Ian Brunton-Smith

Professor Ian Brunton-Smith


Academic and research departments

Department of Sociology.



Research interests

Research projects


Postgraduate research supervision




Sturgis, P., Brunton-Smith, I., and Jackson, J. (2021) ‘Trust in science, social consensus and vaccine confidence’. Nature Human Behavior. 5: 1528–1534

In this paper we make use of a global survey to explore levels of vaccine hesitancy across the world. We demonstrate the importance of trust in science for vaccine uptake with less vaccine hesitancy in countries where trust is high. We also highlight the central role of social consensus, with the positive link between trust and vaccine support only evident in countries where consensus is high. 

You can read the full paper here, or take a look at this short blog


Allum, N., Besley, J., Gomez, L., and Brunton-Smith, I. (2018) 'Disparities in science literacy'. Science, 360 (6391), 861-862.

This was the first detailed study to look at disparities in science knowledge between adults from different racial and ethnic backgrounds. We found that people from black and Hispanic backgrounds were less able to answer questions about scientific facts and processes compared to white Americans. The study, published in Science (!), looked at potential reasons behind the disparity, including differences in basic literacy skills, attitudes to science (some minority groups expressed less trust and confidence in science), demographic factors such as education, gender, where people live and religion. After adjusting for all of these factors, a persistent science literacy gap remains, which could be related to the difference in the quality of education experienced day to day and year over year by underserved groups. This suggests the quantity and quality of science education needs to be looked at and we may also need training and public awareness campaigns to help scientists, teachers and employers to be more sensitive to the subtle manifestations of bias.


Pina-Sánchez, J., Buil-Gil, D., Brunton-Smith, I., and Cernat, A. (2023) ‘The Impact of Measurement Error in Models Using Police Recorded Crime Rates’. Journal of Quantitative Criminology. 39: 975-1002.

Objectives: Assess the extent to which measurement error in police recorded crime rates impact the estimates of regression models exploring the causes and consequences of crime. 

Methods: We focus on linear models where crime rates are included either as the response or as an explanatory variable, in their original scale, or log-transformed. Two measurement error mechanisms are considered, systematic errors in the form of under-recorded crime, and random errors in the form of recording inconsistencies across areas. The extent to which such measurement error mechanisms impact model parameters is demonstrated algebraically, using formal notation, and graphically, using simulations. 

Results: Most coefficients and measures of uncertainty from models where crime rates are included in their original scale are severely biased. However, in many cases, this problem could be minimised, or altogether eliminated by log-transforming crime rates. This transforms the multiplicative measurement error observed in police recorded crime rates into a less harmful additive mechanism. 

Conclusions: The validity of findings from regression models where police recorded crime rates are used in their original scale is put into question. In interpreting the large evidence base exploring the effects and consequences of crime using police statistics we urge researchers to consider the biasing effects shown here. Equally, we urge researchers to log-transform crime rates before they are introduced in statistical models.

McCarthy, D., and Brunton-Smith, I. (2022) ‘International Attitudes towards the Death Penalty: An Assessment of Individual and Country-Level Differences’. European Journal of Criminology.

Research on public attitudes to the death penalty has been predominantly understood through single nation-states, especially within the USA. Examinations of international differences in citizens’ support for the death penalty have been scarce, particularly among continents with a high volume of retentionist nations (e.g. Asia). In this paper, we draw on a dataset of 135,000 people from across 81 nations to examine differences in death penalty support. We find that residents of retentionist nations are generally more supportive of the death penalty than those from abolitionist nations. But this general difference masks important differences both within and between countries. At the country-level, residents of abolitionist nations with autocratic political systems and those with higher homicide levels were more likely to support the death penalty than residents of other abolitionist nations. At the individual level, greater support for a strong dictatorial-type leader and perceptions of political corruption are associated with increased support for the death penalty, but only in abolitionist nations. By contrast, more frequent religious worship, perceived egalitarianism in a nation, and support for the political performance of government reduced death penalty support in abolitionist nations but increased support in retentionist nations, while belief in individual responsibility and critical views towards ethnic minorities increased support for the death penalty across both abolitionist and retentionist nations.

Meadows, R., Brunton-Smith, I., and Ellis, J. (2022) ‘Are sleep quality judgements comparable across individuals, place and space? An interdisciplinary analysis of data from 207,608 individuals from 68 countries’. Sleep Health. Online first

Objectives: Significant research has shown that health is a heterogeneous concept, and one person's poor health may not be comparable with another's. Yet, little consideration has been given to whether sleep quality judgments are also heterogeneous or whether they cohere between individuals. Another possibility is that there are group differences in the ways in which sleep quality is perceived. If this is the case, it is possible known inequalities in sleep are—in part—an artifact of social position influencing how we conceive of sleep problems. The current study explores this possibility.

Design: Cross-sectional, using World Health Organization data from 207,608 individuals; aged between 15 and 101 years of age from 68 countries. Alongside a battery of sleep and demographic variables, data contained sleep and energy vignettes. Random effect anchoring vignette models were applied to investigate interpersonal incompatibility and whether sleep quality perceptions operate differently depending on social location, context, and function.

Results: While sleep quality judgments are largely comparable across individuals, findings also highlight how the relationship between education and self-reported sleep changes following adjustment for reporting heterogeneity. Estimates of threshold parameters suggest that those with more years of education have a slightly increased threshold for reporting mild sleep problems (B 0.005; s.e. 0.001) but a lower threshold for reporting sleep problems as extreme (B -0.007; s.e. 0.001).

Conclusions: Sleep quality judgments occupy a complex position between heterogeneity and coherence. This has implications for both epidemiological methodologies and contemporary debates about social justice, public health and sleep.

Buil-Gil, D., Cernat, A., Brunton-Smith, I., Pina-Sánchez, J. (2022) ‘Comparing Measurements of Crime in Local Communities: A Case Study in Islington, London’. Police Practice and Research: An International Journal. Online first.

Police-recorded crime data are prone to measurement error, affecting our understanding of the nature of crime. Research has responded to this problem using data from surveys and emergency services. These data sources are not error-free, and data from different sources are not always easily comparable. This study compares violent crime data recorded by police, ambulance services, two surveys and computer simulations in Islington, London. Different data sources show remarkably different results. However, crime estimates become more similar, but still show different distributions, when crime rates are calculated using workday population as the denominator and log-transformed. Normalising crime rates by workday population controls for the fact that some data sources reflect offences’ location while others refer to victims’ residence, and log- transforming rates mitigates the biasing effect associated with some multiplicative forms of measurement error. Comparing multiple data sources allows for more accurate descriptions of the prevalence and distribution of crime.

Salmela, J., Brunton-Smith, I., and Meadows, R. (2022) 'Inequalities in recovery or methodological artefact? A comparison of models across physical and mental health'. Social Science and Medicine - Population Health. 17

Considerable attention has been paid to inequalities in health. More recently, focus has also turned to inequalities in ‘recovery’; with research, for example, suggesting that lower grade of employment is strongly associated with slower recovery from both poor physical and poor mental health. However, this research has tended to operationalise recovery as ‘return to baseline’, and we know less about patterns and predictors when recovery is situated as a ‘process’. This paper seeks to address this gap. Drawing on data from the UK Household Longitudinal Study, we operationalise recovery as both an ‘outcome’ and as a ‘process’ and compare patterns and predictors across the two models. Our analysis demonstrates that the determinants of recovery from poor health, measured by the SF-12, are robust, regardless of whether recovery is operationalised as an outcome or as a process. For example, being employed and having a higher degree were found to increase the odds of recovery both from poor physical and mental health functioning, when recovery was operationalised as an outcome. These variables were also important in distinguishing health functioning trajectories following a poor health episode. At one and the same time, our analysis does suggest that understandings of inequalities in recovery will depend in part on how we define it. When recovery is operationalised as a simple transition from poor health state to good, it loses sight of the fact that there may be inequalities (i) within a ‘poor health’ state, (ii) in how individuals are able to step into the path of recovery, and (iii) in whether health states are maintained over time. We therefore need to remain alert to the additional nuance in understanding which comes from situating recovery as a process; as well as possible methodological artefacts in population research which come from how recovery is operationalised.

Buil-Gil, D., Pina-Sánchez, J., Brunton-Smith, I., and Cernat, A. (2022) Bad Data, Worse Predictions: How Measurement Error in Crime Data Affects Crime Prevention. CREST Security Review Magazine.

Increasingly sophisticated methods are applied to predict crime patterns from police records with little regard given to the quality of the data. We explore how this may affect crime prevention.

Brunton-Smith, I., Flatley, J., and Tarling, R. (2022) ‘Prevalence of sexual violence: a comparison of estimates from UK National Surveys’. European Journal of Criminology. 19(5): 891-910.

