IJPDS International Journal of Population Data Science IJPDS 2399-4908 Swansea University 10.23889/ijpds.v10i3.3189 10:3:157 Using SQL for accessing and working with large administrative data in TREs Teyhan Alison 1 Rollings Jasmine 2 Cornish Rosie 1 Brennan Iain 1 University of Bristol, Bristol, United Kingdom Cardiff and Vale UHB - Cedar, Cardiff, United Kingdom University of Hull, Hull, United Kingdom 01 06 2025 2025 8 4 3189 This work is licenced under a Creative Commons Attribution 4.0 International License. This article is available from the IJPDS website at: https://ijpds.org/article/view/3189 Objectives

To overcome respective measurement and sampling limitations in administrative and longitudinal cohort data sets, we compare results on the links between persistent unauthorised absence from school and self-reported and police-recorded violence perpetration using linked national administrative data and data from a longitudinal cohort.

 Methods

We used two sources of data: (i) a linkage between the National Pupil Database (NPD) and the Police National Computer (PNC); and (ii) a linkage between Avon Longitudinal Study of Parents and Children (ALSPAC), NPD and local police records. Persistent absence was defined as missing 10% or more of possible sessions in a term or school year (respectively). In the NPD-PNC data we examined persistent unauthorised absence only; in ALSPAC we examined both persistent absence overall and persistent unauthorised absence . Logistic regression, controlling – via matching and/or covariate adjustment - for a range potential confounders was used to examine the associations.

 Results

In the NPD-PNC data, there were just over 700,000 instances of persistent unauthorised absence included in our analysis. Of these, 3.7% had a conviction or caution for serious violence in the 12-month period following the term in which this persistent absence occurred; the corresponding figure among their matched control was 1.8% (adjusted odds ratio (OR) = 2.01; 95% confidence interval (CI): 1.87 to 2.15). In ALSPAC, there were 3,284 individuals with complete data for the analysis of self-reported violence and 6,436 for the analysis of police-recorded violence. The adjusted ORs for overall absence were 1.86 (1.38 to 2.50) and 1.97 (1.50 to 2.58), respectively. Results for persistent unauthorised absence in ALSPAC are waiting for clearance.

 Conclusion

We used two datasets with different advantages and disadvantages and different measures of violence to examine the link between persistent absence and the risk of subsequent violence perpetration and found, in both datasets, that persistent absence was associated with approximately a doubling of the risk of subsequent violence perpetration.