EQUITY IN EDTECH BY DESIGN

This report aims to provide guidance for improving equitable EdTech design, policy and practice. We identified relevant academic literature and captured best practices in identifying equitable EdTech features, as well as biassed design and organisational practices in EdTech. Our approach draws from existing literature indicating that accepted standards and indicators have generally proven to positively influence developer and consumer awareness, as well as policy-makers’ decision-making.

To operationalise equity for this report -by 'equity' we mean addressing the question of WHO benefits, individually and collectively, from the EdTech solution, particularly considering underprivileged and underrepresented groups and diverse constituent perspectives.While this definition is necessarily simplified, it allows for the incorporation of multiple aspects of equity based on a literature review of equity conceptualisation and its understanding within the context of education and EdTech specifically.As such, in our definition, an EdTech solution is deemed equitable through both the products' inclusive design and the provider's explicit public commitment to embed equity considerations into its operations within its particular context.Aligned to UNESCO's three pillars as mentioned above, an EdTech solution is equitable through its content, capacity, support for students and their teachers in developing their digital competencies, and connectivity, while taking into account individual contexts of use of each particular solution.
Our aim with this report is to explore the concept of equity in relation to EdTech.
To that end, we first outline the current understanding of the concept in education more broadly from a global perspective (Section 2), followed by a literature review to identify key research themes in relation to equity in the EdTech field (Section 3).Section 4 gives an overview of current EdTech industry related equity certification, benchmarks, and indicators.Section 5 then provides a consolidated benchmark that draws on the findings from the previous sections.
This report seeks to support the EdTech community to align effectively with the goals through accepted indicators for monitoring and evaluation (M&E).We aim to consolidate the equity M&E frameworks into shared benchmarks to establish not only shared definitions and understanding of equity in EdTech but also implementable ways of identifying and understanding their presence using a specific set of indicators.Our objective is to facilitate the adoption of a shared language among multiple stakeholders in the EdTech field and to support effective alignment of EdTech equity indicators with the broader UN Sustainable Development Goals (SDGs; https://sdgs.un.org/goals).Several UN SDGs feed specifically into the questions of equity: in particular, SDG 5 (gender equality), SDG 10 (reduced inequalities) and SDG 16 (peace, justice, and strong institutions).The SDG 4 (quality education) also emphasises the significance of inclusive education, focusing on the inclusion of children with diverse needs, addressing the question of equity in education most directly.Carrington, 2022).However, within the field of education, equity considerations extend beyond the legal framework.In very simple terms, the idea of equity "relates to issues of 'who gets what, when and how'" (Rizvi & Lingard, 2010, p. 76).