Accurately measuring the prevalence of sexual violence is difficult. Police-recorded crime figures are known to underestimate the true extent of sexual violence, and so researchers have tended to rely on survey estimates instead. But estimates from surveys are not uniform, with recent estimates from the UK National Survey of Sexual Attitudes and Lifestyles apparently twice as large as official figures from the major crime surveys (the Crime Survey for England and Wales and the Scottish Crime and Justice Survey). In this study we use harmonized data from these three surveys and the UK component of the EU Violence Against Women Survey to explore the features of the surveys that may have contributed to these differences.

Buil-Gil, D., Brunton-Smith, I., Pina-Sánchez, J., Cernat, D., and Moretti, A. (2021) Synthetic crime data in England and Wales (version 0.1).

An artificial dataset constructed by using CSEW data to predict victimisation propensities, and mapping these onto census population figures. Full details and code available at link above.

Cernat, A., Buil-Gil, D., Brunton-Smith, I., Pina-Sánchez, J., and Murrià-Sangenís, M. (2021) 'Estimating Crime in Place: Moving Beyond Residence Location'. Crime and Delinquency.

We assess if asking victims about the places where crimes happen leads to estimates of ‘crime in place’ with better measurement properties. We analyse data from the Barcelona Victimization Survey (2015 to 2020) aggregated in 73 neighbourhoods using longitudinal quasi-simplex models and criterion validity to estimate the quality of four types of survey-based measures of crime. The distribution of survey-based offence location estimates, as opposed to victim residence estimates, is highly similar to police-recorded crime statistics, and there is little trade off in terms of the reliability and validity of offence location and victim residence measures. Estimates of crimes reported to the police show a better validity, but their reliability is lower and capture fewer crimes.

Yesberg, J., Bradford B., and Brunton-Smith, I. (2021) 'Police visibility, trust in police fairness, and collective efficacy: A multilevel Structural Equation Model'. European Journal of Criminology

Areas high in collective efficacy – where residents know and trust one another and are willing to intervene to solve neighbourhood problems – tend to experience less crime. Policing is thought to be one antecedent to collective efficacy, but little empirical research has explored this question. Using three waves of survey data collected from London residents over three consecutive years, and multilevel Structural Equation Modelling, this study tested the impact of police visibility and police–community engagement on collective efficacy. We explored direct effects as well as indirect effects through trust in police. The findings showed levels of police visibility predicted trust in police. Trust in police fairness, in turn, predicted collective efficacy. There was a small indirect relationship between police visibility and collective efficacy, through trust in police fairness. In other words, police presence in neighbourhoods was associated with more positive views about officer behaviour, which in turn was associated with collective efficacy. The findings have important implications for policies designed to build stronger, more resilient communities.

Pina-Sánchez, J., and Brunton-Smith, I. (2021) 'Are We All Equally Persuaded by Procedural Justice? Measuring the ‘Invariance Thesis’ Using Longitudinal Data and Random Effects'. Journal of Developmental and Life-Course Criminology.

A growing number of empirical studies has sought to explore differences in the effectiveness of the procedural justice model across people. Much of this new evidence points at the procedural justice association with both legitimacy and compliance being largely invariant. Here we expand the analysis of this procedural justice ‘invariance thesis’ by introducing a novel life-course perspective to the debate. Specifically, we focus on the variability of the procedural justice effect within individuals across time. To do so, we use mixed effects structural equation models and longitudinal data from a sample of 1,354 young offenders in the US reporting perceptions of the police, and a sample of 511 subjects of the Australian general population reporting on the tax authority. We find the procedural justice within-person association with legitimacy to be highly variant across individuals, which can be negative for more than 10% of subjects in the two samples used, while for at least another 11% of participants the relationship is twice as strong as the average or stronger. We also find variability in the within-person association with compliance; however, this is only the case for a specific measure of procedural justice in the sample of young offenders. These results question the ‘invariance thesis’. Compliance, and especially perceptions of institutional legitimacy, cannot be expected to change uniformly across all subgroups of the population in line with their perceptions of the procedural just actions of those institutions.

Sturgis, P., Brunton-Smith, I., and Jackson, J. (2021) ‘Trust in science, social consensus and vaccine confidence’. Nature Human Behavior. 5(November 2021): 1528-1534.

While scholarly attention to date has focused almost entirely on individual-level drivers of vaccine confidence, we show that macro-level factors play an important role in understanding individual propensity to be confident about vaccination. We analyse data from the 2018 Wellcome Global Monitor survey covering over 120,000 respondents in 126 countries to assess how societal-level trust in science is related to vaccine confidence. In countries with a high aggregate level of trust in science, people are more likely to be confident about vaccination, over and above their individual-level scientific trust. Additionally, we show that societal consensus around trust in science moderates these individual-level and country-level relationships. In countries with a high level of consensus regarding the trustworthiness of science and scientists, the positive correlation between trust in science and vaccine confidence is stronger than it is in comparable countries where the level of social consensus is weaker.

Jackson, J., Brunton-Smith, I., Bradford, B., Posch, K., Rodriguez-Oliveira., and Sturgis, P. (2021) ‘Police Legitimacy and the Norm to Cooperate: Using a Mixed Effects Location-Scale Model to Estimate the Strength of Social Norms’. Journal of Quantitative Criminology. 37(2): 547-572.

Objectives: Test whether cooperation with the police can be modelled as a place-based norm that varies in strength from one neighborhood to the next. Estimate whether perceived police legitimacy predicts an individual’s willingness to cooperate in weak-norm neighborhoods, but not in strong-norm neighborhoods where most people are either willing or unwilling to cooperate, irrespective of their perceptions of police legitimacy.

Methods: A survey of 1,057 individuals in 98 relatively high-crime English neighborhoods defined at a small spatial scale measured (a) willingness to cooperate using a hypothetical crime vignette and (b) legitimacy using indicators of normative alignment between police and citizen values. A mixed-effects, location-scale model estimated the cluster-level mean and cluster-level variance of willingness to cooperate as a neighborhood-level latent variable. A cross-level interaction tested whether legitimacy predicts individual-level willingness to cooperate only in neighborhoods where the norm is weak.

Results: Willingness to cooperate clustered strongly by neighborhood. There were neighborhoods with (i) high mean and low variance, (ii) high mean and high variance, (iii) (relatively) low mean and low variance, and (iv) (relatively) low mean and high variance. Legitimacy was only a positive predictor of cooperation in neighborhoods that had a (relatively) low mean and high variance. There was little variance left to explain in neighborhoods where the norm was strong.

Conclusions: Findings support a boundary condition of procedural justice theory: namely, that cooperation can be modelled as a place-based norm that varies in strength from neighborhood to neighborhood and that legitimacy only predicts an individual’s willingness to cooperate in neighborhoods where the norm is relatively weak.

Fenn. L., and Brunton-Smith, I. (2021) 'The Effects of Terrorist Incidents on Public Worry of Future Attacks, Views of the Police and Social Cohesion' British Journal of Criminology. 61(2): 497–518.

Despite a growing body of research examining the psychological effects of terrorist incidents, there remains comparatively little empirical assessment of their impacts on citizens’ worry about further attacks, perceptions of the police or social cohesion. Drawing on interviews with nearly 100,000 London residents, we find higher levels of worry following most domestic attacks. Improvements in overall ratings of the police are tempered by more negative assessments of their ability to handle future threats. We also find that far-right incidents are less closely linked to changes in public ratings of the police and concerns about future attacks compared to Islamic terror attacks. Effects on social cohesion are less predictable. We find reductions in cohesion following some attacks but increases following others.

Pina-Sánchez, J., and Brunton-Smith, I. (2020) ‘Reassessing the Relationship between Procedural Justice and Police Legitimacy’. Law and Human Behavior. 44(5): 377-393.

Objective: An overwhelming body of research based on cross-sectional data has identified a positive relationship between perceptions of police procedural justice and legitimacy. Following Tyler’s theoretical framework these studies have often interpreted the observed relationship as evidence of an unequivocal causal connection from procedural justice to legitimacy. Here we re-examined the validity of this conclusion by considering the temporal order of that association and the potential biasing effect of time-invariant third common causes.

Hypotheses: 1) Past perceptions of police procedural justice predict future perceptions of legitimacy; 2) Past perceptions of police legitimacy predict future perceptions of procedural justice; and 3) Perceptions of police procedural justice and legitimacy are associated as a result of third common causes.

Method: We fitted random intercepts cross-lagged panel models to seven waves of a longitudinal sample of 1,354 young offenders in the ‘Pathways to Desistance’ study. This allowed us to explore the directional paths between perceptions of police procedural justice and legitimacy, while controlling for time-invariant subject heterogeneity.

Results: We did not find evidence of the assumed temporal association; lagged within subject perceptions of procedural justice rarely predicted within subject perceptions of legitimacy. We did not find evidence of a reciprocal relationship either. Instead, we detected substantial time-invariant subject heterogeneity, and perceptions of legitimacy being self-reproduced.

Conclusions: Our findings challenge the internal validity of the commonly reported positive associations between procedural justice and legitimacy found in studies using cross-sectional data.

