Resources for secondary data analysis

About research at the Youth Endowment Fund (YEF) 

At the YEF, our mission is to prevent children and young people from becoming involved in violence. We do this by finding out what works and building a movement to put this knowledge into practice.  As an organisation, we fund a wide range of research, including systematic reviews, Evidence and Gap Maps, evaluations of programs we fund, and secondary data analysis. 

What is secondary data analysis? 

Secondary data analysis uses existing datasets to address key research questions for policy and practice in relation to children and young people’s involvement in crime and violence. Data sources vary from linked administrative (e.g., Police National Computer-National Pupil Database) and locally held data (e.g., by the local authority, police force, Violence Reduction Unit/Network), to panel/cohort studies (e.g., Millennium Cohort Study) and completed trials (YEF’s evaluation data archive). 

Secondary data analysis can help answer questions that other forms of research find hard – for instance where randomised controlled trials are infeasible, unethical or inefficient. With secondary data, we could draw on large numbers of observations (often millions), which increases the precision and confidence in our findings. They can enable us to track how things have changed over many years. And, they mean we can evaluate the impact of large scale policy and practice changes that affect whole systems or areas of society.  

What types of projects do we fund? 

We broadly think of secondary data analysis (SDA) as covering three different types of analysis: 

• Impact studies – This includes quasi-experimental designs that allow for causal inference on the effects of drivers, policies, practices and interventions on crime and violence outcomes.  
• Relationship testing – This includes regressions (e.g. panel designs/multi-level models) that test for the existence of relationships to crime and violence outcomes, although stops short of being able to make casual inference.  
• Descriptive analysis – This includes descriptions of population or issues that are not well understood or exploration of research questions related to crime and violence, for which there is very little evidence. 

To date, we’ve funded 18 SDA projects. These cover a wide range of questions on YEF’s priority sectors, research areas, datasets and methodologies.  

Our funded projects typically last  between 18 and 24 months. They have three distinct phases, with a deliverable at the end of each phase:  

1. Analysis plan pre-specification phase 
2. Interim reporting phase 
3. Final reporting phase 

All analysis plans and final reports go through a process of external peer review, managed by the YEF. 

How we assess evidence quality 

For impact studies, we apply a standardised evidence security rating system. To achieve our mission of preventing young people becoming involved in crime it is crucial that we can communicate to practitioners, funders and policy-makers to what extent they can trust our published findings. Our security rating system is designed to assess how robust are any causal claims made by the projects we fund. It also helps to bring consistency in how we talk about causality, across the different types of research we fund. The system takes into account the study’s research design, minimum detectable effect size, and various threats to internal validity, including confounding, attrition and missing data, and others.