Developing a measure of social understanding for 15- 24-year-olds that is appropriate, fair, valid, and theoretically motivated.

In adolescence the social lives of young people become increasingly complex. There is already good evidence from middle childhood that variation in the ability to build, manage and maintain social relationships (or 'social competence') matters: poor social competence in this age range predicts loneliness, later mental health difficulties, poorer academic outcomes, and even difficulties in work. Significant recent progress has been made in identifying understanding of others' mental states (sometimes called "mindreading") as an important underlying mechanism that provides a viable target for support and intervention.
There is every reason to expect that mindreading will also be an important basis of social competence in later adolescence and early adulthood, but it is likely that the nature of mindreading and its causes and consequences will change along with the dramatically changing social world of young people. For example, the period from 15-24 involves major social transitions between places of education and work, and between family, peer and romantic relationships. It is also a period of acute risk for mental ill-health, often with lifelong consequences. Understanding the relationships between mindreading and these social and mental health outcomes has important implications for mental health and wellbeing of young people, and will suggest ways in which environments for education and work might better support these outcomes.

This potential is currently difficult to fulfil, firstly because we lack robust measures of mindreading that are sensitive and meaningful for young people.

We will apply gold-standard psychometric approaches to develop a new measure of mindreading that is appropriate, valid, and sensitive in mid-adolescence to early adulthood. We will work with young people in story-exchange workshops to ensure that the mindreading scenarios and questions are relevant and meaningful. We will examine the measure's performance in 3000 young people aged 15-24, and check that measurements are stable, and fair across different demographic groups. We will test prediction of mental health (e.g., depression, anxiety) and social wellbeing (e.g., loneliness, and social support).

To maximise benefits of the new measure we will work with computer scientists to develop a reliable, open system for automated coding of participant's responses. Researchers using the new task will be relieved of hundreds of hours of manual coding, and will instead be able to upload participant responses to a secure remote server, and receive accurately coded responses within minutes. These steps will maximise the benefits to research capacity by enabling a wide range of researchers in psychology, psychiatry, and education, to adopt our methods for their own research.

The development of a new measure of mindreading will enable a wide range of potential benefits to research programmes seeking to advance the wellbeing of young people. Within the proposed work we will lay groundwork for understanding the causes and consequences of mindreading within places of education and work, and how these institutions and organisations can play a role in role in supporting mindreading and related benefits for wellbeing and mental health. To pursue this objective we have planned a series of workshops that will bring our researchers together with educators, mental health professionals, and employers, with young people with a particular interest in mental health. These will help us interpret our findings in relation to the real systems that young people encounter, and lay groundwork for identifying points of particular concern, examples of best practice and opportunities for improved practice that would be pursued in future work.

Background
Mindreading is important for social and mental wellbeing, but we lack psychometrically robust measures of mindreading that are sensitive and meaningful for young people.

Co-production of stimuli
We will use story-exchange workshops to devise and curate stimuli in collaboration with young people, ensuring that the test stimuli are meaningful for participants, and giving young people themselves authority over how they should be interpreted.

Psychometric approach
In developing a new task we will select test stimuli that optimise measurement, and establish: the specificity with which we are measuring mindreading; measurement invariance (test "fairness") for different groups of people; repeatability, and longer-term stability of measurement; validity for predicting social wellbeing and mental health.

Theory
Current theories do not adequately account for mindreading variation in this age range. We will test the novel hypothesis that a participant's flexibility to mindread a range of people and situations is the critical source of variability.

Samples of young people aged 15-24 will be tested in schools/colleges and online, with deliberate over-sampling in some demographic categories to ensure statistical power to establish measurement invariance.

Analysis approaches
We will use latent variable modelling, implemented in Mplus, to test theoretically-motivated hypotheses about the covariance structure of our data.
We will adapt "representational similarity analysis" approaches from cognitive neuroscience to characterise the multidimensional similarity of research participants and the authors of our test stimuli.

To make our new task truly accessible we will use our data to train and test a machine-learning natural-language processor to automatically code responses. The resulting coder will be openly available, and will relieve future researchers of the need to manually code thousands of responses.