# REG EDI Strategy

Warning: This is an internal document of REG, please do not share outside the Turing.

## Intro

The [Turing EDI page](https://www.turing.ac.uk/about-us/equality-diversity-and-inclusion) currently outlines some of the ongoing work towards EDI initiatives at the Turing. The aim of this document is to embed these wider learnings concerning EDI into our work in REG. EDI needs to be at the center of all the work we do.

The REG EDI Strategy and process comprises of three main sections:

1) Outline of our REG specific EDI motivations and approach
2) REG team objectives for the upcoming year
3) A characterisation of the 5 key areas of work within REG, how we can embed EDI within them and specific objectives for each area.

EDI stands for Equality, Diversity and Inclusion. There are different ways to understand these terms, our understanding is as follows:

Equality is about ensuring that every individual has an equal opportunity to make the most of their lives and talents. It is about creating a fair society where everyone can participate and achieve their potential.

Inclusion refers to an individual's experience within the workplace and wider society, and the extent to which they feel valued and appreciated. It is about allowing individuals to authentically express themselves.

Diversity is recognising and valuing differences. It is about creating a culture that recognises and embraces the differences between individuals as a productive force.

# Motivations and approach

## Motivations

There are two primary motivations to care about EDI within REG.

Our first motivation is a moral one. We believe it is important to value and treat everyone with kindness and respect. A person should not be discriminated on the basis of some characteristic that they cannot control. We believe in equality of opportunity and that this requires a sensitivity to individuals' unequal backgrounds imposed by structural inequalities in society. Everyone has their own lived experiences which should be valued and appreciated.

There are many types of power inequality and oppression in the world of data science and broader society. Algorithmic models are at risk of bias by training on datasets that perpetuate existing inequalities and thus [reflect human prejudices](https://www.princeton.edu/news/2017/04/18/biased-bots-artificial-intelligence-systems-echo-human-prejudices). As a group and as individuals, we want to be able to recognise and challenge these inequalities.

Our second motivation is a business one. In the last few years Mckinsey has published [a variety of papers](https://www.mckinsey.com/featured-insights/diversity-and-inclusion/diversity-wins-how-inclusion-matters) promoting the business case for diversity. Organisations in the top quartile in terms of diverse leadership on average outperform those in the bottom quartile by 36%.

We believe that a diverse and inclusive organisation is a successful one. As a research organisation we thrive on the sharing of various ideas and effective collaboration with others. The most complete knowledge comes from synthesising multiple perspectives. It is crucial that we develop a fair and inclusive organisational culture where individuals feel enabled.

These structural inequalities can undermine true equality of opportunity. We want the best people in REG and it is a significant problem if we lose talent based on factors such as discrimination.

## Approach

EDI should not be something *additional* to REG. It should be something that lies at the heart of REG and influences everything that we do; we want to *mainstream* EDI.

“Mainstreaming sets out to affect both cultural and systemic change in an organization such that an equality perspective is integrated into its day-to-day working across all functions” (UK Equality Challenge Unit)

As part of our EDI approach we consider a range of areas, including but not limited to:

- Gender
- Ethnicity / Race
- Marital status
- Pregnancy, maternity and paternity
- Socio-economic
- Mental health
- Sexual orientation
- Religion
- Disability
- Neurodiversity
- Age
- Diversity of backgrounds (both educational and geographical)
- Role seniority and permanence

These are also in line with the **Equality Act 2010**. We appreciate that these areas do not stand alone and that it is important to consider the intersectional nature of systemic discrimination. All our work as part of the EDI service area should be targetted at at least one of the above areas, and often multiple (if not all) of them.