EQUITY IN GLOBAL EDUCATION
There has been a considerable ideological and political discursive struggle over the nuances of the meaning of 'equity' and the key associated societal values (Rizvi & Lingard, 2010).This discursive struggle is closely aligned with the broader discourse on educational justice, far too often narrowly conceptualised as access to institutions without sufficiently considering the "dynamics of educational experiences and their social and economic outcomes, as well as the historical conditions that produce inequalities", as "formal access to schooling does not always translate into effective equity outcomes" (Rizvi & Lingard, 2010, p. 141).
In the globalised field of education, there are three influential and vocal discourse participants with somewhat different ideological stances: UNESCO, OECD, and the World Bank (Rizvi & Lingard, 2010).Equity concerns are an integral part of their discourse, that is how they conceptualise and apply 'equity' within their operations.In order to better understand it, we have compiled a small corpus of texts (ca.200,000 words) associated with these institutions, including, e.g., GEM 2023 report and PISA results reports (for full list of included texts, see Appendix A), based on which we conducted a brief analysis of the use of the word 'equity' in these texts, i.e. its discursive construction.
To this end, we used a linguistic concept of 'collocation', i.e., we looked at words that co-occur frequently and habitually with the word 'equity' in the texts published by these institutions.Collocations have the potential to show the meaning of words in their wider context as words do not have meanings in isolation.The meaning potential of words arises when they are used in discourse, word meanings have histories and they change, and the meanings are continuously negotiated by discourse communities.We used a corpus linguistic software (Anthony, 2023) to generate the collocations, see Figure 1.As Figure 1 shows, the central collocate, that is the word most strongly associated with 'equity' in our corpus, is the word education.We have divided the remaining collocates into several thematic groups: words that relate to 'performance measurement' like efficiency, quality, excellence, attainment; words, mostly adjectives, that characterise a quality (of whatever is measured) such as greater, improved, fundamental; action verbs describing what needs to be done such as ensuring, improve, promote; and words that refer to measurement instruments trends and dimensions.All these collocates represent performance correlated vocabulary that suggests an understanding of equity as having the potential for aligning well with the effectiveness and efficacy agenda, as explained in the introduction.
In fact, the OECD texts repeatedly highlight that 'equity' does not come at the expense of high performance, e.g.PISA Report (2015) states: "PISA … consistently finds that high performance and greater equity in education opportunities and outcomes are not mutually exclusive" (PISA 2015, p. 39).In the OECD conceptualisation of 'equity', achieving greater equity is "not only a social-justice imperative, it is also a way to use resources more efficiently, and to increase the supply of knowledge and skills that fuel economic growth and promote social cohesion" (OECD, Equity in Education, 2022, p. 4).
In the OECD texts, 'equity' is frequently linked with performance, excellence, quality, efficiency, access; in PISA reports, it is also linked with well-being and inclusion.'Equity' is monitored and improvements, changes, progress, decline, differences, gains, levels are noted.PISA reports specifically conceptualise "two dimensions of equity in education: fairness and inclusion" (PISA 2022, p. 111).
'Inclusion' is defined as "the objective of ensuring that all students, particularly those from disadvantaged backgrounds or traditionally marginalised groups, have access to high-quality education and reach a baseline level of skills" and 'fairness' refers to "the goal of removing obstacles to the full development of talent that stem from economic and social circumstances over which individual students have no control, such as unequal access to educational resources in their family and school environments" (PISA 2015, p. 202-203).
In both UNESCO and the World Bank texts, we find strong links between 'equity' and 'inclusion' and how these should be assessed.The idea of "all students" is necessarily extremely complex to operationalise.Rizvi and Lindgard (2010, p. 150)  indigenous peoples, conflict-affected).Among the thematic indicators we find, for example, focus on percentages of children in primary education whose first (or home) language is the language of school instruction (indicator 4.5.2), and allocation of resources to disadvantaged students (indicator 4.5.3).GEM report addresses, in relation to EdTech, 'equity' together with 'quality' and 'efficiency' (see also above), and asks three questions: To conclude, equity concerns remain far too often left out of impact evaluation frameworks.Well-being, inclusion, learning experience and attitudes, and qualities pertaining to equity considerations, are complex and challenging to measure.We operationalise 'equity' as encompassing practices that inclusively address who benefits, individually and collectively, from the EdTech solution.
Thus, equitable EdTech practices are those in which the necessary supports are provided so that all learners, regardless of background or ability level, have full access to a learning community's educational tools, resources, and practices (Prado & Warschauer, 2024;Schuelka & Carrington, 2022).
All learners need to benefit while specific concern for inclusive frameworks paying attention to vulnerable groups and disadvantaged students (vertical equity) is emphasised.As discussed, these groups are contextually defined and delimited and may include, but are not limited to, students with immigrant backgrounds (with specific concerns to those coming from conflict-affected areas), indigenous peoples and ethnic minorities, and students with disabilities.More generally, background factors such as socio-economic status, gender, and race are included.
Learner variability within these groups should be supported by flexible and scaffolded teaching and assessment, as well as accessible and collaborative uses of mainstream and assistive technologies (Prado & Warschauer, 2024).The three what is the context of its use, how adaptable is the solution for various contexts?
'equity' concerns practices that inclusively address who benefits, individually and collectively, from the EdTech solution.
EdTech solutions asks questions: who has access?what is the learner context?how can the learner variability be operationalised in this context?
As discussed in Section 2, 'equity' has been a persistent issue in EdTech specifically, and education more generally, as it is continuously being constructed and reconstructed (Rizvi & Lingard, 2010, p. 76).A primary issue highlighted in current research literature on EdTech is the insufficient emphasis on integrating equity considerations into EdTech solutions.Literature identifies a significant equity gap in the current education system globally with multiple studies underscoring the urgency of addressing this gap and emphasising its implications for education, society, and broader democratic values.
To begin the process of understanding what EdTech equity indicators are given consideration in the research literature, we conducted a rapid literature review.
We selected keywords 'equity' and 'EdTech' (or 'educational', 'technology', 'apps') and searched the SCOPUS database for articles published after 2020.The search was further narrowed down to the field of social sciences and to articles.This initial search yielded 252 articles, which we reviewed for relevance -focus on K12 education and equity being the core foci of the reviewed article.This yielded a set of 20 studies that were analysed in depth for further discussion.During the review, several themes emerged for which we performed additional targeted searches, these included, for example, keywords 'AI' and 'inclusive design'.
Upon reviewing recurring topics and themes across all articles resulting from the literature review, four key themes emerged from this literature review: theoretical perspectives on 'equity', biases incorporated into EdTech, inclusive and participatory design, and deficit-based approaches in Edtech.Discussions of these are followed by studies showcasing mutually reinforcing ideas and interaction.