Sturgis, P., Durrant, G., Maslovskya, O., and Brunton-Smith, I. (2020) 'The interviewer contribution to variability in response times in face-to-face interview surveys'. Journal of Survey Statistics and Methodology.

Survey researchers have consistently found that interviewers make a small but systematic contribution to variability in response times. However, we know little about what the characteristics of interviewers are that lead to this effect. In this study, we address this gap in understanding by linking item-level response times from wave 3 of the UK Household Longitudinal Survey (UKHLS) to data from an independently conducted survey of interviewers. The linked data file contains over three million records and has a complex, hierarchical structure with response latencies nested within respondents and questions, which are themselves nested within interviewers and areas. We propose the use of a cross-classified mixed-effects location scale model to allow for the decomposition of the joint effects on response times of interviewers, areas, questions, and respondents. We evaluate how interviewer demographic characteristics, personality, and attitudes to surveys and to interviewing affect the length of response latencies and present a new method for producing interviewer-specific intra-class correlations of response times. Hence, the study makes both methodological and substantive contributions to the investigation of response times

Brunton-Smith, I., Pina-Sánchez, J., and Li, G. (2020) ‘Re-assessing the consistency of sentencing decisions in cases of Assault: Allowing for within court inconsistencies’. British Journal of Criminology. 60(6): 1438-1459.

Empirical research has repeatedly focused on the potential existence of sentencing disparities. In particular, a growing number of studies have used multilevel models to quantify the extent that ‘similar’ offences are treated alike in different courts. This reliance on multilevel models has resulted in a natural focus on differences in the mean sentence awarded between courts, with the amount of within group variability generally assumed to be the same in each court. In this paper we show how multilevel models can be extended by allowing the magnitude of within-court differences to be different in each court. This provides a natural framework to connect between-court disparities with the sentencing differences that are thought to originate between judges operating within the same court, particularly in the absence of more fine-grained sentencing data about the judge residing in each case. Focusing specifically on cases of assault sentenced in 2011, we show that there are substantial differences in the range of sentences awarded in different courts, with the range almost twice as large in some courts. We also find that it is those courts that appear to show the traits of more homogeneous sentencing that sentence more harshly, and that offences involving the presence of a weapon or evidence of good character and/or exemplary conduct were associated with higher levels of internal consistency.

Sutherland, A., Brunton-Smith, I, Hutt, O., and Bradford, B. (2020) Violent crime in London: Trends, trajectories, and neighbourhood effects. Report for the College of Policing.

The aim of this report is to illustrate some of the ways in which police-recorded crime data can be combined with other sources to provide a deeper insight into the geographical and social distribution of violent crime. Specifically, four different analytical approaches – mapping, predictive models, trajectory models, and machine learning – are used to consider the patterning of violent crime across London. Drawing on a range of administrative and survey-based data, results converge on three central findings. Violent crime is heavily clustered in a small number of lower layer super output areas (LSOAs). These areas tend to be significantly more deprived than others; and the recent increase in violent crime has been primarily confined to a small number of areas, many of which were already relatively prone to violent offending.

The characteristics of areas – such as deprivation – seem to be the most important and consistent drivers of violent crime, although other area characteristics, such as the presence of transport hubs or major shopping centres or night-time economies, are also relevant. The distribution of violent crime over London is therefore predictable, in the sense that it clusters in a relatively small number of generally more deprived areas. It is these areas, moreover, that tend to experience the sharpest increases in violent crime, when and where this occurs. Public policy on violent crime can be effectively targeted at a small number of quite readily identifiable locations. To put it another way, expenditure on place-based violence reduction programmes would, by and large, be wasted in large parts of London because violent crime has not really increased in many areas.

Pina-Sánchez, J., Brunton-Smith, I., Grech, D., and Sferopoulos, D. (2019) ‘Exploring the Origin of Sentencing Disparities in the Crown Court: Using Text Mining Techniques to Differentiate between Court and Judge Disparities’. Social Science Research. 84. 1023-43.

Research on sentence consistency in England and Wales has focused on disparities between courts, with differences between judges generally ignored. This is largely due to the limitations in official data. Using text mining techniques from Crown Court sentence records available online we generate a sample of 7,212 violent and sexual offences where both court and judge are captured. Multilevel time-to-event analyses of sentence length demonstrate that most disparities originate at the judge, not the court-level. Two important implications follow: i) the extent of sentencing consistency in England and Wales has been overestimated; and ii) the importance attributed to the location in which sentences are passed – in England and Wales and elsewhere - needs to be revisited. Further analysis of the judge level disparities identifies judicial rotation across courts as a practice conducive of sentence consistency, which suggests that sentencing guidelines could be complemented with other, less intrusive, changes in judicial practice to promote consistency.

Joliffe, D., Farrington, D.P., Brunton-Smith, I., Loeber, R., Ahonen, L., and Paula Palacios, A. (2019) ‘Depression, Anxiety and Delinquency: Results from the Pittsburgh Youth Study’. Journal of Criminal Justice. 62: 42-49.

Purpose: The main aim of this research is to investigate to what extent within-individual changes in anxiety and depression are related to within-individual changes in theft and violence.

Methods: The youngest sample of the Pittsburgh Youth Study (PYS), a prospective longitudinal survey of 503 boys followed up from age 7 onwards, was analyzed.  Depression and anxiety were measured for boys from ages 11 to 16 as were moderate and serious forms of self-reported theft and violence. A hierarchical linear random effects model was used to investigate anxiety and depression as potential causes or outcomes of these forms of delinquency.

Results: The results showed that the between-individual correlations were consistently higher than the corresponding within-individual correlations, and provided little evidence to discern the directionality of the potential relationships between depression, anxiety and delinquency. Using a random effects approach, there was limited evidence that prior depression or anxiety was related to later offending, but there was evidence that offending (particularly theft and serious violence) was associated with later increases in anxiety, and to a lesser extent depression.   

Conclusions: This study indicates that depression and anxiety were outcomes of offending. If replicated, this would suggest that evidence-based interventions which reduced offending would have a desirable influence in reducing depression and anxiety. However, interventions for depression should still form part of responsive interventions. More research which explores within-individual changes in longitudinal studies with repeated measures is needed.  

Pina-Sánchez, J., Brunton-Smith, I., and Li, G. (2020) Mind the step: A More Insightful and Robust Analysis of the Sentencing Process in England and Wales under the New Sentencing Guidelines. Criminology and Criminal Justice. 20(3): 268-301.

The ‘England and Wales Sentencing Guidelines’ aim to promote consistency by organising the sentencing process as a sequence of steps, with initial judicial assessments subsequently adjusted to reflect relevant case characteristics. Yet, existing evaluations of the guidelines have failed to incorporate this structure adequately, instead concentrating solely on sentence outcomes. We use multivariate multilevel models to offer new insights into the decisions made throughout the sentencing process. Focusing on cases of assault sentenced at the Crown Court we show that the level of compliance with the guidelines is high. However, we also show that some case characteristics are being unduly considered at more than one stage of the sentencing process, meaning existing studies may be underestimating their true influence.

Quick, M., Li, G., and Brunton-Smith, I. (2018) ‘Crime-general and crime-specific spatial patterns: A multivariate spatial analysis of four crime types at the small-area scale’. Journal of Criminal Justice, 58: 22-32.

Purpose: To examine if, and how, spatial crime patterns are explained by one or more underlying crime-general patterns.

Methods: A set of Bayesian multivariate spatial models are applied to analyze burglary, robbery, vehicle crime, and violent crime at the small-area scale. The residual variability of each crime type is partitioned into shared and type-specific components after controlling for the effects of population density, deprivation, residential instability, and ethnic heterogeneity. Shared components account for the correlations between crime types and identify the crime-general patterns shared amongst multiple crimes.

Results: Two shared components are estimated to capture the crime-general pattern for all four crime types and the crime-general pattern for theft-related crimes (burglary, robbery, and vehicle crime). Robbery and violent crime exhibit the strongest positive associations with deprivation, instability, and ethnic heterogeneity. Shared components explain the largest proportions of variability for all crime types. Burglary, robbery, and vehicle crime each exhibit type-specific patterns that diverge from the crime-general patterns.

Conclusions: Crime-general patterns are important for understanding the spatial patterning of many crime types at the small-area scale. Multivariate spatial models provide a framework to directly quantify the correlation structures between crimes and reveal the underlying crime-general patterns shared amongst multiple crime types.

Brunton-Smith, I., and Bullock, K. (2018) ‘Patterns and Drivers of Co-production in Neighbourhood Watch in England and Wales: From Neo-liberalism to New Localism’. British Journal of Criminology, 59(1): 85-106.

Established in England and Wales in the context of the neo-liberal governments of the 1980s and promoted through the New Local agenda of New Labour and beyond, Neighbourhood Watch (NW) is a primary means through which the state and citizens may co-produce crime control. However, whether citizens have the time or inclination to co-produce is debated, and it is generally believed that NW proliferates in advantaged, low crime rate areas that need it least. Drawing on analysis of the Crime Survey for England and Wales (CSEW) (1988–2010/11), this article examines long-term trends in participation in NW. It examines the spread of NW, how household support for NW fluctuates once established and the changing importance of some of the key household drivers of participation in NW. It then assesses the extent to which NW schemes are concentrated in more affluent areas, showing that this is moderated by crime risk.

