Gardênia da Silva Abbad

Regressão múltipla stepwise e hierárquica em Psicologia Organizacional: aplicações, problemas e soluções

Este artigo discute algumas aplicacoes das tecnicas de analise de regressao multipla stepwise e hierarquica, as quais sao muito utilizadas em pesquisas da area de Psicologia Organizacional. Sao discutidas algumas estrategias de identificacao e de solucao de problemas relativos a ocorrencia de erros do Tipo I e II e aos fenomenos de supressao, complementaridade e redundância nas equacoes de regressao multipla. Sao apresentados alguns exemplos de pesquisas nas quais esses padroes de associacao entre variaveis estiveram presentes e descritas as estrategias utilizadas pelos pesquisadores para interpreta-los. Sao discutidas as aplicacoes dessas analises no estudo de interacao entre variaveis e na realizacao de testes para avaliacao da linearidade do relacionamento entre variaveis. Finalmente, sao apresentadas sugestoes para lidar com as limitacoes das analises de regressao multipla (stepwise e hierarquica).

Anàlise fatorial confirmatória da escala de impacto do treinamento no trabalho

Training impact at work is the main indicator of training effectiveness at the individual level. In the training evaluation models, impact is a measure of behavior change on the job. The conceptual delimitation of similar constructs, as transfer of learning, is important to research. The main objective of this study were to test the empirical structure of the training impact at works scale with a confirmatory factor analysis and the adequacy of the construct. In this study, participated 2.966 workers of seven organizations, divided in three sub-samples. The results indicated model modifications, and the analysis was corroborated on the three sub-samples. Thus, the measurement should be improved. However, the conceptual proposition of training impact at work was corroborated.

Estratégias de aprendizagem em curso a distância: validação de uma escala

Pesquisas em avaliacao de treinamentos a distância sao incipientes. O artigo pretende contribuir com a area ao construir e validar estatisticamente o instrumento Estrategias de Aprendizagem. O curso-alvo foi oferecido pelo SEBRAE para alunos do Brasil inteiro. A coleta de dados foi realizada ao final do curso por meio de questionario digitalizado que mede a frequencia com que os participantes utilizam as estrategias ao longo do curso. Foram realizadas analises estatisticas de Componentes Principais (PC) e Fatoriais (PAF) e de consistencia interna (alfa de Cronbach). Apos a validacao estatistica, o instrumento apresentou sete fatores: Controle da emocao, 5 itens (a=0,89); Busca de ajuda interpessoal, 6 itens (a=0,89); Repeticao e organizacao, 5 itens (a=0,77); Controle da motivacao, 4 itens (a=0,84); Elaboracao, 3 itens (a=0,83); Busca de ajuda ao material didatico, 2 itens (a=0,75); Monitoramento da compreensao, 3 itens (a=0,82). Os resultados indicam que a escala e estatisticamente valida e confiavel.

Panorama das pesquisas em educação a distância no Brasil

Investment in training, development and education is a competitive advantage for organizations. In this context, distance learning education (EAD) is regarded as a viable possibility in the construction of mechanisms which encourage continuous learning and acquirement of skills throughout life. The present study is within the set of reflections. Its goal is to briefly describe the context and focus of research on distance training in Brazil, from 2003 to 2009, based on literature and data review at the Brazilian Statistical Yearbook of Education Open and Distance (ABRAED). The production review of knowledge in distance education in Brazil shows the need for more research in corporate education, professional training, and evaluation. The pace of research is incompatible with the rapid growth of this type of learning in organizations. Other considerations are taken and a research agenda is proposed.

Impacto do treinamento no trabalho: investigando variáveis motivacionais e organizacionais como suas preditoras

This research had as a main objective to test a model of Training Impact Evaluation at Work. Three stages were accomplished: (1) the construction and validation of a scale of Training Instrumental Value; (2) the improvement and validation of the scales of Motivation to Learn, Motivation to Transfer and Support Before Training; and (3) the investigation of the relationship among the characteristics of the training participants (Motivation to Learn, Motivation to Transfer and Training Instrumental Value), Reactions and Support to the Transfer of Training; as independent variables, and Impact of Training at Work, as dependent variable. The answers of the participants of 28 courses to the scales were submitted to exploratory data analysis, principal components analysis, factor analysis, reliability analysis and standard multiple regression. The results of the stages 1 and 2 showed valid and accurate one factor scales. In the stage 3, the standard multiple regression analysis showed only three explained variable of self-evaluation of Impact of Training at Work: Manager and Social Support, Training Instrumental Value and Reaction of the Participant to the Instructor. Some theoretical and practical implications of those discoveries are discussed.

