Placing Blame or a Call to Action? An Analysis of Uwezo in the Kenyan Press

Citizen-led assessments were developed in the mid-2000s to gather evidence on children’s learning outcomes following the rapid increase in primary student enrolment in the developing world. Integral to their philosophy is social accountability and community engagement. While most large-scale assessments focus on test-based accountability to promote policy shifts, citizen-led assessments, utilizing the information-accountability causal chain, theorize that information of low learning outcomes, disseminated widely, will engage stakeholders in activities and debates, ultimately leading to improvements in education quality and service delivery. This paper examines Uwezo, the oldest citizen-led assessment in East Africa, and its portrayal in the Kenyan print media. I explore whether discourse concerning Uwezo exposes education quality issues while promoting citizen engagement and how Uwezo’s social accountability philosophy is used in public discourse through the lens of political evolutionary mechanisms. I find that Uwezo’s findings were prominently discussed beginning in 2013, but that shock value of low learning levels has since decreased. Moreover, public discourse has highlighted cross-national and provincial/county comparisons in achievement levels while focusing on blaming teachers and the education system’s culture. Although Uwezo has succeeded in disseminating its findings widely in Kenya, there is minimal engagement in public discourse with its social accountability and community engagement philosophy and discourse does not promote citizen action on a national level. As Uwezo has become a respected education Education Policy Analysis Archives Vol. 28 No. 185 2 policy player in East Africa, and globally, it must find ways to engage communities to act upon its findings to directly improve education quality.