LITERATURE REVIEW OF EQUITY INDICATORS IN EDTECH
3.1 METHODOLOGY

THEORETICAL PERSPECTIVES ON EQUITY IN EDTECH
Persistent digital divides exacerbate existing socio-economic inequalities (Macgilchrist, 2019).The digital divide extends well beyond mere access to the internet and technology; it is recognized as a fundamental right by the UN (Kormos & Wisdom, 2023).Policy makers need to consider not only the persistent wider socio-economic issues posed by digital divide inequities, but also other overlooked and under-represented groups of users, as, for example, highlighted by Bright and Calvert (2023) regarding gifted students.Additionally, the digital divide is not only confined to disparities between developed and developing countries or urban and rural areas; it may be evident even within the same district or neighbourhood (Huffman, 2018).
The authors of this report acknowledge that characterising educational technology and its implementation often stems from particular Western perspectives, which may inadvertently perpetuate educational inequities.To address this, the authors acknowledge the need to expand perspectives and intentionally repair harms that contribute to 'educational debt' (Ladson-Billings, 2006) by aiming to incorporate multiple perspectives, including non-western, to ideate what equitable implementation of educational technologies may look like.
In Rethinking the Digital Divide, Warschauer (2003)  Critical Technology theory is pivotal in equity research, offering a framework for addressing technology-related biases within the education system (Bright & Calvert, 2023).This framework aids researchers in navigating complex issues, including the equitable use of technology and its effects on underrepresented student groups.Notably, Critical Technology theory includes the consideration of groups, which may not be typically viewed as problematic or marginalised, thus highlighting the multidirectional nature of equity considerations. 13 For example, Garcia and Lee (2020) highlight the importance of an equity-centred approach to educational technology, emphasising teaching, pedagogy, and sustained relationships within classrooms.They introduce the concept of Critical Computational Literacy (CCL) as an equity-centred framework to produce technological tools that disrupt and dismantle structures upholding inequality while inventing new tools that sustain a more equitable and humanising world.
CCL integrates critical literacy and computational thinking to create technological tools for transformative social action, aligning with the call for an instructional literacy approach focused on "reading the world and reading the word" as advocated by critical literacy scholars like Freire and Macedo (1987).

BIASES IN EDTECH
Building upon insights from Critical Technology theory, to ensure fair and inclusive educational opportunities for all learners, EdTech developers need to address bias, particularly for technologies relying on AI solutions.When conceptualising 'bias', usually two types are discussed: 1) statistical bias, i.e., systematic differences between the "truths" and "facts" of the world around us and the results of an algorithmic prediction, including systematic differences between a population and the representation of the population in the sample, and 2) historical or social origins of the algorithmic bias (Kinder-Kurlanda & Fahimi, 2024).Researchers have, therefore, proposed, for example, socio-technical understandings of bias (Kinder-Kurlanda & Fahimi, 2024;Lopez, 2021;Poechhacker & Kacianka, 2021).These understandings necessarily include much wider considerations about the data we use to formulate "truths" and "facts" of the world around us -the data we collect (and how), the data we decide not to collect, and the data we do not even think to collect (D'Ignazio & Klein, 2020;Wernimont, 2018).

Historical bias
mismatches between the actual world and desired values, leading models to replicate these mismatches, for instance, student demographics as predictors of grades  Identifying and addressing these, and other, types of representation bias, is crucial for developing algorithmic systems that are fair, inclusive, effective, and sensitive to the specific contexts of diverse populations.For example, in the study by Durham (2024), bias towards multilingual students manifested through deficit perspectives regarding their linguistic abilities (see also Section 3.5).The study revealed that uncritical and indiscriminate implementation of technology may actually harm multilingual students instead of supporting their learning.This suggests a need for educators to address biases that may impact the educational experiences of their diverse students and approach technology implementation in a more contextually informed manner.For example, Nguyen and colleagues (2018) provide an example of disability status bias by highlighting in their discussion of learning analytics a potential benefit of uncovering previously undiagnosed learning disabilities; however, they emphasise, to use learning analytics effectively, there is a vital need for specialised informed consent procedures, particularly for learners with intellectual disabilities.Socio-technical understanding of bias includes both societal and statistical bias.It is imperative to address both as, individually and combined, they distort representations of individuals and groups and may thus potentially have harmful consequences.
Algorithmis systems need to strive to be fair, inclusive, effective, and sensitive to specific contexts of diverse populations.

