Using new technologies to enhance the value of qualitative data in longitudinal studies: an application to health and well-being, and ageing

定性数据，如论文和调查中的自由回答问题，是心理、社会和行为信息的丰富来源。然而，这些信息在传统上是不可能大规模利用的。计算语言学和机器学习的最新进展已经产生了自动内容分析工具，这些工具已经开始用于各种环境，包括Facebook和Twitter等社交媒体环境中使用的文本。我们将首次将这些应用于在一项大型国家出生队列研究中纵向收集的公开回应，以便在健康、福祉和成功老龄化的研究中取得方法和理论上的进步。我们的工作将有助于一场重要的政策辩论，即儿童时期发展的哪些社交和情感技能对一生的幸福至关重要，哪些干预措施可以支持这些技能，以及支撑成功老龄化的因素。所应用的技术还将为跨越广泛的实质性主题领域的研究人员提供一种方法模型，用于从丰富但未充分利用的大规模定性数据集中提取更大的价值，从而在方法和实质性科学基础上进行这种转换研究。该项目包括三个主要步骤。第一步将是数字化转录国家儿童发展研究（NCDS）中包含的13000篇11岁的论文，NCDS是英国世界知名的出生队列研究之一。这些数据包括在11岁和50岁时写的自述文章，以回答以下问题：11岁时：“想象你现在25岁。写下你25岁时的生活，你的兴趣，你的家庭生活和你的工作。”50岁时：“想象你现在60岁了……请写几行关于你的生活（你的兴趣，你的家庭生活，你的健康和幸福，以及你可能正在做的任何工作）。这些回应（11岁时的13669人；50岁时的7383人）提供了一个很大程度上尚未开发的心理和行为信息来源，这些信息可以与同一个人的结果纵向联系起来。接下来，自动内容分析工具将应用于转录的文章，以便对11岁和50岁的文章中表达的单词和概念进行定量分析。个人使用的词汇将被分类为情感、社会关系、文章等不同的类别，从而可以对心理社会内容进行评估。我们将使用“开放词汇”和“封闭词汇”两种方法。然后，通过内容分析从开放文本中得出的分类将用于定量地解决一些研究问题，包括：1969年收集的一大群11岁儿童的文章中使用的语言反映了哪些心理特征和行为？这些揭示出来的心理特征和行为，与整个成年生活中健康和幸福的长期轨迹，以及55岁之前的早期衰老标志之间，有什么联系？在11岁至50岁之间的一生中，语言使用的持久性在多大程度上可以被发现？所揭示的特征和行为的持久性与成年生活中的健康和福祉有何关系？所开发的方法将具有变革性，并将有可能解开包含在许多其他国家纵向数据源中公开回应的信息。研究结果将对政策产生重大影响，为学校、当地社区组织和卫生从业人员提供有关在儿童时期和整个生活中发展社会和情感技能对终身健康和福祉的重要性的信息。

Qualitative data, such as essays and free response questions in surveys, are rich sources of psychological, social and behavioural information. Yet such information has traditionally been impossible to leverage at a large scale. Recent advances in computational linguistics and machine learning have produced automatic content analysis tools, which have started to be used in a variety of settings including in text used in social media settings such as Facebook and Twitter. We will apply these for the first time to the open responses collected longitudinally within a large national birth cohort study in order to make methodological and theoretical advances in the study of health, well-being, and successful ageing. Our work will contribute to an important policy debate about which social and emotional skills developed in childhood are vital for well-being throughout life, what interventions might support these, and the factors that underpin successful ageing. The techniques applied will also offer researchers working across a wide range of substantive topic areas a methodological model for extracting greater value out of rich but underutilised large-scale qualitative datasets, making this transformational research on both methodological and substantive scientific grounds. The project involves three major steps. The first step will be to digitally transcribe 13,000 age 11 essays contained within the National Child Development Study (NCDS), one of the UK's world-renowned birth cohort studies. The data include self-reported essays, written at age 11 and age 50, in response to the following questions: At age 11: "Imagine you are now 25 years old. Write about the life you are leading, your interests, your home life and your work at the age of 25" At age 50: "Imagine that you are now 60 years old...please write a few lines about the life you are leading (your interests, your home life, your health and wellbeing and any work you may be doing)". The responses (13,669 at age 11; 7,383 at age 50) provide a largely untapped source of psychological and behavioural information that can be linked longitudinally to outcomes for the same individuals. Automatic content analysis tools will next be applied to the transcribed essays in order to undertake quantitative analysis of the words and concepts expressed in essays at age 11 and 50. The words used by an individual will be classified into different categories, such as emotions, social relationships, and articles, allowing psycho-social content to be assessed. We will use both 'open vocabulary' and 'closed vocabulary' approaches. The classifications derived from open text through content analysis will then be used to quantitatively address a number of research questions, including: What psychological traits and behaviours are reflected in the language used in the essays of a large group of 11 year olds, collected in 1969? What is the association between such revealed psychological traits and behaviours and long-term trajectories of health and well-being across adult life, and early markers of ageing, captured up to the age of 55? How do future ambitions and expectations, as revealed in age-50 essays, relate to age-55 health statuses and practices What degree of persistence can be found in use of language used across the lifetime - between the ages of 11 and age 50, and how does the persistence of traits and behaviours revealed relate to health and well-being in adult life?The methods developed will be transformative - and will have the potential to unlock information contained open responses in many other national longitudinal data sources. The findings will have strong impact on policy, providing information relevant to schools, local community organisations and health practitioners as to the importance of developing social and emotional skills in childhood and throughout life for lifelong health and well-being.