifications), national/societal (citizenship and settlement), and organization-level (climate of inclusion) factors influence their career success. Overall, we found that: (1) age at migration matters more than length of settlement in predicting skilled migrant career success; (2) citizenship uptake and living in a neighbourhood with a greater number of families from the same country of origin facilitate post-migration career success; and (3) perceptions of one's social/informal networks in the workplace – a dimension of perceived organizational climate of inclusion – also have a positive impact on migrant career outcomes.     

A Cross-Validation Analysis of Neural Network Out-of-Sample Performance in Exchange Rate Forecasting

Econometric methods used in foreign exchange rate forecasting have produced inferior out-of-sample results compared to a random walk model. Applications of neural networks have shown mixed findings. In this paper, we investigate the potentials of neural network models by employing two cross-validation schemes. The effects of different in-sample time periods and sample sizes are examined. Out-of-sample performance evaluated with four criteria across three forecasting horizons shows that neural networks are a more robust forecasting method than the random walk model. Moreover, neural network predictions are quite accurate even when the sample size is relatively small.