Publication | U4 Brief

Anti-corruption and integrity training: learning how to resist corruption

Reconsidering the role of integrity training

More and better training is central to a collective response to corruption. Anti-corruption initiatives typically prescribe and implement integrity trainings, covering compliance, ethics, and anti-corruption principles, as major components of their programming. The UN Convention Against Corruption (2003) and the OECD Anti-Bribery Convention (1997) highlight training and education as a main engagement area. The new international anti-bribery management standard, ISO 37001, assigns a central role to training for organisations in both the public and private sectors,f464f0b6829d while the new European rules for corporate sustainability reporting ask for more granular disclosure on anti-corruption training activities.c01ca70e5f54

Despite the importance of integrity training, researchers and practitioners have limited insight into what works and what does not. The evidence base is fragmented, and studies rarely provide specific details about the training they examine. Integrity training could be improved by drawing on novel insights and practices in other fields, but an exploration of this topic is largely absent from the anti-corruption conversation in both public and private sector organisations.

This U4 Brief summarises recent empirical literature on integrity training produced between 2013 and 2023 and draws out some of most pertinent insights. On this foundation, it suggests pathways for future learning and ideas for innovation to make integrity training more effective and efficient. The discussion should be particularly useful for donor agencies that are engaged in integrity training and wish to explore evidence and ideas on how to further refine their activities in this area.

Black boxes and blind spots

A deep dive into the academic and policy literature of the last ten years has yielded 24 studies and a small set of meta-reviews that empirically explore the effectiveness of integrity training. Taken together, these studies are characterised by three basic shortcomings:

  • A persistent black box. Most studies do not provide detailed descriptions of the training they evaluated, thus limiting their value as a guide to curriculum development.
  • Short time horizons. Studies that seek to capture the effects of training rarely look beyond one and a half years. This precludes insights on longer-term impacts, repeat trainings, and the potential benefits of integrating training with long-term strategies.
  • A gender blind spot. Corruption risks are gender-specific, and gender-specific learning and teaching styles influence the effectiveness of training. Yet none of the studies include a sufficient focus on gender, and many do not consider this crucial dimension at all.

Despite these gaps, intriguing insights have emerged from corruption-specific studies, as well as from research on improving organisational culture and professional conduct in a broad range of fields. Some of the principal insights relate to the types of integrity training, the measurement of impact, and evidence of effectiveness, as detailed in the next three sections.

Different types of training suit different objectives and audiences

The field of integrity training is marked by an astounding diversity of content, formats, and objectives.02cef2313413 For example, a review of 95 business ethics programmes identifies no less than 41 different theoretical approaches in use.42053b5ee579 Moreover, diversity in training approaches is judged to be on the rise.a98e4049adc4

Several authors provide frameworks to organise and discuss different types of integrity training, usually oriented around models of learning,1958a3498dcb ethical behaviour and development,9fefcff2343d or some combination of these.1b65157978c2 Figure 1 summarises different training designs, classified by the types of objectives they pursue.

Matching anti-corruption training with strategic objectives

Source: Dieter Zinnbauer

Specific training programmes mix and prioritise among these objectives in different ways. A given programme may end up situated more towards the information or towards the skills end of the spectrum. The overarching design criterion is to build each training to respond to a specific risk profile, which in turn is determined by the sector, professional, and country context. Other considerations to be taken into account include the sociocultural profile and career stage of the training participants (e.g., entry level vs. senior management). Training implementation parameters are important as well, though often overlooked (e.g., mandatory vs. voluntary; stand-alone vs. integrated into other training). The relation between training formats and effectiveness is likely to be influenced by all these factors and provides fertile ground for future research and experimentation.1646603b477c

Measuring impact is less difficult than often assumed

Indicators for measuring the impact and effectiveness of training vary widely according to the mix of training objectives and the measurement ambition.7e16c0841f3a At the most basic level, indicators aim to measure the participants’ satisfaction with and perception of the training (e.g., through feedback surveys) and test their knowledge of specific compliance or corruption issues. For trainings that emphasise a broader understanding of integrity risks and how to evaluate these risks in specific situations, assessment frameworks can employ tailored scenarios or standardised psychometric tests for gauging shifts in attitudes and competencies. Questions about peer behaviour and external data on helpline use or complaint procedures are applied to arrive at a more granular picture of organisational culture and outcomes.d115e92c1522 More recent additions to the assessment toolbox include the use of experimental cheating games,fe235e42a8c7 textual analysis of organisational files and outputs,d5973a86da46 and the use of so called ‘mystery shoppers’30fb13ba5a35 who probe the conduct of participants by approaching them as clients in practical service situations.