### EDI Working Groups

Many of these areas are also directly linked to different EDI working groups within the Turing. As part of our approach, **we also commit to connecting with and supporting each of the Turing's EDI working groups**:

**EDI advisory group**

**Attracting Diversity, Developing Talent and Public Engagement / Chair: Amy Gallimore**

**Gender and LGBTQ+ equality / Chair: Catrin Evans**

**Health and wellbeing / Chair: Cybele Fernandes**

**Race and Social Economic Equality / Chair: Anneca York**

# REG Team Objectives

Our approach this year (September 2020 - September 2021) is about enabling us to make larger impacts in the future. We will put in some initial work into governance, monitoring, strategy, documentation, relationship building and this will all pay dividends in the future. We currently lack much of the necessary infrastructure to understand what areas are most significant. It is difficult for example, to set recruitment targets without knowing our current position.

As mentioned previously, we believe that EDI should be mainstreamed. It is something that should be embedded within our team and is something that everyone should be working towards together. As a result, we believe that there should be REG Team objectives that are everyone's collective responsibility. REG leadership (and the EDI service area) should be directly responsible (via personal objectives) for them on a higher level, whilst individual team members should be contributing as they can.

1. **Set up frameworks to monitor diversity and compare our current practices against benchmarks.**
- Identify comparable organisations and relevant recruitment pipelines for REG. Gather data on their diversity and use to inform benchmarks for our own recruitment process. _Report by February._
- Collaborate with HR to develop a structure for regular diversity reports for each recruitment round. _Infrastructure by Dec; retrospective report by Feb; rolling reports by April_.
- Annual statement on current status of gender and ethnicity pay variation within REG. _Proposal by Dec; report by April._
- Add guidance to the diversity statement section of the REG project proposal document to highlight what REG considers to be important contributions to project diversity. Implement a process to monitor the content of the statements and their impact on the outcome of the proposal. _In place by Feb._
- Set up regular EDI projects report, capturing projects with EDI focus in the project board and diversity metrics for all projects. What EDI focus is and what metrics we use will be defined. _Discuss definition by November; report by January._
2. **Integrate EDI into REG service areas and internal development.**
- Identify opportunities for collaboration between the EDI service area and other REG service areas to jointly implement at least four of the targets from the rest of this set of this document. _One collaboration + completed task per quarter._
- Hold at least one EDI training event for REG. _Planned by Dec; event by April._
- Produce EDI Guidance
3. **Expand our outreach and recruitment schemes with an EDI focus.**
- Build a connection with at least one charity or community group to lead to an event or short-term placement targeted towards traditionally under-represented groups in tech. _Proposal(s) by October; event(s) by April._
- Strengthen REG connection with the wider Turing EDI work, including attendance to all EDI working groups.
- Formalise the recruitment process for REG placements to allow for EDI monitoring. _Process in place by next recruitment round._

# Specific approach to different types of work

We want EDI to be a part of everything we do, however we also recognise that different types of work will require a different, although linked approach. We want our objectives to be a specific and targetted to the different type of work.

We consider four main types of work: **Outreach, Projects, Internal Career Development** and **Recruitment**.

For each type of work there will be a set of objectives for the 2020-2021 year.

EDI is a collective responsibility, and will require the support of everyone within REG to effectively achieve. In particular, REG leadership should each have an EDI objective based on one or more of the objectives we described above.

## Outreach

As mentioned previously, we believe that diversity brings a range of benefits to an organisation and should be celebrated. In light of this, the REG outreach efforts should not only focus on the usual goals (such as science communication, recruiting or promoting RSE careers), but should also focus on increasing the visibility of under-represented groups and reaching out to wider and more diverse audiences.