Brunton-Smith, I., Sturgis, P., and Leckie, G. (2018) ‘How collective is collective efficacy? The importance of consensus in judgments about community cohesion and willingness to intervene’. Criminology, 56(3): 608-637.

Existing studies have generally measured collective efficacy by combining survey respondents’ ratings of their local area into an overall summary for each neighborhood. Naturally, this approach results in a substantive focus on the variation in average levels of collective efficacy between neighborhoods. In this article, we focus on the variation in consensus of collective efficacy judgments. To account for differential consensus among neighborhoods, we use a mixed‐effects location‐scale model, with variability in the consensus of judgments treated as an additional neighborhood‐level random effect. Our results show that neighborhoods in London differ, not just in their average levels of collective efficacy but also in the extent to which residents agree with one another in their assessments. In accord with findings for U.S. cities, our results show that consensus in collective efficacy assessments is affected by the ethnic composition of neighborhoods. Additionally, we show that heterogeneity in collective efficacy assessments is consequential, with higher levels of criminal victimization, worry about crime, and risk avoidance behavior in areas where collective efficacy consensus is low.

Allum, N., Besley, J., Gomez, L, and Brunton-Smith, I. (2018) ‘Disparities in science literacy’. Science, 360 (6391): 861-862.

Much is known about how adult science literacy varies internationally and over time, and about its association with attitudes and beliefs. However, less is known about disparities in science literacy across racial and ethnic groups (1). This is particularly surprising in light of substantial research on racial and ethnic disparities in related areas such as educational achievement, math and reading ability (2), representation in science, technology, engineering, and math (STEM) occupations (3), and health literacy (4). Given the importance of science literacy to securing and sustaining many jobs, to understanding key health concepts to enhance quality of life, and to increasing public engagement in societal decision-making (5), it is concerning if the distribution of science literacy is unequally stratified, particularly if this stratification reflects broader patterns of disadvantage and cultural dominance as experienced by minorities and educationally underserved populations. We describe here such disparities in science literacy in the United States and attempt to explain underlying drivers, concluding that the science literacy disadvantage among black and Hispanic adults relative to whites is only partially explained by measures of broader, foundational literacies and socioeconomic status (SES).

McCarthy, D., and Brunton-Smith, I. (2018) ‘The effect of penal legitimacy on prisoners’ post-release desistance’. Crime and Delinquency, 64 (7): 917-938.

Studies of procedural justice and legitimacy have shown that where legal actors use formal rules in ways that are perceived to be fair and consistent by those policed, greater compliance with the law can be achieved. A number of studies have assessed how legitimacy and compliance are related using general population samples, but few have tested these links among offending groups. Drawing on data from a longitudinal survey of prisoners across England and Wales, we find that prisoners who perceive their experience of prison as legitimate are more likely to believe that they will desist from crime. However, despite the existence of desistance beliefs, these do not translate into similar effects of legitimacy on proven reconviction rates a year post release.

Skudder, H., Brunton-Smith, I., Druckman, A., Cole, J., and McInnes, A. (2018) ‘The falling carbon footprint of acquisitive and violent offences’. British Journal of Criminology, 58 (2): 351-371.

Cutting carbon emissions, wherever they occur, is a global priority and those associated with crime are no exception. We show that between 1995 and 2015, the carbon footprint of acquisitive and violent crime has dropped by 62 per cent, a total reduction of 54 million tonnes CO2e throughout this period. Although the environmental harm associated with crime is likely to be considered lower in importance than social or economic impacts, a focus on reducing high carbon crimes (burglary and vehicle offences) and high carbon aspects of the footprint (the need to replace stolen/damaged property) could be encouraged. Failure to acknowledge these potential environmental benefits may result in crime prevention strategies being unsustainable and carbon reduction targets being missed.

Skudder, H., Brunton-Smith, I., Tseloni, A., McInnes, A., Cole, J., Thompson, R., and Druckman, A. (2018) ‘Can burglary prevention be low-carbon and effective? Investigating the environmental performance of burglary prevention measures’ Security Journal, 31: 111-138.

There has been limited study to date on the environmental impacts of crime prevention measures. We address this shortfall by estimating the carbon footprint associated with the most widely used burglary prevention measures: door locks, window locks, burglar alarms, lighting and CCTV cameras. We compare these footprints with a measure of their effectiveness, the security protection factor, allowing us to identify those measures that are both low-carbon and effective in preventing burglary. Window locks are found to be the most effective and low-carbon measure available individually. Combinations of window locks, door locks, external and indoor lightings are also shown to be effective and low-carbon. Burglar alarms and CCTV do not perform as strongly, with low security against burglary and higher carbon footprints. This information can be used to help inform more sustainable choices of burglary prevention within households as well as for crime prevention product design.

Sturgis, P. Brunton-Smith, I and Jackson, J. (2018) Regression-based response probing for assessing the validity of survey questions, in Paul Beatty, Willis, G., and Padilla, J-L. (eds) Question Design Testing and Evaluation. Wiley.

We describe a procedure for assessing the validity of survey questions. Response probes are administered which ask respondents to say in their own words what came to mind when answering the question. The verbatim responses are coded to a frame which captures the conceptual content of the responses and are then included as predictors in a regression model, where the question that was used to elicit the verbatim responses is specified as the outcome. Controls are included so that the estimated coefficients of the model can be used to interpret whether and to what extent the different cognitive frames identified in the verbatim data align with responses to the question. The technique can be extended to include split-ballot designs, where variants of the target question are randomised across groups. We illustrate the procedure using two example questions: interpersonal trust and fear of crime.

Brunton-Smith, I. (2018) ‘Chapter 31 – Systematic Social Observation’ in Johnson, S., and Bruinsma, G. (eds) The Oxford Handbook of Environmental Criminology, Oxford University Press.

Systematic Social Observation (SSO) is the application of rigorous, replicable, and generalizable approaches from survey methodology to field observation (ethnography). With its roots in early symbolic interactionism in the late 1920s and 30s - where the emphasis was on participant observation to study social interactions (e.g. Thomas et al., 1933; Whyte, 1943; Barker and Wright, 1955) – SSO provides researchers with independent, robust, and quantifiable data about the forms of social interactions most commonly obtained through qualitative inquiry. Introduced to criminology by Reiss (1968; 1971) in his detailed analyses of violence in police-citizen encounters, it was the work of Taylor and colleagues in Baltimore city which first demonstrated how SSO approaches could also be used to measure qualities of the urban environment including physical disorder, social incivilities, and housing quality (Taylor, Gottfredson and Brower, 1984; Taylor, Shumaker and Gottfredson, 1985). They showed that SSO could provide researchers with independent assessments of the urban environment that were considerably more detailed than available administrative sources like the census and land registry data. SSO was further cemented in the methodological toolbox of environmental criminologists by the work of Sampson and colleagues as part of the Project on Human Development in Chicago Neighborhoods (PHCDN). In particular, with the development of ecometrics – the science of measurement applied to ecological settings – researchers now had a robust methodological foundation from which to judge the quality of ecological measures, and make corrections for measurement error (Raudenbush and Sampson, 1999). 

More recently, SSO for ecological assessment has evolved again. Data about the environment around survey respondents’ houses is now routinely collected by interviewers during the fieldwork process, and this information is now being incorporated in research (Brunton-Smith and Sturgis, 2011). Whilst questions remain about the quality of this data (Andresen et al., 2013; West, 2013; Casas-Cadero et al., 2013), this represents an extremely cost-effective source of SSO data for researchers. Perhaps more interestingly, the rapid growth of the online mapping tool Google Street View now provides an unprecedented amount of recorded observational data on environments across the globe which researchers are only just beginning to engage with (e.g. Badland et al., 2010; Clarke et al., 2010; Odgers et al., 2012). Methodological challenges with this data remain, however Google Street View unquestionably represents an exciting opportunity for the widespread and routine adoption of SSO in studies of ecological settings. 

This chapter proceeds as follows. First, section I provides a general overview of the SSO approach, drawing on the initial insights of Reiss (1971), as well as the work of Mastrofski et al., (2010). Section II describes the application of SSO approaches to environmental settings, from the original studies of Taylor and colleagues (Taylor, Gottfredson and Brower, 1984; Taylor, Shumaker and Gottfredson, 1985; Perkins et al., 1993; Perkins and Taylor, 1996; Taylor, 2001), to more recent work by Sampson and Raudenbush, (1999, see also Raudenbush and Sampson, 1999). The ecometric approach first proposed by Raudenbush and Sampson (1999) is discussed in section III, demonstrating how the science of SSO and the science of ecological analysis were intimately related. Finally, section IV discusses current developments in SSO for environmental analysis. This focuses specifically on three areas: interviewer observational data routinely collected as part of larger social surveys; participant generated observations using space-time budgets; and Google Street View as a repository for environmental audit data. Developments in crowd sourced SSO data like Mapilliary are also briefly discussed. 