Figure 2 summarises and clusters the main approaches to measurement.8d4658bbf589 Some of these approaches map directly onto and probe individual learning objectives regarding knowledge, awareness, and skills. Others assess action intentions as well as actual changes in individual and collective behaviour.

Adapting training measurement to meet specific purposes

Source: Dieter Zinnbauer

Popular at the basic end of the spectrum are one-time surveys among a specific target audience (such as employees,58bdd32425e3 police officers,eb483687bed8 or business students3527e92a9d74), paired with correlation analysis to determine whether exposure to ethics training is associated with specific attitudes or competencies related to integrity or corruption.2083905b87c9 For example, a 2015 study found a significant correlation between individuals’ perceptions of their integrity training as effective and their ability to solve ethical dilemmas.b2db45148ab6

Another cluster of research efforts probe for a causal link, seeking to determine whether and to what extent training contributes directly to observed changes. For this purpose they deploy baseline and endline surveys to measure knowledge and attitudes before and after the training(s). For example, a 2014 study found improvements in ethical intent for employees who received ethics training.97f7227d1eda More rigorous approaches use a control group: that is, they compare changes in knowledge and attitudes before and after training with potential changes over the same period in a comparable group of people that did not receive the training.ffd8eed313d3 Using such an approach, a 2017 study found that training helped to significantly improve anti-corruption knowledge among bank employees.1324e7a98682 More recent studies have begun to test different training formats simultaneously by providing different types of training to different groups and comparing the results.137aae2bfbe9

Some of the most recent and most ambitious studies triangulate results and seek to capture broader outputs by combining treatment group assessments with survey data from external stakeholders7b8ab4be3f01 and administrative data sources.a3dab6b102e8 Such approaches have, for example, helped to determine that training positively affected actual police conduct in India2a96a7ddaee0 and Mexico.958d5b3ce4ce Annex 1 summarises the major studies that have taken place during the past ten years alongside the methods used and results obtained.

Against this backdrop, the persistent concern that training effectiveness cannot be satisfactorily measured appears exaggerated. Innovative data sources are being tapped, assessment strategies are growing in sophistication, and a vast array of customised and standardised indicators and tests is available. Given sufficient planning and funding, an integrated assessment strategy can be developed for specific trainings and built into them from the design phase onwards. In this way it should be possible to arrive at confident claims about the effects or non-effects of training on important antecedents for and dimensions of integrity and anti-corruption.

It should be noted that none of the empirical studies covered by this literature review set out to test a causal link between training and direct, population-level corruption outcomes, such as bribery rates. Such a question is perhaps too linear and general to be operationalised in a rigorous way. While the existing body of research has largely sidestepped the complex task of establishing a direct causal relationship between training interventions and curbing corruption, there is a nuanced landscape where training does exhibit demonstrable impact.

Evidence of effectiveness is positive but contingent

There is a solid empirical evidence base, far beyond the proof of concept, to show that training can make an effective contribution to improving integrity outcomes. However, there are also a significant number of (mainly older) studies with mixed or inconclusive results. This suggests that the effectiveness of integrity training is far from guaranteed and indeed is highly contingent, meaning that it is influenced by a variety of training features and contextual factors.

It remains unclear which types of pedagogies, modules, and other specific features of training design work better than others and under what circumstances. Given the enormous breadth and diversity of training parameters, objectives, and measurement approaches, paired with the scant description of the actual training interventions in many studies, it is all but impossible to arrive at high-confidence empirical insights about optimum training design features. Nevertheless, the current literature points to several features that may make success more likely. They include:

  • A balanced mix of training components focussed on values, knowledge, and skills.0eea23cd1ed0 An approach that seeks to improve ethical competence without enhancing knowledge of support resources or practical resistance skills is likely to be insufficient. So too is focussing on practical scenarios without anchoring them in reflection on higher-level values and purpose.
  • Careful attention to group composition and group dynamics before and during the workshop. Training work teams together may reinforce existing hierarchies, prevent candid conversations, and breed cynicism.1d096565811b Letting peer groups role-play scenarios without proper guidance may augment peer pressure and corruption stereotypes, rather than deconstruct them. Special attention must be paid to the gender-related composition of training groups, which is discussed in a later section.
  • Learner-driven, interactive formats guided by clear objectives. The need to introduce participants to vast amounts of new knowledge and information makes it tempting to rely on a one-directional lecture format. This can put participants into passive absorption mode. More conducive to active learning are interactive and small-group exercises and case work that give participants sufficient leeway to steer conversations to topics they find relevant. Such free-flowing approaches, however, must be accompanied by a clear explanation of objectives and a guided reflection on outcomes. It should be kept in mind that some participants appreciate a more structured learning approach.