Placing Blame or a Call to Action? An Analysis of Uwezo in the Kenyan Press
Education through formal schooling became an integral part of the global development agenda in the late 1980s and early 1990s. While the initial Education For All (EFA) goals included targets on literacy, early childhood care and lifelong learning, the preoccupation was achieving universal primary education. This singular goal, measured by enrolment in basic schooling, was put in the global spotlight with the second Millennium Development Goal (MDG) and became the dominant objective within the global education agenda (Barrett et al., 2015;McGrath, 2014;Mundy, 2006).
To achieve EFA and MDG 2's central education goals, numerous countries in the developing world, especially in Africa, began implementing Free Primary Education (FPE) policies at the national level. While many countries had attempted to increase enrolment at the primary level beginning in the 1960s and 1970s, providing free, or almost free, education became the preeminent policy to achieve universal primary education (Burns et al., 2003;Lewin, 2009;Riddell, 2003). The introduction of FPE in the late 20 th and early 21 st century allowed for millions of children who previously would not have enrolled in formal school to begin attending. The number of children enrolled in primary school in sub-Saharan Africa increased by approximately 67.5 million between 1990 and 2014. Rapid and large increases in primary school enrolment can also be observed in South Asia, the Middle East and North Africa (World Bank, n.d.). As countries looked to mass education to improve their development prospects while following the goals set out by international organizations, measurement, funding and understanding of the impact of FPE on education quality was largely left to the side (Napier, 2014).
Beginning in the mid-20 th century, some education organizations in countries where enrolment was rapidly rising, much to the celebration of governments and the public, noticed that there was little evidence about whether children were actually learning after the push for universal primary education. As the global education agenda moved from focusing on access to learning, the citizen-led assessment movement, starting with Pratham, an Indian education NGO, began conducting citizen-led assessments in 2005 to answer the question "are our children learning?" Since 2005, 13 other developing countries have implemented, or are planning to conduct, a citizen-led assessment (Byker & Banerjee, 2016;PAL Network, 2017a). The citizen-led assessment movement is an attempt by civil society organizations to gather evidence on basic learning outcomes countrywide to complement the largely successful increases in school enrolment and to use it for two purposes: to increase public awareness of the poor quality of education and to stimulate actions by numerous stakeholders to improve learning levels (Eberhardt & Burnett, 2015).
In East Africa, Uwezo (meaning "capabilities" in Kiswahili), the continent's first citizen-led assessment, began operating in 2009. Uwezo operates in Kenya, Tanzania and Uganda under the guidance of Twaweza, a civil society research group. Uwezo, like other citizen-led assessments, evaluates basic reading and numeracy skills using trained volunteers who conduct the assessment one-on-one within a household setting for all children between the ages of 6 and 16. The philosophy behind citizen-led assessments is that a citizen-driven approach to social change, in which citizens hold schools and governments accountable for poor education quality as well as enabling communities to put forth localized solutions, will be a catalyst for improving education quality (Uwezo, n.d.a).
This paper examines a main aspect of Uwezo's theory of change: the dissemination of information and results to create awareness, debate and action (Uwezo, n.d.b). To spur debate and engagement from the public, Uwezo disseminates the results of its assessments in two ways.
Volunteer assessors 1 provide instant feedback to the head of household concerning their child(ren)'s learning level and Uwezo publishes reports used by media, researchers and policymakers for broad communication about the results. In this paper, I examine the nature of the latter using Kenya's two most read newspapers, the Daily Nation and the Standard. News media was analyzed to gather information on the portrayal of Uwezo and discourse surrounding its findings in widely dispersed mediums. This information was compiled and analyzed to understand whether there was any mention of actions citizens need to take or skills they need to possess in order to improve education outcomes, which is the basis of Uwezo's theory of change and an integral step in information accountability.
Media analysis of education issues and learning assessments have recently been used by comparative education scholars to understand how historical, political and cultural contexts can influence education policy and public opinion. Newspapers, as well as other media outlets, are examined to understand how public problems are framed and who is given the opportunity to comment. Analyzing media provides an insight on how public discourse is shaped concerning education issues and how it can sway policy and public action (Green Saraisky, 2016). Integral to public discourse's influence on policy and action is understanding how the public understands and uses new information and how this information can influence policymakers (Steiner-Khamsi, 2003).
I will examine public discourse through the lens of political evolutionary mechanisms, or how new policies are introduced in the political and public realm through variation (development of new ideas due to challenges or dissatisfaction), selection (identification of solutions deemed suitable by the public) and retention (policymaker action; Jessop, 2010;Verger et al., 2017). Specifically, I relate Uwezo's theory of change to the variation and selection cycles of the evolutionary mechanisms. However, out of the scope of this paper is an examination of how citizens may engage to improve education in their communities based on their interaction with Uwezo results, or how other policy actors use Uwezo results to advance public policy in Kenya (retention).
The mass dissemination of its results, to demonstrate a need for variation, is integral to Uwezo's goals. While the first step is conducting a valid assessment of learning on a national level, the second step, dissemination and creating an informed public, is the initial push towards engagement and accountability. As an integral part of Uwezo's theory of change, the way in which assessments results are presented and how public discourse is impacted by results are important for understanding how the public may interact or "select" solutions to the education crisis that they can enact or push policymakers to legislate. While the presentation of information has a role in allowing citizens to act and hold state actors accountable, Uwezo's goal is to promote citizen-led solutions alongside assessment results in the hopes of spurring engagement. However, we know relatively little about the media coverage of citizen-led assessments, and whether the discourse surrounding these assessments solely uncovers challenges to quality education or if it also identifies and promotes solutions that are rooted in community action. The goals of this paper are to add to literature concerning social accountability projects by researching a growing assessment and engagement phenomenon in the developing world, to understand how public discourse is impacted by a new large-scale assessment, and to examine whether the discourse includes this endeavor's philosophy on stimulating local engagement to improve education.
To understand how this dissemination influences public discourse, I asked the following three research questions a) How is Uwezo portrayed in the Kenyan print media? b) How is 1 Volunteers are selected from each district and trained to conduct a household survey and short reading and numeracy tests in each household selected. Over 9,100 volunteers were mobilized in over 69,000 households for Uwezo 2015 in Kenya. According to Eberhardt and Bennet (2015), volunteers reliably assess children's competencies and are integral to the assessment's functioning.
information provided by Uwezo used in the print media to frame education issues and to spur debate, responsibility and possible solutions? and c) Is Uwezo's philosophy of citizen-led engagement or theory of change presented in the print media?