### Actions

- **Develop a REG outreach agenda that focuses on EDI principles**:
- Working with recruitment to do EDI focused outreach of new roles/placement scheme (attending networking events, universities, etc).
- Extend our outreach activities to include school(s) in the Camden area. Try to reach out to pupils from under-represented groups for a short placement.
- Seek collaborations with charities and organisations that work with disadvantaged people (e.g. Red Cross).
- Attend events to promote the RSE movement, and REG in particular, to under-represented groups in tech (or for an international audience that has not been exposed to the RSE movement).
- Provide TOIL for events occuring after work hours.
- **Gather data on the on diversity represented in the different REG outreach efforts**.
- Use this data to inform future content for the existing projects (E.g The Turing podcast, Twitter take over series, REG spotlight series, Turing blog, individuals giving outreach talks).
- **Provide EDI related documentation for people doing or organising outreach events**.
- List of resources for running inclusive and accesible events/presentations/other outreach activities.
- A checklist for individuals doing or organising outreach.
- **Share best practices for organising and running inclusive events** Members of REG are often involved in organising and hosting events of small/medium size at the Turing, e.g. workshops associated with a project or training events.
- Develop a guide for organising inclusive events (see [#469](https://github.com/alan-turing-institute/Hut23/issues/469)).
- Contribute the non-Turing-specific material to The Turing Way's [_Guide for Communication_](https://the-turing-way.netlify.app/communication/communication.html) book.
- **Support the Turing-wide EDI working groups** The four working groups regularly run events around their own particular theme, and rely on their members to support those events.
- _Attracting Diversity, Developing Talent and Public Engagement_: Louise and Anna will support National Inclusion Week events in September.

## Projects
The projects we choose to do as a team and the way we work on them should be aligned with our EDI commitments and motivation. We want to do more work that has diversity embedded in it, focuses on addressing structural inequalities through data science and cultivates collaborations with organisations and communities that will lead to positive change.

### Actions
- **_Decide EDI criteria for prioritising projects and capture relevant data about the state of the REG projects portfolio_**: We need to define what types of projects we want to see more, e.g. projects that address inequalities in DS/society, projects with more diverse collaborators, etc and collect data.
- Organise team meetups to discuss criteria for EDI-focused projects.
- Regularly measure and report percentage of EDI-focused projects in main board and 22 days board, set up diversity metrics for all project teams.
- **_Set up targets for EDI project percentage and diversity metrics_**:
- Decide what percentage of our projects should have an EDI focus and what diversity targets we should set for collaborations (if any). We could set up an objective like "At the end of the year, REG's project board has a larger (or X) percentage of projects focusing on ...".
- Quantify the risks and costs arising from a possible shift (e.g. some other projects would not be selected).
- **_Increase REG's involvement in EDI-focused projects and connections to academics and organisations that focus on EDI_**:
- Modify our project selection criteria by extending the definition of project diversity beyond academic disciplines/sectors and introducing a way to reward projects that address issues of inequality/bias/fairness (either in the data world or in society) or involve collaborations with minoritised groups.
- Investigate whether we can institute a "discounted REG rate" for EDI-focused collaborations with organisations that cannot afford the full rate and/or a pro-bono scheme. The cost recovery model should be examined in this case.
- **_Set up EDI guidelines for any project work we undertake_**.
- Set up a checklist for EDI awareness and good practice - focusing on REG project scoping and project work. This could cover guidelines to avoid common forms of algorithmic/data bias, build accessible software, prevent misuse of our work, make sure projects do not impact communities without their consent. Could be a by-product of EDI training/reading groups.

## Internal Career Development

Actions in this area are specifically motivated by our overall goal of promoting equality in the way people enter and progress in the team. We want to guarantee that REG is a welcoming and inclusive environment that equally supports all team members (regardless of role seniority / permanance) through their career, effectively addressing any form of disparity in treatment.

### Actions

- **Assessment of the Current Paygap Situation.** *(In conjunction with REG Leadership and HR).* It is important to have a clear overview of potential initial and long-term disparities. We aim to achieve this by:
- Have an internal analysis (by Leadership and HR) of potential gender/ethnic/etc. paygaps in REG, both when people were initially hired (including the negotiation phase) and in the current stage for same role+experience+performance.
- Start with a yearly anonymised statement to the group.

- **An EDI shaped Competency Framework.** *(In conjunction with the Competency Framework group).* It is essential that we define a clear set of milestones for career development that would lead to raise in salary and seniority. Our role in the collaboration aims at guaranteeing that:
- We allow people to achieve such milestones during their everyday job, for instance by giving opportunity of mentorship, leading collaborations or attracting funds.
- We offer support and training opportunities for specific skills needed (e.g., training courses for mentorship or project management)
- The framework recognises that people have different collaboration and leadership styles and that competencies can be demonstrated in many different ways.