INCLUSIVE DESIGN
Inclusive design serves as a pivotal approach to supporting equity in EdTech.As the name implies, it aims to encompass, in design solutions, all groups, ensuring equitable access and representation.Implementing inclusive design often involves leveraging learning analytics to identify and address disparities, ultimately fostering an environment where all learners can thrive.Baek and Aguilar (2023), for example, discuss learning analytics as a means to assess and improve the design of educational technologies for students with disabilities.By leveraging data-driven insights, including learning analytics for disabled students -such as multimodal learning analytics in tracking the students' experience or analytics able to capture students' emotions (see also Boulton et al., 2018) learning experiences for students with diverse disabilities could be substantially enhanced by creating more inclusive educational environments.
Multisensory approaches to design are another critical aspect of inclusive design in that they pay attention to the engagement of all senses, without privileging the so-called "higher senses" of vision and hearing but providing equal access and engagement possibilities through the "lower senses" of touch, smell, gustation (taste) and proprioception (kinesthetics and sense of being in the space).
Examples of EdTech prototypes that draw on multisensory design principles include oBooks, which are digital books available on iPad with embedded sounds, images, texts, hotspots to touch and connected smell release (when activated from within the book, see Kucirkova & Tosun, 2023) 19 Finally, the principle of multiple means of representation focuses on understanding how learners perceive and comprehend information, recognizing that individuals have diverse educational backgrounds, languages, abilities, and cultural contexts, which influence their processing of information.This principle emphasises the importance of providing various methods and formats for presenting content to accommodate these differences, ensuring that all learners have equitable access to learning materials and opportunities for comprehension and engagement (Dzaman et al., 2022).

Things to consider in inclusive design:
inclusivity of learning analytics multisensory approaches universal design for learning (UDL) principles assisstive and accessible technologies, e.g.AR Participatory design, also referred to as co-design, is another crucial approach focused on equity in EdTech design and evaluation.Participatory design prioritises the active involvement of usually unrepresented groups.By engaging varied stakeholders, such as teachers and children, participatory design not only acknowledges their perspectives but also incorporates specific strategies to ensure their effective participation and inclusion of design solutions that might have not been considered otherwise.This inclusive process not only fosters a more equitable design and evaluation of EdTech but also empowers diverse user (e.g.Tare & Gugha, 2023).This research has recently expanded to participatory AI design with children (e.g.Wang et al., 2023), with strong connections to the children's rights literature and ethical AI design in EdTech (Livingstone & Third, 2017;Hadfield-Hill & Zara, 2024).
Children's access to data (as well as Internet) was highlighted as a barrier to inclusive digital play and so was the paid nature of many digital games: Equity also extends to issues around the financial cost of digital play experiences.For a comprehensive evaluation and a holistic understanding of EdTech impact, efficacy and effectiveness studies must incorporate considerations of equity, ethics, and environment (Kucirkova, 2024).This entails conducting holistic evaluations that centre on marginalised groups, addressing the crucial question of who benefits from the technology and conducting examinations and analyses with this focus in mind.
Holistic evaluations stand in contrast to deficit approaches in EdTech (and education more broadly).Deficit approaches often underlie many design and evaluation studies, focusing on identifying and remediating problems within Asset-based frameworks, such as the Funds of Knowledge framework (Moll et al., 1992), as an alternative to deficit-based approaches, emphasise students' existing knowledge, competencies, and diverse problem-solving strategies, recognizing the importance of cultural context and resources in supporting learning, including those gained from outside of formal learning settings (see, e.g.Esteban- Guirtart & Moll, 2014;Verdin et al., 2021).

ASSETS VERSUS DEFICITS BASED APPROACHES IN EDTECH
Things to consider in participatory design: active involvement of users, including underrepresented groupd child-centred design practices ethical considerations, including advertisement placement Leverage Existing Learner Assets Using System Features and Functions.

3.
Develop New Assets Using System Features and Functions.4.
Make assets visible for learners, educators, and other stakeholders.