Social Accountability
The association between an informed public, social accountability and citizen engagement has been central to the study of the quality of democracy. Social accountability, sometimes referred to as community accountability, does not have a universally agreed upon definition, but can be understood as a citizen and community-led approach to improve public accountability and to further engage these stakeholders to improve services and governance (Joshi & Houtzager, 2012;Malena & McNeil, 2010;O'Meally, 2013). Theoretically, information plays an integral role in ensuring that citizens can act to keep state actors accountable. Relevant information allows citizens to understand the reality of the quality of governance and services and, in the right circumstances, pushes them to act if the quality does not meet expectations (Bjorkman-Nyqvist et al., 2014;Olken & Pande, 2012;Stromberg, 2015). The idea that information can be a first step in empowering citizens to hold governments accountable for social services as well as engage them to act collectively or individually to improve services has been researched in numerous fields, especially since community accountability projects have been increasingly funded by international organizations in recent years (see Bukenya & King, 2012).
Information and social accountability studies often focus on improving service delivery, especially concerning civil and political actions and in the fields of health and education. Studies have demonstrated positive results when examining the link between informed citizens and participation in voting, protesting, and overall political engagement (Banerjee et al., 2011;Booth, 2012;Gerber & Green, 2000;Hossain, 2009Pande, 2011. While there is an abundance of literature on the positive impact of information on improving community accountability, there are also instances where there is no effect on political engagement (Chong et al., 2012;Humphreys & Weinstein, 2012). Social accountability regarding health service delivery has also garnered a fair amount of research. Informed citizens have been shown to monitor community health facilities (Bjorkman & Svensson, 2009) and to push governments for increased funding for indigenous health services (Shankland, 2010).
Education and school quality have also been subject to social accountability studies, with citizen-led assessments acting as one facet of the research. Studies have focused on how general information campaigns about the importance of education and community involvement (Jensen, 2010;Pandey et al., 2009), community radio programming on parental involvement in schools (Keefer & Khemani, 2011), newspaper campaigns about education funding (Reinikka & Svensson, 2005) and student report cards given to parents (Andrabi et al.,, 2017) lead to improvements in service delivery and motivating citizens to act. While social accountability projects are recognized as cost-effective education development projects by the international community (see International Commission on Financing Global Education Opportunity, 2016), studies over the past two decades have demonstrated both successful and discouraging results on their effect of improving service delivery and engaging citizens. This paper therefore adds to this literature by focusing on whether public discourse can include both information (assessment results) and a concerted push to increase awareness and responsibility of citizens in hopes of increasing engagement and locally-led initiatives.

Information-Accountability
Informed citizen action is at its root a localized endeavor. The preceding examples of social accountability research demonstrates the micro-level analysis and context specific cases in which these activities and campaigns occur. To tease out how information affects micro-level social accountability projects, Lieberman, Posner and Tsai (2014b) and Joshi (2014) proposed causal chains to best understand the micro-contexts which lead to successful results. While incorporating social accountability theories, the causal chain focuses on the theories in practice. An informationaccountability causal chain (see Lieberman et al., 2014b, p. 77) begins with information aimed at the public, which under the right circumstances leads to citizen engagement and pressure, causing improved service delivery and citizen action, ultimately improving the quality of services. For information to lead to welfare improvements: it must be understood; it must cause people to update their prior beliefs in some manner; and it must speak to an issue that people prioritize and also believe is their responsibility to address. In addition, the people at whom the information is directed must know what actions to take and possess the skills for taking these actions; they must believe that authorities will respond to their actions; and, to the extent that the outcome in question requires collective action, they must believe that others in the community will act as well. And, of course, they cannot already be doing everything that is possible for them to do (Lieberman et al., 2014b, p. 70) These conditions must be met for social accountability projects to succeed. The proposed causal chain could be used to study the effectiveness of social accountability initiatives, as well as to set guidelines to use in planning and implementing new projects, such as citizen-led assessments (Joshi, 2014). This paper, in similar conceptual fashion to Lieberman, Posner and Tsai (2014b), explores whether public discourse developed in the media prioritizes the fourth step in the causal chain (does the citizen now feel responsibility to act) and the fifth step (is the citizen now aware of what actions to take), ultimately leading to new citizen action. These two elements of the causal chain are closely related to the variation and selection policy evolutionary mechanisms, which provide a useful framework for understanding how information leads to effective understanding of the problem and possible solutions to the issue at hand.

