'Does the Crown Court Discriminate Against Muslim-Named Offenders? A Novel Investigation Based on Text Mining Techniques' by Jose Pina-Sánchez Julian V Roberts and Dimitrios Sferopoulos
in (2018)
The British Journal of Criminology comments
Most research in sentencing discrimination in the United Kingdom has relied on aggregate analyses comparing disparities by ethnic group. These studies fail to consider differences in the individual characteristics of the cases processed. To circumvent the lack of official data, we scraped sentence records stored in a commercial website, from which a sample of 8,437 offenders sentenced to custody in the Crown Court from 2007 to 2017 was generated. Using the names of the offenders, we have been able to classify 8.6 per cent of our sample as having a traditional Muslim name. We find that Muslim-named offenders received sentences 9.8 per cent longer than the rest of the sample. However, this difference disappeared once we accounted for the type of offence and other key case characteristics.
The authors argue
Is there evidence of discrimination at sentencing in England and Wales? The 2017 Lammy Review has provided a timely reminder of the need for more—and better—research into the criminal justice treatment of racial, ethnic and immigrant minorities in England and Wales. A significant body of research addresses the differential impact on these groups at all stages of the criminal process (e.g. Chigwada-Bailey 2003; Hood et al. 2003; Cole and Wardak 2006; Earle 2011; Phillips 2012; Phillips and Bowling 2017; Irwin-Rogers 2018). Sentencing—the most visible and symbolic stage of that process (Ashworth 2010)—has been subject to far less academic scrutiny. The most significant study of race and sentencing is now over a generation old (Hood 1992). Since then, empirical research has been intermittent,1 with the Ministry of Justice undertaking much of the work on this topic. As part of its section 95 duties, the Ministry of Justice publishes annual statistics relating to race and criminal justice, including sentencing (e.g. Ministry of Justice 2017). These reports provide bivariate statistics, highlighting relationships between race and sentencing outcomes, but are unable to control for relevant case characteristics that might explain those relationships.
The Lammy Review demonstrated racial disparities in sentencing outcomes for certain offence categories. More specifically, the review reported that within drug offences the odds of a prison sentence were 240 per cent higher for defendants who self-identify as Black, Asian or Minority Ethnic (BAME) compared with White defendants. The review’s analysis took some relevant case characteristics into account. For example, previous convictions and plea were considered, but not other mitigating or aggravating circumstances or indeed the possibility that BAME drug offenders had been convicted of more serious drug crimes (see Hopkins et al. 2016; Lammy 2017: 33). Regrettably, the Lammy Review failed to conduct or commission original empirical research, which might have accounted for other relevant case characteristics. Nor did the review draw upon existing databases that could have helped answering the key question of whether and to what extent racial minorities are treated differently.
The Ministry of Justice biennial report ‘Statistics on Race and the Criminal Justice System’ has consistently documented sentencing differentials between BAME defendants and White defendants accused of the same offence. The most recent Ministry report (2017) found that BAME defendants had a higher custody rate than White defendants. In addition, since 2012, the average sentence length has been consistently longer for all non-White ethnic groups. In 2016, of all offenders sentenced to immediate custody, Black and Asian offenders received an average sentence length of 24 and 25 months, respectively, compared with 18 months for White offenders (Ministry of Justice 2017: 53). These disparities are cause for concern (see The Secret Barrister 2018: 285). However, they do not constitute incontrovertible evidence of discrimination at the sentencing stage since we do not know whether the differential outcomes can be explained by legally relevant factors2 (Green 1961; Hall and Simkus 1975; Raynor and Lewis 2011; Pina-Sánchez and Linacre 2016) such as those determining the harm of the offence or the culpability of the offender. If for example, BAME defendants were less likely to plead guilty—as suggested by Thomas (2010) and Hood (1992)—we would expect to see differences in sentencing outcomes, all other characteristics being equal.
This methodological challenge is not new. In 1987, Zatz described the comparison of group means as an obsolete approach to investigate sentence disparities. Multivariate approaches are superior when it comes to detecting the presence of discrimination in sentencing. These methods can be used to control simultaneously for the relevant aggravating and mitigating factors present in different cases, which is key to be able to distinguish legitimate disparities in sentencing from truly discriminatory practices. Hundreds of such studies have been conducted in the United States, with regression modelling being the predominant method of choice (Baumer 2013). In the United Kingdom, however, official sentencing data have traditionally been presented in an aggregated format, precluding the use of regression modelling techniques. In response to critics,3 the Ministry of Justice released a large data set of 1.2 million cases sentenced from 2007 to 2011 at the magistrates and Crown Courts.4 These individual cases included some important demographic characteristics of the defendant, such as age, gender and ethnic group; however, they did not contain any relevant case characteristics other than the broad offence type, thus preventing researchers from differentiating between warranted and unwarranted disparities.