- **EDI and Promotions** *(In conjunction with REG Leadership and HR).* In case there are limitations in the overall Turing budget for fully implementing the competency framework strategy, it is important that:
- The REG leadership would liaise with HR to avoid choices that would further increase (potential) disparities across the group.
- In case resources are available for future hirings (e.g., hiring a senior position), we always first check whether current members could move up in their career trajectory.

### Internal Data Science EDI training

#### Objectives
- **_Organise a Data Science EDI training event or reading group_**: This would focus on topics such as data ethics, algorithmic bias, addressing privilege and bias (theirs or others') when involved in data science projects.
- Set up the reading group or course. This could be a collaboration between EDI and Training service areas or a team of volunteers and could also involve external experts running sessions. The material in Data Feminism is a useful first source to base this on.
- Make reading the relevant sections in the Turing Way (or some other source of information about bias/inequality in data science) part of the starters checklist.
- Seek collaboaritons with TPS and Turing Way on the above topics.

## Recruitment

As referenced in the Motivations section, organisations are able to flourish in rapidly changing business and economic environments when they have a diverse workforce with a wide variety of experience, skills, perspectives and ideas. We want to ensure that our recruitment process is inclusive and fair for all groups of people.

### Actions

- **_Define a clear set of milestones_** We want to define a set of milestones for recruitment. For example, this could include setting specific goals regarding the diversity of applicants we hope to attract. In order to set such goals, we need to identify baselines that will evaluate our current position on recruitment with respect to EDI (attraction, shortlisting, interview and offer) and enable tracking of any progress we make. We will also need a way of ensuring we make informed decisions when considering recruitment goals we wish to set for the demographics of applicants relating to EDI-protected characteristics.

- In order to identify baselines, where possible and with consent, we should collate data from recruitment calls regarding diversity of applicants, as well as diversity-related demographics of REG. We cannot neccessarily assume such demographics and these baselines are important if we consider setting EDI-related recruitment goals. Ideally, we want a pipeline whereby such data is easy to periodically obtain for the future.
- We may consider aiming to attract _X_\% applicants with a certain protected characteristic - but how do we decide the value for _X_? We need to identify and analyse relevant external data for this task. We might also want to solicit support from other REG members for how to come up with these numbers.
- We should also seek discussion/review of these with HR and leadership when setting recruitment milestones, as we may be unable to independently set/make these decisions in some cases.

- **_Training for staff involved in recruitment \& selection processes and decisions_** There should be EDI and unconscious bias training for all staff involved in recruitment decisions. For example, staff would review this material before short-listing or interviewing candidates to promote awareness of the impact of unconscious bias during interviews. This EDI training could be a nice accompaniment to some general interview training for staff too. A well-run interview is likely to bring out the best in all candidates, especially helpful in cases where a candidate lacks confidence in an interview setting and offering training for staff might also increase the pool of people willing to interview.
- This would require developing a relevant guide/training material.
- We might also want to discuss the possibility of appointing someone (or a _Chair_) that will challenge the thinking of staff throughout shortlisting/recruitment decision making processes.

- **_Ensure our job opportunities reach a diverse audience_** We'd like to minimise the possibility of over-relying on standard mainstream places to post our job advertisements to and make sure we advertise jobs through avenues that might enable greater outreach into diverse groups.
- Understand where we have previously and currently been advertising available posts.
- Review Amy Gallimore's list of contacts she sent us for suggestions and contribute other relevant options we find.

- **_Offer support to potential candidates prior to applications_** It would be nice to encourage pre-application discussions with potential candidates to provide them with light mentoring/tips for the application process. This may help candidates have a better awareness of what will be expected.
- One option is to emphasise in the job advertisement that we are happy for (and encourage) candidates to contact _X_ before applying to find out more about the position, interview process, and answer any concerns or questions they may have.