Test-Based Accountability vs. Social Accountability in Citizen-Led Assessments
Using assessments to promote accountability of education actors, namely teachers and school administrators, has become one of the most persistent global education policies in recent decades. Whether through the growth of cross-national or national assessments, using tests has become an integral component for governments and the public to hold those exerting leadership in schools and classrooms accountable (Anagnostopoulos et al., 2013;Lewis & Lingard, 2015;von Davier et al., 2013). Test-based accountability focuses extensively on results and the numbers as objective measures of effectiveness, and provides ample space for comparisons, blame and scandalization. There is a strong relationship between data or results, politics and accountability mechanisms, with very little room for a democratic process of deciding why these results matter, what to do with these results and how to turn results into improvements (Henig, 2013). Lingard, Martino and Rezai-Rashi (2013) call this use of tests "vertical, one-way, top-down" accountability, which ultimately stifles any other type of accountability (p. 544).
In comparison to test-based accountability, citizen-led assessments use their testing agents to promote social accountability. While the use of an assessment to measure achievement or learning outcomes, on various levels and using disparate methods, are common, the nature of accountability mechanisms, who they attempt to keep accountable and who leads these processes, are very different. In theory, social accountability invigorates civil society and the public to act towards their own benefit, ultimately pushing service providers to work with and listen to the community (Bjorkman-Nyqvist et al., 2014). In practice, ordinary citizens are the primary actors using information and their actions aim to hold fellow community members, and those who hold power positions, accountable. For citizen-led assessments, information stemming from immediate feedback to parents, reports and media dissemination of findings, provides an opportunity to promote citizen actions that hypothetically lead to educational improvements through greater government accountability and community involvement (Lieberman et al., 2014a). In theory, Uwezo attempts to increase public awareness about the learning crisis and incentivizes citizens to take responsibility and to learn skills for taking action to create an environment for improving education.

Literature Review
As a relatively new and growing phenomenon in the developing world, there is a lack of academic literature on citizen-led assessments and its social accountability philosophy, and even less about media dissemination and public discourse. The most complete examination of citizen-led assessments is a Results for Development Institute report by Eberhardt and Burnett (2015) which explored four tests, including Uwezo, to determine how well these assessments measure learning, how well the assessments' processes work and if the assessments stimulate action and awareness. The report used media sources briefly as part of its examination on action and awareness. It concluded that dissemination of results is strong nationally, especially using print media, but that actions on local levels stemming from this information have been difficult to observe.
Other studies have explored different aspects of citizen-led philosophy and methodology. Banerjee et al. (2010) examined the impact of the Indian citizen-led assessment ASER, which was then used to create community report cards, in one state. The study found that this type of intervention, along with the assessment, made more people aware of the low learning levels but did not increase parental involvement with schools or learning outcomes. Byker and Banerjee (2016) examined ASER's participatory action research methodology to understand how volunteer assessors conducted and engaged with the assessment. The mixed-method study of 37 volunteers found that most volunteers want to serve to improve education in their communities and that ASER's methodology is fundamental to citizen capacity building and facilitates active citizenship. The Australian Council for Education Research (ACER) has also done research on citizen-led assessments, including reports on their validity (see ACER, 2015), their impact (see Tobin & Lietz, 2016) and their monitoring tools (see ACER, 2014).
Research on Uwezo conducted by Lieberman, Posner and Tsai (2014a;Carlitz & Lipovsek, 2015) for Twaweza, and the subsequent Lieberman, Posner and Tsai article (2014b), have been the most substantial academic work done on a citizen-led assessment and its theory of change. The study explored how the direct reporting of assessment results to parents along with materials given to parents about the importance of their involvement to improve learning impacted expansion of parental action. In short, the study found that the information interventions did not have any significant effect on parent's behaviors or attitudes, ultimately demonstrating a flaw in Uwezo's citizen engagement philosophy. Although this study did not include research on the widespread dissemination of results using media, the results do demonstrate a need for Uwezo and other citizenled assessments to better understand and act upon the conditions in which information leads to actions.
Research on media dissemination of citizen-led assessments has been undertaken by assessment organizations and their partners. The most prominent examples include media surveys conducted by ASER Centre (2013) and Khamis, Mwaipopo and Mugiumi (2013) for Twaweza. Both reports demonstrate the high visibility of citizen-led assessments in the Indian and East African media, especially in newspapers, creating strong national awareness of the assessment results. In addition, the People's Action for Learning (PAL) Network, the global network of citizen-led assessments, includes a section on "media coverage" with links to relevant online sources. 2 However, there is no significant research on what is included in these newspaper articles and how public discourse is influenced by the assessments.