5.
'Detecting assets' principle can be followed, for example, through learners' selfreporting and expanding learner models, while using these in conjunction with real-time detection techniques that are adjusted so that not to be framed in a deficit style as is currently often the case.Ocumpaugh and colleagues (2023) emphasise that "pedagogies are not deficit based when they find places where students need support.They are deficit based when they only find areas where students are deficient." Principle no. 3 proposes to increase the number of pedagogical and student problem-solving approaches -for example, Hunt and colleagues (2022) suggest that while students with learning disabilities benefit from open-ended problem solving, educators tend to underestimate the extent (and type) of their prior knowledge; Gobert and colleagues (2013) have developed models that detect a student's understanding of the concept of experimentation while using a different strategy than the expected approach; Nasiar and colleagues (2023) have built models that recognise a new strategy proposed by the student; Crossley and colleagues (2014) have modelled different approaches to successful writing.
Tools that support self-regulated learning (e.g., Azevedo et al., 2022;Roll et al., 2011;Roscoe & Craig, 2022) can empower learners to develop their own assets with greater agency and independence.Principle 4 emphasises the variety of ways that students obtain knowledge, and techniques and tools to be used in supporting them.Broadening learning content to offer culturally rich experiences serves all students, including those that are typically "privileged", it offers opportunities to make new connections and learn new skills and strategies.
Principle 5 encompasses everything from including the home and cultural assets of learners to re-designing the data visualisations and dashboards used to communicate about the learner.Ocumpaugh and colleagues (2023) stress that, for example, in designing dashboards, careful attention needs to be paid to how students are labelled as the evidence shows labelling (e.g., high performing) affects students' own self-conceptions and other people's expectations of their capabilities (see also, e.g., in O' Donnell & Sireci, 2022;Walker et al., 2023;and Bertrand & Marsh, 2021).
Assets based approaches are equitable as they emphasise students' existing knowledge, competencies, and diverse problem-solving strategies, and the importance of their cultural contexts.
Assets based approaches stand in contrast to deficits based approaches that emphasise what learners lack or where they fall short of normative standards, locating issues within the student rather than considering the broader context.
In sum, while a significant portion of learning sciences literature focuses on the innovations, effectiveness, and efficiencies of educational technology, there is limited research examining its role in tackling systemic and entrenched educational inequities.Our literature review highlighted a number of issues and biases embedded in the design of EdTech and calls for more attention paid to how EdTech can contribute to promoting equity and addressing disparities in  addresses equity issues to some degree, however there is no publicly available documentation on how these topics are assessed.'The Ten Verticals' of evaluation address equity issues by considering, for example, age appropriate design, cyber security, ethics, algorithmic justice, ethics, duty of care, data responsibility.Moreover, equity is specifically mentioned, in the Pedagogy section, and provides clear values to teaching and learning, to equity, and quality education.

EDGE Standards and Certification
EDGE Standards and Certification (EDGE Foundation -https://www.edgecert.org/) is a standard for Diversity, Equity, and Inclusion, centred on workplace gender and intersectional equity.The standards offer a framework against which organisations can measure where they stand on gender and intersectional equity, as such the standard is company, rather than product, focused, for example, Abt Global (https://www.abtglobal.com/ ) has become the first U.S. government implementing partner to have been awarded the EDGE Certification, the certification process included an audit of Abt's gender equity practices, policies, employee perceptions -benchmarking them against other organisations.Communities are empowered to assert their rights and play leadership roles in decision-making processes.

EQUITY BENCHMARKS AND INDICATORS
Participatory processes are inclusive and nondiscriminatory, reflecting the diversity of the community.
Efforts are made to address power imbalances within communities, promoting equity in engagement.
Mechanisms are in place for communities to provide ongoing feedback on the quality and accessibility of services.
When developing a consolidated benchmark for equity in EdTech, we focused on the equity principle within the EdTech product, considerations of its end-users, and also the company itself.
In relation to Edtech product evaluation, key indicators for monitoring and evaluation concern the target users and context of the product use.Ideally, equitable EdTech solutions should aim to offer inclusive and accessible design as well as diverse content that is culturally localised for the contexts in which the EdTech solutions are to be used.
EdTech solutions are considered to be equitable when they can be adapted to the variable needs of diverse learners.Diverse learners include disadvantaged or marginalised students due to their socio-economic status, gender, race, disability, immigration, minority or language learner status.EdTech solutions are equitable when they build on students' diverse knowledge building pathways, and support self-efficacy and agency, rather than relying on deficits paradigms.
EdTech solutions are equitable when, through their inclusive and/or participatory design features, they address the contextually situated needs of diverse learners through flexible, accessible, and adaptable design features.
Equitable EdTech solutions are also designed with considerations taken towards addressing bias across all stages of product development (see Tables 2 and 3), but also implementation and scale across different environments.
Finally, EdTech products are equitably designed when the EdTech company offers explicit public commitment to user accountability and transparency through clear, easy to understand, consent processes.Company-level equity indicators include access, skills, use and innovation (see Table 4).