There is no golden rule as to the optimum length and format of training. Some studies suggest that training interventions lasting two days may produce good results, while others find evidence for a shorter duration or spread-out training at intervals.15c175163d19 Few studies directly compare different treatment lengths or consider longer-term training schedules with refreshment cycles. This suggests that training length may be determined more often by budgets and participants’ availability than by the merits of a particular time frame. Similarly, the evidence does not suggest a clear advantage for either on-site or online format.171ccd685c04 Online-only pedagogies may find it difficult to achieve maximum results, but when budgets are tight, hybrid formats that mix online and in-person learning might yield better value than one-off in-person sessions.7fcccf745f08

Is training overall getting better over time? Some meta-reviews suggest so,67298490f8ca but they need to be taken with grain of salt. The number of empirical studies is very small relative to the vast universe and diversity of training interventions, and study choices are shaped by research trends and publication imperatives. The body of empirical evidence is therefore unlikely to give a full picture on whether integrity training on average has become better over time.

Pathways to training innovation

Inspired by insights from outside the integrity training field and drawing on the author’s experience in designing innovative training approaches, this section suggests some approaches that could be tried out or explored further to augment the impact of integrity training.

Integrity trainings are often construed as stand-alone interventions, but they can also serve as launchpads for lasting peer support networks aimed at bolstering integrity. Convening practitioners who confront shared challenges and giving them opportunities to interact, compare experiences, and work in teams can forge supportive relationships and solidarity. Continuing opportunities to learn from and support one another may even have more long-run impact than shifts in individual attitudes and knowledge. In Zambia, for example, an exploratory training of urban planners explicitly focussed on building lasting relations and support networks among participants. This was done through simple means such as adding several social events to the training curriculum, creating a WhatsApp group for attendees, and following up frequently after the training.f2842bbf36dc

New approaches based on socialisation are also emerging within development aid agencies.ed6d0aa74940 For instance, at Sida, risk managers and corruption experts are the ones training their colleagues: ‘Our risk management and corruption experts take the lead to organise training sessions and discuss different aspects of risks related to corruption, for example sharing their experience on how whistle-blowing cases are managed.’95861d1f7a16 Similarly, at the French Development Agency, a learning practice is based on peer-to-peer dialogue with other development agencies and among employees, who share their expertise and past experiences in dealing with particular risk situations: ‘During training, a lot of space is dedicated to exchange with colleagues. At the compliance department, we are not specialists in all sectors and geographic areas, so it is very useful to have experts explaining specific risk situations.’285fe505753b

The advantages of such longer-term approaches are evident. Follow-up meetings, for example, have long been thought to be useful for locking in short-term training impacts,f10a92c80d6c refreshing knowledge,a22ed24fc05e and reinforcing commitment.c1c4f3c0b243 A large empirical literature on how to facilitate social connection in trainings and workshops, including in online environments, stands ready to support a stronger emphasis on these aspects.3fac241db927

It is also important to build, in parallel to the training events, a broad support infrastructure of live expert advice and updated accessible resources that will be available not just during the training but also after it. Such complementary efforts can help give participants a confident sense of being part of an empowered peer network with ready, on-demand access to expert support. This is preferable to a training experience that sends participants back to face their daily challenges without any continuing support. As there has been little effort to devise such support networks and integrate them into other programming elements, this strategy offers much room for experimentation.

One approach is to make expertise and resources that participants are introduced to during the trainings easily and continuously accessible after the sessions. Training sponsors could consider retaining expert practitioners part-time for on-demand ‘office hours’ or establish a helpdesk that former participants can consult. Alumni networks can be strengthened post-training by organising follow-up reunions and related policy events. And participants that turn out to be particularly highly motivated could be offered train-the-trainer sessions and resources so they can convene mini-trainings for peers at their workplace. Figure 3 summarises related options.