Brunton-Smith, I. (2018) ‘Fear 2.0: Worry about cybercrime in England and Wales’ in Lee, M., and Mythen. G. (eds) International handbook of fear of crime. Routledge. Oxon.

Much has been written about the fear of crime, and there now exists a considerable body of evidence about the causes and consequences of increased levels of fear amongst particular groups. Yet one area that has received comparatively little attention is fear of cybercrime (Henson and Reyns 2015).  This is a surprising omission, with recent evidence suggesting that a substantial proportion of the population may be a victim of online crime each year, and as many as 120,000 new phishing websites appearing each month (APWG 2016). The rapid growth of smart-phone and other handheld computer technology is providing users with almost uninterrupted access to online spaces, and at the same time people are sharing more and more personal information online, placing them at increasing risk of online victimisation. The constantly evolving landscape of cyberspace also represents an increasingly uncertain social space for users, with the possibility that these anxieties will also translate into higher levels of fear of crime. This chapter provides one of the first in depth analyses of fear of cybercrime, using data from a nationally representative survey to identify those individuals that are most fearful, as well as the key drivers of this fear. The link between fear of online and offline crimes is also examined, to determine the extent that people transfer their offline anxieties into the online sphere, or whether fear of cybercrime stands alone.

Jackson, J., Bradford, B., Brunton-Smith, I., and Gray, E. (2018) ‘In the eye of the (motivated) beholder: Towards a motivated cognition perspective on disorder perceptions’ in Lee, M., and Mythen. G. (eds) International handbook of fear of crime. Routledge. Oxon.

Neighbourhood disorder refers to those cues in one’s social and physical environment that signal first the erosion of shared commitments to dominant norms and values, and second the failure of community members and authorities to regulate behaviour in public space. Disorder is dependent on an individual defining his or her surroundings and a number of U.S. studies have examined factors related to disorder perceptions. Our goal in this chapter is to present the findings from two U.K. studies into the instrumental and relational nature of public judgements about what characterises disorder. We frame our discussion in the context of psychological work on motivated social cognition – i.e. the ways in which various psychological needs, goals and desires (a) shape information processing and (b) lead to conclusions that individuals wish to reach rather than ones demanded by adherence to logic and/or evidence. We argue that disorder may not only be ‘in the eye of the beholder,’ it may also be ‘in the eye of the motivated beholder.’

McCarthy, D., and Brunton-Smith, I. (2017) ‘Prisoner-family ties during imprisonment: Reassessing resettlement outcomes’. Prison Service Journal, 233: 23-27.

Prison life can be hard time for both those serving time and for their families on the outside. Prisoners who maintain ties with family members during their sentence can often see their relationships tested by the physical isolation and social strains which imprisonment brings and the value of a family support network for prisoners has been recognised across a number of prison service policies. Successive studies have shown that familial ties are important for prisoners as a mode of social support during their sentence, as a motivation to behave inside prison in order to improve their chances of early-release, as well as for resettlement outcomes including finding accommodation, desisting from drug use, and reducing reoffending risk.  Despite these important positive outcomes, few studies have sought to understand what actually happens to prisoner family relationships across the course of a sentence.

During any prison sentence a lot can happen to an offender, whether it be anxiety adapting to a sentence, victimisation, loss of privileges, or a host of other events which may impact on the overall experience of confinement. These experiences no doubt are dynamic and open to change, not least because some prisoners are able to adapt to their sentence more effectively than others.  They also have obvious implications for ties with family. For the families of offenders too, life paths may change—family members may die, new romantic relations may be developed, and children may be born. Taking stock of these factors, policy makers require a clearer insight into whether or not prisoner-family ties change during a prison sentence, and what the implications of these changes are for resettlement outcomes such as reoffending, drug use after release, and chances of gaining employment.

Brunton-Smith, I., and Tarling, R. (2017) ‘Harnessing paradata and multilevel multiple imputation when analysing longitudinal survey data’. International Journal of Social Research Methodology, 20 (6): 709-720.

Missing data (attrition and non-response) are a feature of most surveys especially longitudinal/panel studies. And many such studies now have multilevel designs and hence multilevel data structures. Recent advances in imputation methodology now offer social researchers opportunities to address issues of missing data in a statistically principled way. Paradata can offer great insights in understanding the nature and causes of missingness and can be used to construct auxiliary variables to be included in imputation models. In this paper we present multilevel multiple imputation which has recently extended MI to incorporate multilevel data, making it a flexible and robust strategy for many research settings. We illustrate the procedures by analysing data drawn from a longitudinal study of prisoners. We show how paradata of that study was instrumental in guiding our approach and subsequent analysis.

Sturgis, P., Williams, J., Brunton-Smith, I., and Moore, J. (2017) ‘Fieldwork effort, response rate and the distribution of survey outcomes: a multi-level meta-analysis’. Public Opinion Quarterly. 81 (2): 5223-542.

This study assesses how survey outcome distributions change over repeated calls made to addresses in face-to-face household interview surveys. We consider this question for 541 survey variables, drawn from six major face-to-face UK surveys that have different sample designs, cover different topic areas, and achieve response rates between 54 and 76 percent. Using a multilevel meta-analytic framework, we estimate for each survey variable the expected difference between the point estimate for a proportion at call n  and for the full achieved sample. Results show that most variables are surprisingly close to the final achieved sample distribution after only one or two call attempts and before any post-stratification weighting has been applied; the mean expected difference from the final sample proportion across all 541 variables after one call is 1.6 percent, dropping to 0.7 percent after three calls and to 0.4 percent after five calls. These estimates vary only marginally across the six surveys and the different types of questions examined. Our findings add weight to the body of evidence that questions the strength of the relationship between response rate and nonresponse bias. In practical terms, our results suggest that making large numbers of calls at sampled addresses and converting “soft” refusals into interviews are not cost-effective means of minimizing survey error.

Sindall, K., McCarthy, D., and Brunton-Smith, I. (2017) ‘Young people and the formation of attitudes towards the police’ European Journal of Criminology. 14 (3): 344-364.

Much is now known about public trust and confidence in the police, especially regarding the important role of procedural justice in police–citizen engagements. However, less is known about perceptions of the police amongst young people and how their views are formed. We use survey data from more than 1500 young people aged 10–15 years whose parents were also interviewed in the Crime Survey for England and Wales (2010–12) to explore the extent that children’s views of the police correspond with those of their parents. We find a strong and consistent link between the views of children and their parents – a relationship moderated by perceptions of police visibility, experience of victimization and the age of the child.

Brunton-Smith, I., Sturgis, P., and Leckie, G. (2017) ‘Detecting and understanding interviewer effects on survey data by using a cross-classified mixed effects location-scale model’. Journal of the Royal Statistical Society, Series A. 180 (2): 551-568.

We propose a cross-classified mixed effects location–scale model for the analysis of interviewer effects in survey data. The model extends the standard two-way cross-classified random-intercept model (respondents nested in interviewers crossed with areas) by specifying the residual variance to be a function of covariates and an additional interviewer random effect. This extension provides a way to study interviewers’ effects on not just the ‘location’ (mean) of respondents’ responses, but additionally on their ‘scale’ (variability). It therefore allows researchers to address new questions such as ‘Do interviewers influence the variability of their respondents’ responses in addition to their average, and if so why?’. In doing so, the model facilitates a more complete and flexible assessment of the factors that are associated with interviewer error. We illustrate this model by using data from wave 3 of the UK Household Longitudinal Survey, which we link to a range of interviewer characteristics measured in an independent survey of interviewers. By identifying both interviewer characteristics in general, but also specific interviewers who are associated with unusually high or low or homogeneous or heterogeneous responses, the model provides a way to inform improvements to survey quality.

Skudder, H., Druckman, A., Cole, J., Ansaloni, G. P., Brunton-Smith, I., and McInnes, A. (2017) ‘Addressing the carbon-crime blind spot: A carbon footprint approach’. Journal of Industrial Ecology. 21 (4): 797-1041.

Governments estimate the social and economic impacts of crime, but its environmental impact is largely unacknowledged. Our study addresses this by estimating the carbon footprint of crime in England and Wales and identifies the largest sources of emissions. By applying environmentally extended input‐output analysis–derived carbon emission factors to the monetized costs of crime, we estimate that crime committed in 2011 in England and Wales gave rise to over 4 million tonnes of carbon dioxide equivalents. Burglary resulted in the largest proportion of the total footprint (30%), because of the carbon associated with replacing stolen/damaged goods. Emissions arising from criminal justice system services also accounted for a large proportion (21% of all offenses; 49% of police recorded offenses). Focus on these offenses and the carbon efficiency of these services may help reduce the overall emissions that result from crime. However, cutting crime does not automatically result in a net reduction in carbon, given that we need to take account of potential rebound effects. As an example, we consider the impact of reducing domestic burglary by 5%. Calculating this is inherently uncertain given that it depends on assumptions concerning how money would be spent in the absence of crime. We find the most likely rebound effect (our medium estimate) is an increase in emissions of 2%. Despite this uncertainty concerning carbon savings, our study goes some way toward informing policy makers of the scale of the environmental consequences of crime and thus enables it to be taken into account in policy appraisals.