GOLD SILVER BRONZE
Equitable design comprehensively addresses indicators both at the product and organisation levels.
All features of equitable design clearly support each other and are adaptable to various learner groups and contexts of use.
All three of the following indicators are met: The product has documented inclusive and participatory design features.
The product addresses bias in its design.
The product developer publicly provides clear documentation of its practices and demonstrates transparency and accountability in addressing users' feedback.
Equitable design addresses indicators at either the product or organisation levels.
All features of equitable design clearly support each other and are adaptable to various learner groups and contexts of use.
At least two out of three of the following indicators are met: The product has documented inclusive and/or participatory design features.
The product addresses the issue of bias in its design.
The product developer is publicly committed to its users, provides clear documentation of its practices and shows demonstrates transparency and accountability in addressing users' feedback.
Equitable product design partially addresses indicators at either the product or organisation levels.
All features of equitable design clearly support each other and are adaptable to various learner groups and contexts of use.
At least one out of three of the following indicators are met: The product has documented inclusive and/or participatory design features.
The product and addresses the issue of bias in its design.
The product developer is publicly committed to its users, provides clear documentation of its practices and shows demonstrates transparency and accountability in addressing users' feedback.
Transforming the literature-based indicators into a benchmark involved categorising the key indicators into three levels in order to establish a comprehensive benchmark that can offer valuable guidance for EdTech users, providers, and evaluators alike.Certification providers can feed the consolidated benchmark into an adaptable rubric with clear parameters for technology characteristics at product level and organisation's characteristics more broadly.A rubric with clear criteria for each level can effectively guide the EdTech field toward adopting best practices.These will need to be evaluated holistically to ensure a representative score.An 'Example rubric' in Table 6 illustrates criteria focused on equitable design that can be developed into an evaluation and rating system for EdTech tools: This report aims to provide guidance for improving equitable EdTech design, policy and practice.We identified relevant academic literature and captured best practices in identifying equitable EdTech features, as well as biassed design and organisational practices in EdTech.Our approach draws from existing literature indicating that accepted standards and indicators have generally proven to positively influence developer and consumer awareness, as well as policy-makers' decision-making.
The benchmark and example rubric developed for this report are designed to be dynamic resources that should undergo continuous evaluation and modification in response to developers' and consumers' evolving needs, research, and insights from industry and practice.
In this conceptualisation, we also hope to initiate a dialogue among community constituents invested in the development and evaluation of equitable EdTech products, tools, and resources .This includes discussion at the intersection of digital equity and inclusion taking the position that "the implementation of inclusive educational practices is a matter of equity and social justice" (Prado & Warschauer, 2024, p. 17) and "the cornerstone of the modern democratic state" (Schuelka & Carrington 2022, p. 3).
Discussion of equity in EdTech also necessitates conversations about how constituents can support development of EdTech practices and tools that support students of all ability levels in achieving full access -building on the premise that inclusion of students within their learning communities supports educational equity (Prado & Warschauer, 2024).
Finally, we hope that these resources will foster the continued development of equity in Edtech knowledge-base that enhances the standards and indicators used to design EdTech that is equitable in its aims to serve a diverse range of users across multiple contexts.