The Case: Kenya
Kenya is one of the wealthiest countries in Eastern Africa, with a GNI per capita of nearly $1,600 in 2018, and one of the few that have reached lower-middle income status according to the OECD. According to the World Bank, over 70% of the population lives in a rural area, and there are nearly 20 million children under the age of 15 in the country, accounting for just over 40% of the total population (World Bank, n.d.). Public expenditure on education in Kenya is 5.3% of GDP. In addition, Kenya received between $125 and $225 million annually between 2012 and 2016 in official development assistance specifically for education (OECD, n.d). Although Kenya is a strong economic actor in the region, political instability and corruption persist, and issues concerning presidential elections and subsequent violence have made international headlines in recent years.
A first attempt at FPE was introduced in Kenya in the late 1970s, but government shifts and a need to cut spending abolished the program a decade later. In the early 2000s, with MDG 2 recently put in place, the Kenyan government began to work towards another FPE program. In 2002, the newly elected government adopted FPE as one of its essential political promises and initiated FPE countrywide in 2003 (Muyanga et al., 2009;Riddell, 2003). The impact of FPE was immediate; primary enrolment surged from 6 million students to over 7.1 million between 2002 and 2003. The rapid increase in enrolment also affected classroom environments and school materials. For instance, the student-teacher ratio increased from 28 in 2002 to 38 in 2003, and has since remained around 40 (World Bank, n.d.). Although FPE increased primary enrolment and ensured that Kenya was moving towards universal primary education, the program impacted numerous education quality indicators, which have been difficult for the government to improve.
According to the Ministry of Education, Science and Technology, the main goal of the National Education Plan 2013-2018 is to ensure quality basic education and to improve school outcomes. The shift from access to outcomes, while still ensuring that all children are enrolled in school, has guided the ministry's education priorities (Ministry of Education, Science and Technology, 2013). The Kenya National Examination Council (KNEC) is the government branch charged with implementing assessments and exams countrywide. The two main examinations administered by KNEC are the Kenyan Certificate of Primary Education (KCPE), the results of which are used to determine placement at secondary schools, and the Kenya Certificate of Secondary Education (KCSE). In addition, KNEC is charged with administering and organizing national assessments.
The National Assessment System for Monitoring Learner Achievement (NASMLA) compliments the public examinations, and in similar fashion to other national, regional and international assessments, acts as a diagnostic of the education system rather than individuals or schools. NASMLA results have been stable over time and demonstrate gender and location disparities. Kenya has also participated in all four rounds of SACMEQ (The Southern and Eastern Africa Consortium for Monitoring Educational Quality). SACMEQ has completed four assessments since 1995 of students in Grade 6 in 15 countries. The assessment has a strong focus on building capacity and technical expertise in each participating country and generates research which can inform policy and enables cross-country comparisons. The high country participation in the region and the assessment's leadership in training researchers and policymakers as well as its inclusion in SDG 4.1 monitoring has led to the growth and legitimacy of SACMEQ (Ramalingam, 2017). Kenya has repeatedly scored high on the SACMEQ assessments and has been one of the top performers in the region. Kenya's above average performance is highlighted in the most recent national report (Karago et al., 2019;SACMEQ, n.d.).
Uwezo's assessment presents a more critical image of Kenyan students' achievement. Since its inaugural assessment in 2009, Uwezo has found that most children aged 6 to 16 are not learning at grade level, in relation to the national curriculum. For instance, the first Uwezo cycle found that two out of three Standard 2 students could not read a Standard 2 level paragraph (the second most difficult stage assessed) in English or Kiswahili. Early reports also found that nearly ten percent of students in Standard 8 could not complete simple division, a concept included in Standard 2 math curriculum (Uwezo, 2012). Uwezo's findings have remained consistent since 2009. Figure 1 presents Uwezo's findings in Kiswahili, English and Math for all children assessed using its highest level achieved scoring methodology. 3 Overall, Uwezo's results demonstrate a learning crisis, in which many children are not mastering the basic content of the Standard 2 curriculum.

Figure 1
Uwezo Kiswahili, English and numeracy results of children aged 7-13 at Standard 2 level, 2011-2014 3 Volunteers administer the assessment and move from one level to another in the test booklet and places the child at the appropriate level. For example, if a child is successful at reading words aloud but cannot complete an entire paragraph, the child's level is "Word." This highest-level methodology is used in the numeracy assessment as well. Standard 2 level in Kenya is "story" in literacy and "division" in numeracy.

