Training Workshops on Modelling Mechanisms of Change Using Longitudinal Archived Data

许多社会科学家对研究纵向变化很感兴趣。例如，教育研究者可能对儿童时期的学术能力如何发展感兴趣，经济学家可能对经济衰退期间个人债务水平如何变化感兴趣，或者政治科学家可能对人们对政党的态度如何变化感兴趣。现在有一些可用的统计方法，可用于描述群体和个人层面的纵向变化模式。潜在增长模型提供了一种强大而灵活的方法来描述纵向变化。目前统计建模的发展使社会科学研究人员能够超越简单地描述纵向变化，并对这些变化背后的机制和过程进行建模。因此，例如，研究人员可以检查那些可能影响或预测儿童时期学术能力发展的个体（如自尊、个性）、社会（如社会经济地位、父母教育成就）和环境（如学校和教师特征）因素。这样就可以更好地理解决定变化的机制和过程。这种模型可以使用一种灵活而强大的统计框架来制定和测试，这种框架被称为潜在变量建模（LVM）。尽管英国社会科学家可以利用LVM对纵向变化的描述性模型进行培训，但在使用这些模型以检查机制和过程方面，没有提供全面的培训。本项目旨在为英国社会科学家提供应用这些模型的高质量培训。研讨会将每年举办三次，为期三年，在三个不同的地点（贝尔法斯特，伦敦和斯特林）。每个工作坊将持续两天，包括讲座和实践环节。讲习班的一个重要方面是，它们将基于英国数据档案馆的数据集，该档案馆提供了英国社会科学领域最大的数据集。当统计数据和实质性问题明确结合时，使用二手数据分析的好处已得到证明。为了最大限度地发挥讲习班的影响，将有后续的网络研讨会，讲习班的培训材料也将存档在一个专门的网站上，专业软件将保留给主办机构。培训方案将使英国的研究人员和研究生有机会参加最新统计方法方面的培训，这些方法可以制定和检验与变化的性质和机制有关的复杂假设。

Many social scientists are interested in investigating longitudinal change. For example, a educational researcher may be interested in how academic ability develops during childhood, an economist may be interested in how levels of personal debt changes during a recession, or a political scientist may be interested in how attitudes towards political parties change. There are now statistical methods available which are useful in describing the patterns of longitudinal change, both at the group level of the individual level. Latent growth models provide a powerful and flexible approach to describing longitudinal change. Current developments in statistical modelling now allow social science researchers to move beyond simply describing longitudinal change and model the mechanisms and processes that underlie such changes. So, for example, a researcher could examine those individual (e.g. self-esteem, personality), social (e.g. socioeconomic status, parental educational achievement), and contextual (e.g. school and teacher characteristics) factors that may influence, or predict, how academic ability develops during childhood. This allows for the mechanisms and processes that determine change to be better understood. Such models can be formulated and tested using a flexible and powerful statistical framework known as latent variable modelling (LVM). Although UK social scientists can avail of training in descriptive models of longitudinal change using LVM, there is no provision for comprehensive training in the use of those models that allow for the mechanisms and processes to be examined. This project aims to provide high quality training for UK social scientists in the application of these models. Workshops will be delivered three times a year, for three years, in three different locations (Belfast, London, and Stirling). Each workshop will last two days and comprise lectures and practical sessions. One important aspect of the workshops is that they will be based on datasets from the UK Data Archive which provides the largest collection of data in the social sciences in the UK. The benefits of using secondary data analysis have been demonstrated when there is clear integration of statistics and substantive issues. In order to maximise the impact of the workshops there will be follow-up webinars, the training materials from the workshops will also be archived on a dedicated website, and the specialist software will remain with the host institutions.The programme of training will provide UK researchers and research students the opportunity to take part in training in state of the art statistical methods that allow complex hypotheses relating to the nature and mechanisms of change to be formulated and tested.