Figure 1 .
Figure 1.Collocates of the word equity in our 'equity' corpus (see Appendix A for the included texts).The collocates were calculated with AntConc software: Llikelihood statistical measure was used with a window of five words to the left and right.Figure 1 includes all the collocates generated by the software except functional words in and and, and text organising words: chapter, volume, guide.
questions the GEM Report (2023) asks translate into evaluation criteria of equity concerns in EdTech solutions: First, who is targeted by the EdTech and what is the logical mechanism that the solution works for these specific groups?Second, voices to contribute meaningfully to the development of equitable educational technology solutions.Birch and Demmans Epp's study (2023)  in the context of a project involving young music students serves as an example of how participatory research and design enabled students to express their preferences and needs.The students rejected already available apps that mainly detected mistakes or criticised their playing speed.Instead, they were motivated to design a solution that would enhance social interactions among their peers who played the piano.This approach empowered the students to shape the design process based on their own experiences and desires, ultimately leading to the creation of a solution that aligned more closely with their preferences and goals.By involving end-users in the research process, EdTech developers can use participatory research and design to support equity and reduce bias by integrating diverse perspectives.In the report 'Responsible Innovation in Technology for Children', UNICEF (2022) researchers highlight the need for more research on digital play to include the voices of children from low-income countries and ensuring that diversity, equity, and inclusion are central to the design of digital play for all.3.5 PARTICIPATORY DESIGNDesigning for inclusion also means co-designing with children who are often marginalised.For example, Kahoot! recently published their White Paper Designing for Inclusion(Rosenheck & Limpiti, 2024), in which they discuss their co-design process with neurodiverse users.Participatory design and co-design approaches enhance the potential of technology to foster a range of cognitive, social, and communicative skills, as Fernández-Batanero and colleagues' (2024) study with ASD (autistic spectrum disorder) students shows.To fully realise the potential of EdTech to become "a fundamental ally to cultivate cognitive, social, and communicative skills"(Fernández-Batanero et al., 2024), educational technology should not only open up learning opportunities but also enhance active participation in society.However, to reach that potential, the EdTech industry must prioritise participatory and inclusive design practices to address biases effectively.There is a long tradition of child-centred design in children's media and humancomputer interaction studies, which can be considered the hallmark of participatory design of digital technologies with children.Various techniques are used by researchers-designers in these studies to elicit children's responses, in a cycle of iterative development that incorporates children's views directly into the design of the technologies.For example, drawing, photography, cultural probes (toolkits with various materials including craft-making materials), and arts-based methods have been used as techniques to expand adults' perspectives on design individual students using EdTech tools.These models unintentionally emphasise what learners lack or where they fall short of normative standards, locating issues within the student rather than considering the broader context.While some EdTech systems can empower students with different problem-solving strategies, defining students solely by what they lack can have severe consequences, particularly for those from diverse cultural backgrounds.
These principles aim to recognise the multidimensionality of learners and learning experience, acknowledging differences in knowledge, skills, misconceptions, strategies, metacognition, emotions, engagement, interests, values etc. Research on assets-based approaches emphasises the resources that students bring from their home and community, including linguistic and cultural practices (e.g.MacSwan, 2020).Conceptualising and defining learner assets expansively (principle no.1), "differs from the common practice of simply identifying and fixing shortcomings in [learner's] knowledge" by focusing on "developing methods to incorporate students' existing knowledge"(Ocumpaugh et al., 2023) by, e.g., drawing on research on measuring sociocultural factors that influence student engagement (e.g.Duran et al., 2020 on students' sense of belonging) and incorporating these as additional data sources that have the potential to reveal a wider spectrum of students' strengths and can assist educators in better understanding their learning contexts.Within the context of AIED (Artificial Intelligence in Education), based on the Funds of Knowledge and asset-based approaches, Ocumpaugh and colleagues (2023) propose five principles to guide the development of EdTech: Conceptualise and Define Learner Assets Expansively.1. Detect Assets that Students Possess or Have Access to. 2.
orientation rather that deficit based COMMON SENSE MEDIA While industry-related equity certifications are not yet widely utilised, there are several worth noting.Equity related questions are being addressed by various emerging frameworks and rubrics, these often fall into the areas of data interoperability or data privacy, for example, Common Sense Media Privacy ratings (https://privacy.commonsense.org/resource/privacy-ratings),though not specifically focused on EdTech only, are based on Quick, Basic, and Full privacy evaluations.Their 2021 'State of Kids' Privacy Report' focuses on practices in e.g.data sharing, collection and practices but also includes ads and tracking.Common sense also focuses on digital equity, for their work see https://www.commonsensemedia.org/what-we-stand-for/digital-equity.
offers 'Prioritising Racial Equity in AI Design' certification (https://productcertifications.microcredentials.digitalpromise.org/explore/1prioritizing-racial-equity-in-ai-design-2),which focuses on mitigating racial bias in EdTech products through intentional design efforts.Key features include: a) identifying points of risk for racial bias in algorithmic training, b) establishing processes for accountability and transparency, implementing practices to minimise or eliminate racial bias in design, c) regularly updating of an Assumption Log to ensure accountability for actions taken to mitigate bias, d) identifying significant associated risks in products driven by AI or machine learning models, e) using practices to mitigate racial bias in datasets, training, or design of AI algorithms, f) integrating user-friendly feature allowing educators to override model decisions for individual learners, g) providing view access to primary inputs informing model decisions, and h) posting public statements on the website identifying specific ways the product minimises racial bias in datasets and training algorithms.Other product certifications by Digital Promise that cover aspects of equity include the 'Learner Variability Product Certification' (for EdTech that have earned this certification, see https://productcertifications.digitalpromise.org/learnervariability-certified-products/)and 'UDL Product Certification'.Earlier this year, Digital Promise's Centre for Inclusive Design released a white paper 'A New Narrative: How Unlocking the Power of R&D Through Inclusive Innovation Can Transform Education' (Smith & Young, 2024) and their product certifications have been revised to include 'competency-based' and 'research-driven' certifications.EDDS EDDS's (https://www.edds-education.org/)five-point 'Manifesto for Education and Technology' addresses equity indirectly by focusing on the importance of diversity.Their EdTech Quality Framework section, 'Lawful, ethical and safe', 'Socio-ethical requirements' and 'Algorithmic fairness and human rights' At the organisation level, particularly with the increasing volume of AI-based applications, equitable process for addressing bias is of utmost importance; this includes identification of bias type and points of risk in algorithmic training and implementation of clear and transparent practices to minimise or eliminate bias in design.It also includes: a) regular updating of assumption logs to ensure accountability for actions taken to mitigate bias, b) use of practices to mitigate bias in datasets, training, or design of AI algorithms, c) integration of user-friendly features allowing educators to override model decisions for individual learners, d) provision of view access to primary inputs informing model decisions, e) measurement and deployment of assets to historically marginalised communities, with context and assumptions provided, and f) completion of equity audit and implementation of advocacy strategy and transparency practices.
However, among the collocate themes we have identified (Figure1), there are also two smaller thematic groups: one of them concerns agency, that is who is involved, represented by the collocate students and the other group comprises concepts related to 'equity' -in addition to inclusion and fairness discussed In the World Bank's framework paper (What Matters Most for Equity and Inclusion in Educations Systems, 2016), two types of equity are conceptualised -vertical and horizontal: above, mobility also occurs.The collocate students suggests a more personalised and nuanced view of 'equity' in these texts.It shows that 'equity' links with concerns for well-being, learning experience, learning opportunities, attitudes, education outcomes.It shows that "strategies to include disadvantaged students" are needed (UNESCO, Wang, 31/10/2023); the aim is to ensure success for students "from all social backgrounds" (PISA 2015) and ensuring upward social mobility is likewise essential (UNESCO, Wang, 31/10/2023).