Methodology Data Collection
To examine how Uwezo is portrayed and how public discourse on education issues is shaped by its findings, I constructed an original dataset of newspaper articles published online between 2010 and 2017. 4 I searched for articles in two Kenyan newspapers: the Daily Nation and the Standard. The Daily Nation is the largest newspaper by daily circulation in Kenya, followed by the Standard. The Daily Nation is considered the paper with the most balanced reporting though it tends to lean liberal while the Standard is described as more conservative (Maina, 2006). Although newspaper circulation in Kenya is low, 320,000 per day in 2010, readership is significantly higher considering that on average about 10 people share a newspaper (Nyabuga & Booker, 2013). In addition to print circulation, the Daily Nation and the Standard are the two most popular news websites in Kenya with almost national reach and rank 10 th and 15 th respectively among the most popular websites in the country (Nyabuga & Booker, 2013). Both newspapers also have sizeable social media followings on Facebook and Twitter (nearing or over a million followers on each platform). Both newspapers publish articles in English. Additionally, the Daily Nation's website includes articles in Kiswahili, which are written by the same journalist as the originally published English articles and are circulated in Taifa Leo, the largest Kiswahili newspaper. I only read articles in English on the Daily Nation's website.
I used the search option for both the Daily Nation's and the Standard's website as well as each newspaper's Twitter feed to find relevant articles. Regarding the online search protocol, I inputted the following key words: "Uwezo Education", "Uwezo Learning", "Uwezo Assessment", "Learning Assessment", and "Learning Levels". Searching simply "Uwezo" returned results primarily concerning the Kenyan government's Uwezo Fund for financial and business support. I read through each article before finalizing the dataset to ensure that the articles referenced Uwezo as either the assessment or the organization. Four articles were subsequently removed from the dataset due to lack of relevance. The final dataset included 77 newspaper articles. Nearly three-fifths (58%) of the articles were published in the Standard, while slightly more than two-fifths (42%) were published in the Daily Nation. The dataset has two immediate limitations. First, the dataset includes only articles that were found on the newspapers' websites. According to a 2014 University of Nairobi Master's thesis, the online version of the Daily Nation lacked several articles that were present in the print newspaper (Ngoge, 2014). Therefore, the dataset may be missing articles, especially near the beginning of the dataset's timeframe. Second, while the two newspapers used have the largest readership in Kenya, online articles may not be as accessible to people who do not have regular access to the internet. Therefore, readership may be concentrated primarily in certain areas. Nevertheless, in the context of this paper, the dataset provides a strong insight into Uwezo's presence in the print media.

Coding
To analyze public discourse of Uwezo, I developed a detailed qualitative codebook and questionnaire for background variables. I employ Pizmony-Levy's (2016) methodology for analyzing public discourse of assessments, which includes documenting background information, identifying themes repeated across articles, as well as speech acts and presentation of assessment findings. Coding was initiated by reading each article in the dataset to understand the central education issues presented, as well as by reflecting on how other scholar examined media and assessments (see Green Saraisky, 2016;Pizmony-Levy, 2016;Wadlow et al., 2014). The codebook was created to understand main themes that were recurrent in the articles and how Uwezo informed or initiated these themes. For instance, I unsurprisingly found numerous articles which referenced the low education quality of Kenyan primary and secondary schools. This became a theme titled "low learning levels". Some articles referenced education issues in the context of national development, or more specifically Kenya's national development plan Vision 2030. This became the "national development" theme. There were also some instances of questioning the validity of Uwezo's assessment and data, which became the basis of the "negative perceptions of Uwezo" theme.
The dataset also included background and content variables. Variables included month and year of publication, location of the article (1=news section, 2=opinion, letter to the editor or feature, 3=re-published piece), objective of the article (1=release of Uwezo report and data, 2=Uwezo data as part of an article on an education issue, 3=Uwezo representative as part of an article on an education issue) and regional focus (1=East Africa, 2=Kenya, 3=County). I also included three content variables to better understand what information was presented to the reader. Content variables included information about Uwezo (1=Uwezo's goals or operations, 2=Uwezo as an NGO or thinktank, 3=Uwezo as a lobbying organization, 4=no information), inclusion of Uwezo data (0=no, 1=yes) and inclusion of Uwezo representative analysis (0=no, 1=yes).

