George and Teresa Smith Award Winners 2021

ts award

We are delighted to announce the winners of the 2020- 2021 George and Teresa Smith Awards. These are awarded annually for the best theses submitted by our Master's students. Find out more about the history of the awards

George Smith Award

The George Smith Award for best thesis in the academic year 2020-21 for Comparative Social Policy has been awarded to: 

Leon David Küstermann for Afraid of automation? Do active labour market policies improve the subjective wellbeing of workers who are threatened by automation?

Teresa Smith Award

The Teresa Smith Award for best thesis in the academic year 2020-2021 for Evidence-Based Intervention and Policy Evaluation has been awarded to:

Many congratulations to Leon, Alejandro, Sarani and Sarah for this fantastic achievement amidst strong competition. You can read more about their theses below.

Afraid of automation? Do active labour market policies improve the subjective wellbeing of workers who are threatened by automation?

Leon David Küstermann

Automation causes massive disruptions in the labour market and leaves many workers with outdated skills in a difficult position between older superfluous jobs and emerging occupations. However, little is known about how automation is perceived from a worker’s perspective, on whether there are wellbeing implications for workers threatened by automation and whether suggested policy responses such as active labour market policies are perceived as supportive. Therefore, this paper seeks to answer the following research question:  How do ALMPs affect the subjective wellbeing of workers at risk of automation in Europe?
I use cross-classified multi-level regression to analyse data from the European Social Survey for 28 European countries and compare the routine task intensity (RTI) and the Frey/Osborne index as measures for automation. My results show that automation risks reduce subjective wellbeing especially in Eastern European countries. These wellbeing losses are mitigated by active labour market policies. While my assumption that training ALMPs are more effective than public employment services cannot be confirmed, I find indications that benefit sanctions undermine the effectiveness of ALMPs.

Don’t Mess With My Children? A Qualitative Evidence Synthesis Exploring Parent Perceptions of Sexuality Education

Alejandro Biondi

Development of sexuality and socialisation in gender norms impact life from an early age. However, many young people reach adulthood without access to knowledge and skills related to sexuality and equitable gender norms. This exposes them to negative outcomes that undermine their autonomy. School-based Sexuality Education (SE) can change this, as shown by growing evidence of its effectiveness. However, SE implementation remains a challenge around the world. Little is known about the factors shaping this, including the meanings that powerful stakeholders assign to SE. Parents and carers are particularly relevant, due to their significant roles in the gender socialisation of their children. Therefore, to expand access to SE for children and adolescents, it is critical to understand parental perceptions of it.

This review seeks to explore and synthesise parent and carer perceptions of school-based SE around the world. An exhaustive search was conducted. This comprised a standard search of 16 electronic sources, hand-searching of three journals, 11 websites of relevant organisations and two search engines, and contacting experts and using citation searching. To be included, studies had to sample parents/carers of children under 19, focus on SE delivered at school by adults, be primary research and qualitative or mixed methods in design, evaluate participant perceptions of SE, be available in English, Spanish, French, or Portuguese, from all world regions and in the period 1995-2021. Data from included studies was extracted and study quality was critically appraised. Data was analysed using thematic synthesis to inductively generate descriptive and analytical themes. Sensitivity, the overall confidence in the evidence and reflexivity were evaluated.

15 studies were included. 10 were rated high quality, 3 medium and 2 low. Coding of their Results and Discussion sections yielded 88 codes, and a conceptual framework with 18 descriptive and 5 analytical themes emerged from thematic synthesis. Parents expressed three types of perceptions. Descriptively, they expressed what they conceived the essence of school-based SE to be, defining it as a form of education but expressing uncertainty about its content. Their experience with absent or deficient SE in their youth, plus their internalised social norms, shaped this conceptualisation of SE. Parents also voiced arguments in favour/against SE, around themes such as SE instilling or countering their own values, protecting or harming, and promoting or disrupting their children’s well-being. Finally, parents made prescriptive judgements about SE implementation, including on its approach and objectives, coordination with stakeholders (especially, with families), curriculum, pedagogy, timing, and implementers.

These findings extend and increase the confidence in previous literature and can be contrasted with synthesised perceptions of other stakeholders (notably, students). Implications for policy include strengthening school outreach to parents, from a collaborative approach seeking to implement SE in a context-aware way. Further research into perceptions of other stakeholders, systematic reviewing of quantitative evidence and realist reviewing of implementation processes would help confirm this approach.

Making TVET Work. A Systematic Review and Meta-Analysis of Technical and Vocational Education and Training for Youth in LMICs, and its Effective Characteristics

Sarani Jaywardena


Aligned with a global development agenda, low- and middle-income countries (LMICs) are investing in technical and vocational education and training (TVET) to improve labour market outcomes for youth.


Through a systematic review and meta-analysis, this study aims to update Tripney et al.’s (2013) Campbell review on TVET for youth in LMICs with recent and rigorous evidence; extend the evidence base by exploring effective intervention characteristics; and thereby inform the design, implementation, and evaluation of TVET in LMICs.