Brunton-Smith, I., and McCarthy, D. (2017) ‘The effects of prisoner attachment to family on re-entry outcomes: A longitudinal assessment’. British Journal of Criminology. 57 (2): 463-482.

Strong family support networks are regularly identified in the search for effective inhibitors of criminal behaviour but have rarely been empirically examined in the context of the prison population. Furthermore, we know little about the factors that may weaken or indeed enhance these bonds during a prison sentence. Using data from a longitudinal survey of male prisoners in England and Wales, we address this deficit. We show that visits from parents are influential in improving prisoners’ relations with their family. Furthermore, those prisoners that experience improved family relations are significantly less likely to reoffend whilst also being more likely to find work and desist from class A drug use.

Brunton-Smith, I., and McCarthy, D. (2016) ‘The social characteristics of online piracy offenders and their offending pathways’. Victims and Offenders. 11 (4): 509-533.

The Internet has been widely acknowledged as facilitating many forms of youth offending. Existing research has identified important drivers of young people’s involvement in online crime, yet this has overwhelmingly relied on school or college samples. As such, it tells us little about those young people that have left the formal education system—a group who are more likely perpetrators of juvenile crime more generally. Focusing on young people’s involvement in online piracy offenses, our analysis draws on data from a nationally representative survey of England and Wales to better understand the dynamics of involvement in online crime across the population. We assess the potential overlaps between online and offline offending, the role of differential association and deviancy neutralization techniques in shaping offending behavior, as well as the protective effect of strong family support networks in reducing involvement in piracy. We find that illegal downloaders tend to be young, male, and have a higher number of delinquent friends. We also find that many of these offenders do not confine their offending to online spaces, with involvement in offline property offenses also high among this group.

Brunton-Smith, I., and McCarthy, D. (2016) ‘Prison legitimacy and procedural fairness: the view from prisoners across England and Wales’. Justice Quarterly. 33 (6): 1029-1054.

The procedural justice model has been widely used as an explanation for understanding legitimacy and compliance with the law, particularly within the context of policing. Central to this model is the importance of procedural fairness—in which the treatment of citizens and offenders by criminal justice agents can play a key role in building legitimacy and influencing compliance with legal rules and values. This paper examines the relationship between procedural fairness and legitimacy within the context of corrections. Drawing on data from a longitudinal survey of more than 3,000 prisoners across England and Wales, we identify an important link between procedural fairness and prisoner perceptions of legitimacy. We further examine variations in legitimacy in terms of individual prisoner characteristics, conditions within prison, as well as differences between prisons.

Saunders, C., Brunton-Smith, I., Garrod, B., Stepanek, M., Vignoles, A., and Sutherland, A. (2016) ‘Missing Key Stage 2 data in LSYPE2: technical report.’ Report for the Department for Education.

The Second Longitudinal Study of Young People in England (LSYPE2) provides a resource for evidence-based policy development. A significant barrier to achieving this purpose could be the missingness present in LSYPE2, owing to a boycott of Key Stage 2 (KS2) testing in 2010. Specifically, in 2010, 15,518 maintained schools were expected to administer KS2 tests, but around one-quarter (4,005 schools; 26 per cent) of these did not administer it. Boycotts of national tests leave gaps in pupils’ attainment records and, in the case of LSYPE2, threaten to undermine a large-scale longitudinal study with substantial policy relevance. This project sought to find a way to calculate values for pupils who attended schools that boycotted KS2 tests in 2010 and/or partly mitigate the effect of the boycott on this study.

Prior attainment data is something that should be incorporated into even basic analyses of LSYPE2, and so analysts and researchers need to decide whether the missing data arising from the boycott will cause difficulties when they are making inferences and conclusions in their work, and they then need to take appropriate steps to deal with these difficulties if necessary.

The results of the analyses undertaken for this project suggest that complete-cases analyses using only pupil-level data that include a random effect for primary sampling unit (i.e. secondary school) should be unbiased. Comparing complete-case analysis with multiple imputation (MI) suggests that MI would be more efficient than the complete-case approach – i.e. standard errors (SE) would be smaller, meaning this approach should be used if statistical inference is the aim of a given analysis.

In this report we present an introduction to LSYPE2 and the KS2 Standard Assessment Tests (SATS) boycott in Chapter 1 and to statistical issues with missing data, and to methods for addressing these in Chapter 2. In Chapters 3 and 4 we explore predictors of missing KS2 test scores among LSYPE2 cohort participants, as well as predictors of KS2 attainment. Chapters 6 to 9 describe the methodological approaches and challenges to the MI and inverse probability weighting (IPW) approaches taken. Sensitivity analyses are presented in the appendix to this report. The user guide accompanying this report walks potential users of the imputed/inverse probability weighting variables through descriptive and multivariable analyses.

We have taken a pragmatic approach to this work, balancing the need for practical solutions for analysts with the desire to exhaustively explore options for dealing with missingness. This report includes an assessment of the strengths and limitations of the approaches taken, in particular, the assumptions made in the development of the MI data.

Davies, M., Brunton-Smith, I., Pina-Sanchez, J., and Sutherland, A. (2016) ‘Assessing the impact and implementation of the Sentencing Council’s Drug Offences Definitive Guideline'. London: Office of the Sentencing Council.

Analysis of trend data, disposals data and survey data was used to assess the impact of the Sentencing Council’s Drug Offences Definitive Guideline. This was the first guideline on these offences which covered both the Crown Court and the magistrates’ courts, coming into force in February 2012. The analysis focused on the effect of the guideline on sentence outcomes.

Stoneman, P., and Brunton-Smith, I. (2016) ‘Chapter 5 - The foundations of quantitative research’ in Stoneman, P., and Gilbert, N. (eds) Researching Social Life, 3rd Edition.

Key points: 

Quantitative research is an essential approach to social research which allows us to challenge or support different ideas about social reality.

While today’s quantitative researchers are positivists in origin, the vast majority are very much aware of the limitations of quantitative research in terms of describing and explaining the full extent of social reality.

The strengths of quantitative research are based on some basic scientific credentials – it can allow us to generalise findings, explain phenomena by citing causal mechanisms, and sometimes it affords us the ability to predict future scenarios.

Quantitative research is best used when you have a research question which requires you to describe variance in a population, measure degrees of difference and/or test hypotheses for why something has occurred.

Quantitative research should always follow a clear research design, starting from clear conceptual definitions and working towards the testing of hypotheses which have been clearly derived from a theoretical framework.

A lot of quantitative data has already been collected by others (known as secondary data) and much of this is publicly available for you to use. A great variety of data is available, covering topics such as employment, health, crime, education, social attitudes, political behaviour, and expenditure data (on items like food and energy use).

Bullock, K., Meadows, R., and Brunton-Smith, I. (2014) ‘‘DiscoverQuants’: Integrating quantitative methods (QM) and substantive teaching for first year undergraduate sociology students’ Enhancing Learning in the Social Sciences, 6 (2): 6-20.

This paper considers the rationale for, design and outputs of a project, based at the University of Surrey UK and funded by the Economic and Social Research Council (ESRC), which sought to integrate aspects of teaching substantive and Quantitative Methods (QM) teaching across first year sociology undergraduate programmes using a blended approach. The paper considers the nature of concerns regarding teaching QM within social science undergraduate programmes. It goes on to describe the rationale for this project, its design and its primary outputs. We consider a range of data related to student attitudes towards studying QM at university as well as their perspectives on the project and the implications for practice.

Brunton-Smith, I., Carpenter, J., Kenward, M., and Tarling, R. (2014) ‘Multiple Imputation for handling missing data in social research’ Social Research Update, no 66.

Missing data are ubiquitous in quantitative social research and can lead to incorrect inferences. Statistically principled approaches have been developed to address the problem of missing data. Of the available approaches, we favour multiple imputation (MI) for its flexibility, accessibility and ease of use. MI is described and a worked example, using the statistical software Stata, is presented.

Brunton-Smith, I., Sutherland, A., and Jackson, J. (2014) ‘Bridging structure and perception; On the social ecology of beliefs and worries about neighbourhood violence in London’ British Journal of Criminology. 54 (4): 503-526.

Applying Robert Sampson’s (2012) work on interdependent spatial patterns in a new setting, we link structural characteristics of the neighbourhood to public beliefs and worries about neighbourhood violence via two intermediate mechanisms: (1) collective efficacy and (2) neighbourhood disorder. Analysing data from face-to-face interviews of 61,436 individuals living in 4,761 London neighbourhoods, we find that the strength of informal social control mechanisms and the extent of low-level breaches of common standards of behaviour communicate information about the prevalence and threat of violent crime in one’s neighbourhood. Moreover, collective efficacy partially mediates many of the statistical effects of structural characteristics of the neighbourhood on beliefs and worries about violent crime. Theoretical implications of the findings are discussed.