Strategies for nurturing post-training engagement

On demand expert support

  • Direct line to retained experts
  • Access to helpdesk / expert network
  • Clinic / office hours

Peer network cultivation

  • Online group network
  • Re-union events, perhaps linked to policy event
  • Periodic alumni outreach / news sharing

Follow-on offerings

  • Training-the-trainer for selected participants
  • Local snap trainings

Source: Dieter Zinnbauer

Such experimentation with fresh ideas for refining training formats and expanding training functions also opens new opportunities for a broader research and learning agenda in this area. For example, research studies would benefit from being more explicit about which training modules and pedagogies are part of the intervention. They could vary some of these elements systematically and thus test different training designs simultaneously, and in comparison to a control group that does not receive any training, in order to compare different levels of effectiveness for different types of training. Likewise, the range of potential impacts to be assessed could be expanded in future research projects in order to capture some of the longer-term networking, community-building, and potential spillover impacts. This in turn would require working with longer-term research horizons and also expanding the group of stakeholders to be included when tracing impact.

Addressing the gender blind spot

One clear insight from the literature review is that gender is not sufficiently considered in either training intervention or assessment design. Only a few of the empirical studies of integrity training impact between 2013 and 2023 even offer a basic breakdown of their training participants by gender. Two studies almost approach gender parity in training/survey participation,f9178375f58c but none of the others seem to have made an effort to assemble a representative sample. Some include gender as control variable without engaging in a more in-depth discussion of this dimension. Not a single recent study focuses specifically on a gender-sensitive or gender-specific integrity training and its assessments.6f89eb6b8882 This is surprising and unfortunate for several reasons.

First, there is a clear gender dimension in exposure to corruption and integrity risks. Women and girls are often among the most vulnerable when it comes to corruption, facing a number of gender-specific risks such as sextortion. As a result, they may perceive corruption challenges differently from men.57a8ef242ed7 This alone suggests that gender-specific training should be a priority and requires careful development.

Second, there is some provisional evidence that strategic attention to gender can help to improve integrity in organisations and communities, at least temporarily. This may include strategies related to female leadership, female representatives in government,4b27d1394945 and the share of male and female staffd5437794459b or voters1417c88dc3fa in specific teams or communities. It may also include gender-sensitive ways to work with norms and attitudinal change.ad74ad5fb8fb

Finally, there is a considerable trove of practical experience with training women on life skills and livelihoods,b6a898f68c99 negotiation capabilities,d99728380368 economic and social empowerment,afb3aa060e9f and de-escalation and personal safety.763549cedcc7 These experiences provide an important source for cross-disciplinary learning and innovation in integrity training. As many corruption situations involve highly asymmetric positions of power and elements of extortion,4226014a9a75 a broader focus on life skills and self-empowerment, with elements learned or borrowed from training in those fields, can help to improve integrity training for both women and men.544de06092db

Addressing the gender blind spot in integrity training and in the empirical studies that examine its impact is both timely and essential. A continuing lack of attention to this issue means that training will be poorly suited to and leave behind one of the main beneficiary groups, reaffirming gender inequalities along the way. It also means that many opportunities for maximising training effectiveness, strategic impact on integrity systems, and inspirational learning will remain untapped.

Looking ahead

This U4 Brief has examined the evidence and learning landscape for integrity training and has identified several insights of value to practitioners and researchers in this area. Empirical research over the past ten years has firmly established the proof of concept, finding that integrity training can indeed make an impactful contribution to tackling corruption and improving governance. The next step is to take a closer look at which types of training work better in which contexts. A growing and increasingly sophisticated toolbox for measuring different types of impacts is available to support these efforts, and some of the most recent studies have begun to more fully utilise these methods and step up the generation of actionable evidence.

Further experimentation with novel training designs and more learning from experiences with training in other areas offers a high potential for innovation and future refinements. Similarly, a stronger focus on how integrity training can be complemented by continuous support and capacity-enhancing efforts opens up another fruitful avenue for improvements. Last but not least, realising the full potential of integrity training urgently requires paying more attention to gender. Recognising how integrity risks vary by gender and how strategies of resistance and training formats can be made more gender-sensitive can go a long way towards making integrity training work effectively for anti-corruption and good governance.


Annex 1: Empirical studies of integrity training (2013–2023)
Annex 2: Selected meta-studies and overviews (2013–2023)

Download a file with the two annexes here.