EVASÃO EM CURSO VIA INTERNET: EXPLORANDO VARIÁVEIS EXPLICATIVAS

There are few studies on the evaluation of distance courses, especially dropout rates, in spite of their importance as a tool to develop human competences for the work. The purpose of the present paper is to identify variables that explain dropout rates in a free distance course offered through Internet at national level. In this study we consider dropout when the student quits definitely the course at any moment. The antecedent variables included demographic data and the use of electronic resources. We choose the analysis of logistic regression according to the categorical nature of the criterion variable (those who finished versus those who did not finish the course). The results show that students who did not finish the course are those who are not inclined to use the electronic resources of interactivity (notice boards, chats, exchange of electronic messages). The theoretical and methodological implications of the conclusions are discussed, as well as the need to include motivational, cognitive and contextual variables, and their possible interactions, in multivariate models explaining dropout.There are few studies on the evaluation of distance courses, especially dropout rates, in spite of their importance as a tool to develop human competences for the work. The purpose of the present paper is to identify variables that explain dropout rates in a free distance course offered through Internet at national level. In this study we consider dropout when the student quits definitely the course at any moment. The antecedent variables included demographic data and the use of electronic resources. We choose the analysis of logistic regression according to the categorical nature of the criterion variable (those who finished versus those who did not finish the course). The results show that students who did not finish the course are those who are not inclined to use the electronic resources of interactivity (notice boards, chats, exchange of electronic messages). The theoretical and methodological implications of the conclusions are discussed, as well as the need to include motivational, cognitive and contextual variables, and their possible interactions, in multivariate models explaining dropout.

Treinamento: análise do relacionamento da avaliação nos níveis de reação, aprendizagem e impacto no trabalho

The present paper aims to report the validation of training reaction scales, as well as to report the results from an analysis of the relationship among the three most common levels: reaction, learning and impact of training on work. The scientific literature has shown that the relationship between these levels has not always been significant, or toward the predicted direction. Despite their relevance, valid and reliable reaction measures are very rare. The used reaction questionnaire contains 39 items. It was applied in a sample of 229 training events. Approximately 3,700 valid questionnaires were subjected to factorial analysis (PAF - direct oblimin) and internal consistency analysis (Cronbachs alpha). Two scales with high level of reliabily were obtained: one of them concerning training reactions and the another concerning the instructor performance. Finally, very strong relationships were found between the reaction and impact levels, but poor relationships between the reaction and the learning level. The results are discussed in terms of their practical and theoretical implications for training evaluations.

Autoconceito profissional e suporte à transferência e impacto do treinamento no trabalho

This research investigated the relationship between professional self-concept, support to transfer of training and training impact at work. The samples of participants (model 1 self-assessment as the dependent variable, N = 117; model 2 hetero-assessment as the dependent variable, N = 100) was obtained from two Brazilian federal government organizations. The data were collected at two distinct moments: on the first day of training, when the Professional Self-Concept scale was administered to the participants, and three months after the training, when the scales of Impact and Support were administered to the trainees and their respective supervisors. All the instruments were statistically validated. Only Psychosocial Support perceived by the trainees (sr2 = 0,25) and Material Support perceived by the supervisors (sr2 = 0,16) contributed, respectively, to the explanation of the variability of self and hetero-assessment of Training Impact. These results corroborated other national and international studies concerning the predictive power of Psychosocial Support over the Impact level. Self-Concept did not predict Impact of Training. However, this variable should not be discarded from the evaluation models, since it could explain the impact of educational programs and career systems over the individual competence profile.

Locus de controle: validação de uma escala em situação de treinamento

Locus of control: validation of a scale in a training context. This article presents the results of the development and validation process of a locus of Control instrument, in a sample (n = 1845) of training participants at public and private organizations in the Federal District. The valid responses to the developed instrument have constituted three factors, namely Internality, Externality/Chance and Externality/Others, each one of them with 12 items, with reliability scores higher than those reported in national and international studies (0.89, 0.92 and 0.80, respectively). This study also sought to investigate the sample of participants demographic profile, based on their scores in each one of the three mentioned factors. Given the created response categories, a higher number of significant differences were observed between the groups in the Internality factor. One reason for this may be that most respondents have reached high scores in this factor, because they could not correctly distinguish their judgments concerning external sources of control.

Reação aos procedimentos instrucionais de um curso via internet: validação de uma escala

Research into the evaluation of distance training is still in the early stages. This article intends to contribute to this area by validating statistically the tool Reaction to Educational Processes. The course being evaluated was offered by the Servico Brasileiro de Apoio as Micro e Pequenas Empresas to students across Brazil. The data collection was conducted at the end of the course by means of a questionnaire that measures the satisfaction of the participants with regard to the educational processes offered in the course. Analyses were carried out of the main, factorial components and Internal Consistency (Cronbachs Alpha Coefficient). Following statistical validation, the questionnaire demonstrated two factors: Traditional Processes, 12 questions, α=0.91, factor loads ranging from 0.53 to 0.79; Web Resources, 3 questions, confidence level α=0.76, and factor loads 0.40, 0.76 and 0.88. The results indicated that the scale presented evidence of construct validity.