Sturgis, P., Brunton-Smith, I., Jackson, J., and Kuha, J. (2014) ‘Ethnic diversity and the social cohesion of neighbourhoods in London’ Ethnic and Racial Studies, 37 (8): 1286-1309.

The question of whether and how ethnic diversity affects the social cohesion of communities has become an increasingly prominent and contested topic of academic and political debate. In this paper we focus on a single city: London. As possibly the most ethnically diverse conurbation on the planet, London serves as a particularly suitable test-bed for theories about the effects of ethnic heterogeneity on prosocial attitudes. We find neighbourhood ethnic diversity in London to be positively related to the perceived social cohesion of neighbourhood residents, once the level of economic deprivation is accounted for. Ethnic segregation within neighbourhoods, on the other hand, is associated with lower levels of perceived social cohesion. Both effects are strongly moderated by the age of individual residents: diversity has a positive effect on social cohesion for young people but this effect dissipates in older age groups; the reverse pattern is found for ethnic segregation.

Barrett, M., and Brunton-Smith, I. (2014) ‘Political and civic participation: Towards an integrative perspective’ Journal of Civil Society, 10 (1): 5-28.

This paper presents an overview of current understandings in the study of political and civic participation, drawing in particular on innovations which have emerged from the PIDOP project. The different forms that political and civic participation can take are outlined, and the factors that are related to different patterns of participation are reviewed. These factors operate at many different levels, and include distal macro contextual factors (e.g., electoral, political and legal institutions, and the historical, economic and cultural characteristics of a country), demographic factors (e.g., SES, ethnicity and gender), proximal social factors (e.g., factors stemming from the family, education, the peer group, the workplace, the mass media and non-political organisations) and endogenous psychological factors (e.g., political knowledge, anger towards perceived social injustices, internal efficacy, external efficacy, institutional trust, and motivations for participating). Some findings from the secondary analysis of existing datasets (including the European Social Survey and Eurobarometer) in the PIDOP project are reported. These findings show that forms of participation vary as a function of complex interactions between different macro, demographic and psychological factors. It is argued that a multi-level integrative theory which takes into account the specific circumstances of particular demographic subgroups living within particular national and cultural contexts is required to understand the drivers of political and civic participation, and that policies and interventions aimed at enhancing citizens’ levels of participation need to take this multi-level complexity into account.

Brunton-Smith, I., and Barrett, M. (2014) ‘Political and civic participation: findings from the modelling of existing survey data sets’ in Barrett, M., and Zani, B. (eds) Political and Civic Participation: Theories and Findings from the PIDOP Project. Routledge.

In this chapter, we examine four broad categories of political and civic participation: Voting; Other forms of conventional political activity, such as contacting a politician, being a member of/working for a political party, donating money to a political organization, wearing a political party campaign badge, etc.; Non-conventional political activity, such as participating in lawful demonstrations and illegal protests, buying/boycotting certain products, signing petitions, etc.; Civic engagement, such as being involved in a social club, education or teaching group, religious or church organization, cultural or hobby group, sports or outdoor activity club, environmental or humanitarian organisation, business or professional group, or a trade union.

While these various forms of participation are often examined through the collection of new data, they can also be investigated using existing survey data sets. Indeed, the European Union and its constituent national governments have invested considerable resources in the collection of survey data relating to the extent and nature of political and civic participation amongst its citizens (e.g., the European Social Survey; International Social Survey Programme; Eurobarometer; and World Values Survey). This has provided researchers with a formidable resource for the empirical examination of the complex range of political, social, and psychological factors that influence people’s tendency to vote, participate in other forms of conventional and non-conventional political activity, and be civically engaged citizens.

Using these existing data sets has a number of distinct advantages over collecting new data. The data sets are based on representative samples from different countries, which means that the differences which emerge from cross-national comparisons cannot be attributed to sampling biases within countries; the large number of participants who have contributed the data means that powerful statistical techniques can be used to analyse the data; the large number of participants also facilitates fine-grained analysis across key demographic sub-groups of interest; and the secondary analysis of existing data, in comparison to the collection and analysis of new data, is a highly cost-efficient research strategy. 

Brunton-Smith, I., and Hopkins, K. (2014) ‘The impact of experience in prison on the employment status of prisoners after release: Findings from the first 3 waves of Surveying Prisoner Crime Reduction (SPCR)’. Report for Ministry of Justice.

The importance of employment in supporting reducing re-offending has long been recognised.

There is less understanding of why some prisoners are able to secure work, whilst others do not. Improving our understanding is a key priority for those involved in the management and rehabilitation of offenders.

This report presents findings from data gathered during SPCR about factors, such as programmes and interventions in prison, that are associated with employment after release, for longer-sentenced prisoners (between 18 months and four years).

Hopkins, K., and Brunton-Smith, I. (2014) ‘Prisoners’ experience of prison and outcomes on release: Results from Surveying Prisoner Crime Reduction (SPCR).' Report for Ministry of Justice.

This report summarises the key results from Wave 2 (in-custody, pre-release) and Wave 3 (post-custody) of the Surveying Prisoner Crime Reduction survey. Using interviews with prisoners the report examines prison routine, prisoners’ expectations of life after custody and actual outcomes on release, including employment, accommodation, drugs and alcohol, and finance, benefits and debt.

Brunton-Smith, I., Carpenter, J., Kenward, M., and Tarling, R. (2014) ‘Surveying Prisoner Crime Reduction (SPCR) Using multiple imputation to recover missing data from the SPCR’. Report for Ministry of Justice.

This report is concerned with a large longitudinal survey, Surveying Prisoner Crime Reduction (SPCR), and the recovery of ‘missing’ data from the survey. SPCR involved interviewing a large group of prisoners during and after their prison sentences. In some cases, the interviews were not conducted, as the prisoner could not be contacted or did not want to be interviewed. This meant that the prisoners’ answers were ‘missing’ from the dataset, meaning that the dataset was smaller than planned, and potentially biased. This report explains how statistical techniques were used to ‘recover’ the missing data, where possible, allowing more robust analysis of the survey data to be conducted, and more rigorous findings to be produced.

Sturgis, P., and Brunton-Smith, I. (2014) ‘Examining potential bias in the sampling methodology for the Green Deal Assessment Survey’. Report for the Department of Energy and Climate Change.
Sutherland, A., Brunton-Smith, I., and Jackson, J. (2013) ‘Collective efficacy and violence in London’ British Journal of Criminology, 53 (6): 1050-1074.

This paper examines the importance of neighbourhood context in explaining violence in London. Exploring in a new context Sampson’s work on the relationship between interdependent spatial patterns of concentrated disadvantage and crime, we assess whether collective efficacy (i.e. shared expectations about norms, values and goals, as well as the ability of members of the community to realize these goals) mediates the potential impact on violence of neighbourhood deprivation, residential stability and population heterogeneity. Reporting findings from a dataset based on face-to-face interviews with 60,000 individuals living in 4,700 London neighbourhoods, we find that collective efficacy is negatively related to police-recorded violence. But, unlike previous research, we find that collective efficacy does not mediate the statistical relationship between structural characteristics of the neighbourhood and violence. After finding that collective efficacy is unrelated to an alternative measure of neighbourhood violence, we discuss limitations and possible explanations for our results, before setting out plans for further research.

Brunton-Smith, I., and Hopkins, K. (2013) ‘The factors associated with reconviction following release from prison: Findings from the first 3 waves of Surveying Prisoner Crime Reduction (SPCR). Report for Ministry of Justice.

This study looked at risk factors before prison, experiences of prison, and outcomes on release amongst a sub-sample of SPCR which was matched to the Police National Computer (PNC). Associations between these factors and proven re-offending one and two years after release from prison were identified. These included pre-custody factors such as: employment and accommodation status; drug use; and criminal history. They also included experiences of prison: being worried/confused; attending paid work; prison punishments; contact with family; and prison interventions. Post-release outcomes such as employment and accommodation status and drug use were also explored. All the risk factors were entered into a model to test for independent associations with re-offending. Factors remaining in the model included criminal history, employment and accommodation status before custody, being worried/confused about prison, receiving additional punishments, and reporting using Class A drugs after release.

Brunton-Smith, I., Sturgis, P., and Williams, J. (2012) ‘Is success on the doorstep correlated with the magnitude of the interviewer design effect?’ Public Opinion Quarterly, 76 (2): 265-286.