Search methods

Search terms were formulated according to the ‘PICOS’ framework. Thirteen bibliographic, specialist, and grey literature databases, and the websites of eight organisations were searched electronically. Three key reviews and three relevant journals were hand searched for references.

Selection criteria

Experimental and quasi-experimental studies examining the impact of TVET on labour market outcomes of youth in LMICs published in English after the year 2000 were included in the search criteria. Only experimental studies were included in the review.

Data collection and analysis

Intervention characteristics were mapped based on the included studies and external literature to pre-specify subgroup analysis variables. Data was extracted and risk of bias evaluated for each study. Meta-analysis was conducted on RevMan using random effects models with the generic invariance variance method. Heterogeneity was investigated through subgroup analysis by gender and pre-specified program features. Sensitivity, publication bias, and the overall confidence of evidence were evaluated.


5941 search records were retrieved for title and abstract review and 158 studies were retrieved for full text review, of which 62 passed screening. Data was extracted only from the 22 experimental studies, and 19 were used for meta-analysis. The studies evaluated 18 TVET programs in five regions. Risk of bias was low in five studies, moderate in 13, and high in four.

TVET produced small, positive significant effects on three primary outcomes: employment status (0.02 SMD (95% CI 0.00, 0.03), moderate certainty), weekly working hours (0.61 hours (95% CI 0.05, 1.15), low certainty), and monthly earnings (5.44 USD (95% CI 1.30, 9.57), low certainty). Gender significantly affected employment, with benefits only accruing to women. TVET had positive but non-significant effects on wage-, self-, and formal employment. 

Including classroom-based learning improved earnings, while programs only for women and girls improved employment, and targeting the socio-economically disadvantaged improved working hours. Providing stipends was detrimental to employment. Inclusion of workplace-based learning or non-technical skills, duration, cost, implementing agency, and targeting urban areas did not have significant subgroup effects.

The results of all meta-analyses were robust to excluding studies with high risk of bias, but there was potential publication bias for findings on working hours and earnings.


This is the first study to synthesise research on TVET’s effectiveness for youth labour market outcomes in LMICs based only on experimental evidence, and to explore intervention characteristics. Strengthening prior findings, TVET benefits employment status, weekly working hours, and monthly earnings; but it may potentially negatively impact employment among men. Certain program features appear to improve outcomes, but these findings should be evaluated further.

Investigating equity effects of parenting programmes on child maltreatment in low- and middle-income countries. An individual participant data meta-analysis of four Parenting for Lifelong Health randomised trials

Sarah Melville


Child maltreatment (CM) is a form of interpersonal violence with widespread health and social consequences and is patterned by socioeconomic disadvantage due to differing exposures to risk factors. Evidence suggests that parenting programmes are effective at preventing CM in high-income countries (HICs) and low- and middle-income countries (LMICs), however little is known about effectiveness for subgroups in LMICs. Understanding subgroup effectiveness is important for understanding for whom parenting programmes are most effective, and hence ensuring that they do not increase socioeconomic inequities. The lack of evidence on subgroup effectiveness thus means that there is little understanding of the equity effects of parenting programmes in LMICs.


To investigate the equity effects of the Parenting for Lifelong Health (PLH) for young children (2–9 years) parenting programme implemented in LMICs by testing whether socially stratifying PROGRESS-Plus factors moderate PLH’s effectiveness in terms of CM.


  1. An individual participant data (IPD) meta-analysis to pool individual-level data from four randomised controlled trials of PLH implemented in LMICs (N=593).
  2. A moderator analysis using the pooled IPD sample to identify differential subgroup and hence equity effects of PLH in LMIC populations.


The odds of high maltreatment (the primary outcome, dichotomised from a continuous score measuring CM) decreased in all treatment groups compared to control groups (South African pilot trial: OR=0.12, 95% CI 0.02–0.56; South African main trial: OR=0.47, 95% CI 0.21–1.01; Thailand trial: OR=0.47, 95% CI 0.14–1.56; Philippines trial: OR=0.26, 95% CI 0.09–0.77; pooled sample: OR=0.35, 95% CI 0.21–0.59). No variables significantly moderated the odds of high maltreatment; PLH is thus effective for boys and girls (p-value=0.72), caregivers with less than and more than a secondary school education (p-value=0.83), single and married caregivers (p-value=0.96), caregivers with self-reported good and poor health (p-value=0.13), households that experience hunger and households that do not (p-value=0.95) and households with at least one member employed and households with no one employed (p-value=0.74). It is thus unlikely to increase inequities in populations receiving the intervention.


Findings from the pooled sample of caregivers from LMICs indicate that PLH is effective for subgroups with differing levels of disadvantage, and is thus unlikely to increase inequities in populations receiving the intervention. All factors tested were found not to moderate the odds of high maltreatment, and PLH was found to be effective at reducing high maltreatment in LMIC populations. These findings cohere with evidence on subgroup effectiveness in HICs and are somewhat similar to what little evidence exists in LMICs. More research is needed to confirm these results, and on factors that affect dimensions of equity not explored in the present study, such as initial access to and targeting of parenting programmes.