  1. European Commission 2023.
  2. Chen 2022.
  3. Mulhearn et al. 2017.
  4. Kreismann and Talaulicar 2021.
  5. Warren, Gaspar, and Laufer 2014.
  6. Maesschalck and Schrijver 2015; Hauser 2020.
  7. Watts et al. 2017. Perhaps most prominently, the theory of planned behaviour developed by Ajzen (1991), and its variants, anchors a number of studies.
  8. Steele et al. 2016.
  9. It is beyond the scope of this Brief to discuss the mediating influence of all these parameters, but the gender dimension is explored in more detail in a later section.
  10. Canales, Magaña, and Santini 2020.
  11. Azulai et al. 2020.
  12. Harris et al. 2022; Meyer-Sahling, Schuster, and Mikkelsen 2022.
  13. Navin et al. 2020.
  14. Kirkpatrick and Kirkpatrick (2006) provide a general typology of training assessments that is often referenced in this context.
  15. For overviews of other assessment frameworks for integrity training, see, for example, Kreismann and Talaulicar (2021) and Hong et al. (2021).
  16. Ruiz et al. 2015.
  17. Park and Blenkinsopp 2013; Ruiz et al. 2015; Kaptein 2015; Crim 2018; Hauser 2019; Kancharla and Dadhich 2020; Navin et al. 2020; Knight 2023.
  18. Parks-Leduc, Mulligan, and Rutherford 2021.
  19. Banerjee et al. 2021; Harris et al. 2022; Meyer-Sahling, Schuster, and Mikkelsen 2022.
  20. Kancharla and Dadhich 2020.
  21. Pallai and Gregor 2016; Christensen, Cote, and Latham 2018; Cote and Latham 2019; Azulai et al. 2020; Singer and Diab 2021.
  22. Baxter, Holderness, and Wood 2017.
  23. Van Montfort, Beck, and Twijnstra 2013; Skogan, Van Craen, and Hennessy 2015; Baxter, Holderness, and Wood 2017; Van Droogenbroeck et al. 2019; Canales, Magaña, and Santini 2020.
  24. Warren, Gaspar, and Laufer 2014.
  25. Canales, Magaña, and Santini 2020.
  26. Banerjee et al. 2021.
  27. Azulai et al. 2020.
  28. Canales, Magaña, and Santini 2020; Banerjee et al. 2021.
  29. Kreismann and Talaulicar 2021.
  30. Maesschalck 2019; Azulai et al. 2020.
  31. Hess 2021.
  32. Medeiros et al. 2017.
  33. Antes et al. 2009; Van Montfort, Beck, and Twijnstra 2013; Pallai and Gregor 2016; Watts et al. 2017.
  34. Kreismann and Talaulicar 2021.
  35. Siame, Nkula-Wenz, and Zinnbauer 2020; Alexander et al. 2022.
  36. Emilie Loiseau, anti-corruption officer and legal advisor for Agence française de développement, interview on 7 September 2021.
  37. Lena Fassali, programme manager, Swedish International Development Agency, interview on 10 August 2021.
  38. Nicaise 2022.
  39. Chopik and Oh 2022.
  40. Van Montfort, Beck, and Twijnstra 2013.
  41. Gino et al. 2011.
  42. Weber 2007.
  43. Two older studies seek to assess gender-specific impacts and arrive at opposite conclusions, one finding a positive effect only on male participants, and the other only on female participants. See Maesschalck and Schrijver (2015) for a discussion of these studies.
  44. Warren, Gaspar, and Laufer 2014; Pallai and Gregor 2016.
  45. Boehm and Sierra 2015; Stensöta, Wängnerud, and Svensson 2015; Breen et al. 2017; Bauhr and Charron 2020; Bicker Caarten, van Heugten, and Merkle 2022.
  46. Christensen, Cote, and Latham 2018.
  47. Alexander, Bågenholm, and Charron 2020.
  48. Bauhr and Charron 2021.
  49. Iyer et al. 2012.
  50. For more on possible pathways for innovation in integrity training, see Zinnbauer 2022.
  51. Rose-Ackerman 2010.
  52. Hollander 2018.
  53. Bandiera et al. 2020.
  54. Ashraf et al. 2020; Recalde and Vesterlund 2020.
  55. Malhotra and Elnakib 2021.