Evidence is now beginning to accumulate that shows that interviewer attitudes, personality, and behavior are predictive of success in achieving contact and cooperation with sampled households. A less frequently explored possibility, however, is that these same characteristics might also be the source of variability in the extent to which interviewers follow best practices in the implementation of standardized interviewing. That is to say, there may be a correlation between interviewer-induced nonresponse bias and measurement error. In this article, we provide the first empirical investigation of the direction and magnitude of the relationship between interviewer skill in obtaining contact and cooperation and correlated interviewer error. Drawing on face-to-face interview data from a large, multistage probability sample of the British population, we use cross-classified multilevel models with a complex error structure to examine how the interviewer variance component varies as a function of historical measures of interviewer skill in obtaining contact and cooperation. Our results show that, across a broad range of variables, interviewers with a history of obtaining poor rates of contact and cooperation exhibit higher levels of correlated interviewer error than their better-performing colleagues. For cooperation, we find some evidence of a U-shaped relationship, with the least and  the most successful interviewers having the largest interviewer variance component.

Brunton-Smith, I., and Sturgis, P. (2011) ‘Do neighborhoods generate fear of crime?: An empirical test using the British Crime Survey’. Criminology, 49 (2): 331-369.

For a long time, criminologists have contended that neighborhoods are important determinants of how individuals perceive their risk of criminal victimization. Yet, despite the theoretical importance and policy relevance of these claims, the empirical evidence base is surprisingly thin and inconsistent. Drawing on data from a national probability sample of individuals, linked to independent measures of neighborhood demographic characteristics, visual signs of physical disorder, and reported crime, we test four hypotheses about the mechanisms through which neighborhoods influence fear of crime. Our large sample size, analytical approach, and the independence of our empirical measures enable us to overcome some of the limitations that have hampered much previous research into this question. We find that neighborhood structural characteristics, visual signs of disorder, and recorded crime all have direct and independent effects on individual-level fear of crime. Additionally, we demonstrate that individual differences in fear of crime are strongly moderated by neighborhood socioeconomic characteristics; between group differences in expressed fear of crime are both exacerbated and ameliorated by the characteristics of the areas in which people live.

Brunton-Smith, I. (2011) ‘Untangling the relationship between fear of crime and perceived disorder: Evidence from a longitudinal study of young people in England and Wales’. British Journal of Criminology, 51 (6): 885-899.

Over the last 40 years and more, a growing number of researchers have explored the links between perceptions of disorder and fear of criminal victimization. Many of these studies have posited a causal link from perceptions of disorder to subsequent fear, with disorderly cues in the environment signalling to individuals that an area is in decline and unable to control deviant behaviour. But a growing body of evidence approaches this question from the opposite direction, emphasizing the socially constructed nature of perceived disorder and the potential role that fear may have in giving meaning to ambiguous disorderly cues present in the environment. This conceptual uncertainty stems, in part, from the reliance of existing research on cross-sectional data, making it impossible to say whether it is perceptions of disorder that shape fear or whether fear drives perceived disorder. A cross-lagged panel design is applied to longitudinal data from the Offending Crime and Justice Survey to more carefully explore the causal links between fear and disorder.

Sturgis, P., Brunton-Smith, I., Read, S., and Allum, N. (2011) ‘Does ethnic diversity erode trust?: Putnam’s ‘Hunkering-Down’ thesis reconsidered’. British Journal of Political Science, 41 (1): 57-82.

We use a multi-level modelling approach to estimate the effect of ethnic diversity on measures of generalized and strategic trust using data from a new survey in Britain with a sample size approaching 25,000 individuals. In addition to the ethnic diversity of neighbourhoods, we incorporate a range of indicators of the socio-economic characteristics of individuals and the areas in which they live. Our results show no effect of ethnic diversity on generalized trust. There is a statistically significant association between diversity and a measure of strategic trust, but in substantive terms, the effect is trivial and dwarfed by the effects of economic deprivation and the social connectedness of individuals.

Sturgis, P., and Brunton-Smith, I. (2011) ‘An evaluation of the potential utility of interviewer observation variables for reducing nonresponse error in the National Survey for Wales’. Report for the Welsh Assembly Government.
Sturgis, P., Brunton-Smith, I., and Fife-Schaw, C. (2010) ‘Public attitudes to genomic science: An experiment in information provision’. Public Understanding of Science, 19 (2): 166-180.

We use an experimental panel study design to investigate the effect of providing “value-neutral” information about genomic science in the form of a short film to a random sample of the British public. We find little evidence of attitude change as a function of information provision. However, our results show that information provision significantly increased dropout from the study amongst less educated respondents. Our findings have implications both for our understanding of the knowledge–attitude relationship in public opinion toward genomic science and for science communication more generally.

Allum, N., Sturgis, P., Tabourazi, D., and Brunton-Smith, I. (2008) ‘Science knowledge and attitudes across cultures: A meta-analysis’. Public Understanding of Science, 17 (1): 35-54.

The correlation between knowledge and attitudes has been the source of controversy in research on the public understanding of science (PUS). Although many studies, both quantitative and qualitative, have examined this issue, the results are at best diverse and at worst contradictory. In this paper, we review the evidence on the relationship between public attitudes and public knowledge about science across 40 countries using a meta-analytic approach. We fit multilevel models to data from 193 nationally representative surveys on PUS carried out since 1989. We find a small positive correlation between general attitudes towards science and general knowledge of scientific facts, after controlling for a range of possible confounding variables. This general relationship varies little across cultures but more substantially between different domains of science and technology. Our results suggest that PUS research needs to focus on understanding the mechanisms that underlie the clear association that exists between knowledge and attitudes about science.

Brunton-Smith, I., Buil-Gil, D., Pina-Sánchez, J., Cernat, A., Moretti, A. (2024) ‘Using synthetic crime data to understand patterns of police under-counting at the local level’, in Huey, L, and Buil-Gil, D (eds) The Crime Data Handbook.
Pina-Sánchez, J., Brunton-Smith, I., Buil-Gil, D., Cernat, A. (2023) ‘Exploring the Impact of Measurement Error in Police Recorded Crime Rates through Sensitivity Analysis’. Crime Science. 12(14).

It is well known that police recorded crime data is susceptible to substantial measurement error. However, despite its limitations, police data is widely used in regression models exploring the causes and effects of crime, which can lead to different types of bias. Here, we introduce a new R package (‘rcme’: Recounting Crime with Measurement Error) that can be used to facilitate sensitivity assessments of the impact of measurement error in analyses using police recorded crime rates across a wide range of settings. To demonstrate the potential of such sensitivity analysis, we explore the robustness of the effect of collective efficacy on criminal damage across Greater London’s neighbourhoods. We show how the crime reduction effect attributed to collective efficacy appears robust, even when most criminal damage incidents are not recorded by the police, and if we accept that under-recording rates are moderately affected by collective efficacy.

Brunton-Smith, I., Cernat, A., Pina-Sánchez, J., Buil-Gil, D. (2024) 'Estimating the reliability of crime data in geographic areas'. British Journal of Criminology, Revise and Resubmit.

Crime data is problematic: Crimes that are never reported undermine its validity and differences in police recording practices affect its reliability. However, the true extent of these problems is not well known, with existing studies suffering from a number of methodological limitations. We examine the quality of police recorded crime data and survey-based crime estimates recorded in England and Wales using a robust latent trait model that effectively represents the competing sources of error. We find that whilst crime rates derived from police data systematically underestimate the true extent of crime, they are substantially more reliable than estimates from survey data. Reliability is lower for violence and criminal damage and is getting worse over time. 

Sturgis, P., and Brunton-Smith, I. (2023) ‘Personality and survey satisficing’. Public Opinion Quarterly. Online first.

In this paper, we consider the role of personality as a component of motivation in promoting or inhibiting the tendency to exhibit the satisficing response styles of midpoint, straightlining, and Don’t Know responding. We assess whether respondents who are low on the Conscientiousness and Agreeableness dimensions of the Big Five Personality Inventory are more likely to exhibit these satisficing response styles. We find large effects of these personality dimensions on the propensity to satisfice in both face-to-face and self-administration modes and in probability and nonprobability samples. People who score high on Conscientiousness and Agreeableness were less likely to be in the top decile of straightlining and midpoint distributions. The findings for Don’t Know responding were weaker and only significant for Conscientiousness in the nonprobability sample. We also find large effects across all satisficing indicators for a direct measure of cognitive ability, where existing studies have mostly relied on proxy measures of ability such as educational attainment. Sensitivity analysis suggests the personality effects are likely to be causal in nature.

Sturgis, P., Brunton-Smith, I., Allum, N., and Fulsang, S. (2024) ‘Testing the cultural-invariance hypothesis: A global analysis of the relationship between scientific knowledge and attitudes to science’. PLOS ONE.

A substantial body of research has demonstrated that science knowledge is correlated with attitudes towards science, with most studies finding a positive relationship between the two constructs; people who are more knowledgeable about science tend to be more positive about it. However, this evidence base has been almost exclusively confined to high and middle-income democracies, with poorer and less developed nations excluded from consideration. In this study, we conduct the first global investigation of the science knowledge-attitude relationship, using the 2018 Wellcome Global Monitor survey. Our results show a positive knowledge-attitude correlation in all but one of the 144 countries investigated. This robust cross-national relationship is consistent across both science literacy and self-assessed measures of science knowledge.