Table 1 .
Adapted exemplify this in their discussion of gender equity in education by highlighting how dramatically different this may be in Global North and South.While in Global North, it is often the boys who are the focus of gender policy, in Global South, the main concern remains girls' access to schooling.What follows from Rizvi and Lingard's discussion is that the most nuanced approach to gender equity asks 'which boys and which girls' as gender cannot be considered contends that achieving equitable access to digital technology entails more than just supplying devices and internet access; it must consider supporting users' full engagement with digital technologies by integrating an understanding of language and literacy differences, interpersonal and communal connections, communities, and institutional frameworks in pedagogy and design.The 'digital divide' is now commonly used to characterise both global and local inequalities.Equity in EdTech encompasses addressing the digital divide or digital equity, which is particularly relevant in the Global South but not limited to it, see, for example, Ventrella and Cotnam-Kappel's (2024) case study in Canada and the 2024 US Department of Education 'National Educational Technology Plan' framing three key divides: digital access, digital use, and digital design (U.S.Department of Education, 2024).
(Mitchell et al., 2021)rom socio-technical perspectives on bias, EdTech mirrors numerous statistical and societal inequities.A distinction between statistical and societal bias is important, as focusing on bias solely from the perspective of nationally protected classes (such as gender, race, ethnicity) may overlook serious impacts on other under-investigated groups.Societal bias pertains to concerns regarding objectionable social structures represented in the data(Mitchell et al., 2021), while statistical bias encompasses more technical issues like sampling bias and error measurement.Both forms of bias can contribute to overall algorithmic bias and lead to real-world discrimination and harms.This emphasises the need for a broader and contextualised examination of bias beyond nationally protected classes to ensure equity and fairness in algorithmic systems.
Baker and Hawn (2022)a model in inappropriate ways, such as designing it for one purpose but using it for a different purposeResearchers have identified some of the characteristics beyond traditional categories that are vulnerable to bias, including, for example, urbanicity, militaryconnected status, or speed of learning.Baker and Hawn (2022)summarise key types of bias, such as historical and representation bias, see Table2for an adapted overview.14
exemplified by the 2020 UK GCSE and A-Level grading controversy; secondly, varied definitions and interpretations of algorithmic bias, from statistical definitions to broader societal implications, underscoring the complexity of addressing bias in automated systems; third, categorization of harms into allocative and representational forms highlighting the potential impact of biassed algorithms on the distribution of opportunities and resources, and erroneously portraying underrepresented groups.When social biases merge with statistical biases in algorithms, they distort representations of individuals and groups, leading to various manifestations based on different characteristics, which may have harmful consequences.In EdTech, examples of such socio-statistical biases include, for example, socioeconomic status or parental education background.Examples of the characteristics, based on Baker and Hawn (2022) are summarised in

Table 3 . Bias Characterisation Socioeconomic status individuals
from different socioeconomic backgrounds are not equally and adequately represented in datasets, leading to algorithmic systems that may not consider specific challenges or circumstances faced by, for example, individuals from lower socioeconomic status backgrounds
These indicators are precise, assume responsibility for equitable practice at the company level, and define inclusivity across four measurement areas: access, skills, use, and innovation, see Table4for the detailed criteria.

Table 4 .
Summary of key deliverables of 2023 Digital Inclusion Benchmark Report.

Equity Indicators on the Product Level Equity Indicators on the Organisation Level
In Table5below, we mapped equity indicators at the product and organisation levels onto existing frameworks to formulate equity indicators for EdTech products and organisations.

Table 5 .
Equity monitoring and evaluation indicators in EdTech solutions.

Table 6 .
An adaptable example rubric of M&E criteria of equitable design.