Descriptive Patterns
To begin, I examined the descriptive statistics of the dataset. The number of articles which referenced Uwezo was low, between three and eight, from 2010 until 2012, the first three years in which Uwezo released assessment reports. However, following the release of the 2012 Kenyan assessment report in July 2013, the number of articles concerning Uwezo increased to above 10 and remained between 10 and 15 from 2013 to 2017. Figure 2 presents the number of articles published per year. It is evident from the spike in articles that it took nearly three years for Uwezo to have gained recognition as an important and trustworthy assessment of Kenyan students on par with other national assessments and as a respectable education organization.

Figure 2 Number of articles referencing Uwezo, per year
The publication of articles is associated with the release of Uwezo's Kenyan and East African assessment reports. Uwezo has released five Kenya assessment reports between 2010 and 2015, the exception being a report for 2013. Each assessment is accompanied by a report published the following year. In addition to the national reports, Uwezo also publishes East African regional reports in select years. Reports do not have a set release date and therefore have been released in different months. There are 24 articles (32%) which were released immediately following the publication of an assessment report. Of these articles, 13 were published in tandem with the release of a Kenyan assessment report, seven focus on a regional report, three examine data released by Uwezo specific to certain counties and one article provides information on the Uwezo team who worked on the 2012 Kenyan report. Regarding location of these articles, 20 were published in the news section, three were immediate opinions about the results and one was the piece on the Uwezo team. Interestingly, while most articles published concerning Uwezo are written in close proximity to the release of a report, there was also a cluster of articles which utilize Uwezo data written in late December and January, the timeframe for the release of national assessment results.
Regarding the location of Uwezo-related articles in the newspapers, 38 articles (49%) were published in the news section while the remaining 39 articles (51%) were opinion pieces, features or weekend essays. As stated above, 24 articles (32%) examine the release of Uwezo reports, while 32 articles (42%) use Uwezo findings in the presentation of an education problem or news story. The data used in these articles corroborate narratives that there are serious faults in the Kenyan education system. While most of the data references low learning levels, other aspects of Uwezo data, such as teacher absenteeism, education material and school infrastructure, are mentioned as well. The remaining 21 articles (27%) are either opinion pieces written by Uwezo staff members or are articles in which Uwezo is mentioned due to quotes by Uwezo representatives. These articles have a much wider focus and go beyond discussing low learning levels. For example, Uwezo staff members have been given speech acts in articles concerning the new national curriculum, national primary and secondary assessments and the growth of private schools.
Uwezo's learning assessment, as well as the household, school and village surveys which accompany the yearly assessment, is the predominant output of the organization. The importance and originality of this data is clearly demonstrated in the articles. 55 articles (71%) include some sort of data in which Uwezo has collected and released. The shock and outrage of the Uwezo data, especially after the first few reports, is exemplified through headlines such as "Shocking report on literacy levels in school" (Standard, 2010), "Class Four pupils 'can't do Class Two work'" (Standard, 2011) and "Kenyan schools churning out 'illiterates'" (Standard, 2012). Information presented in the articles about Uwezo's work, philosophy or structure was also collected. 56 articles (73%) did not include any further information other than stating Uwezo Kenya or Uwezo East Africa, or cited it as an initiative, assessment or survey. Two articles (3%) used the term "lobby" to describe Uwezo while an additional 11 articles (15%) portrayed Uwezo as an education thinktank or NGO. Only eight articles (11%) included information about Uwezo's goals, philosophy or endeavors.

Strong Dissemination of Results, but Shock Value is Waning
The results of Uwezo's learning assessments have been successfully disseminated in the national Kenyan print media. The 24 articles written specifically about the release of new Uwezo reports, along with the more than 50 articles that incorporate Uwezo education statistics, demonstrates the wide-ranging dissemination of Uwezo's findings in Kenya. Overall, of the 77 articles in the dataset, 56 articles (73%) specifically mention, in some capacity, the low levels of learning and poor quality of education in Kenyan primary and secondary schools as demonstrated by Uwezo's assessment.
The specific way Uwezo issues its findings regarding inadequate learning levels has been emulated by the Daily Nation and the Standard. The following excerpt, written in "Shock of illiterate KCPE pupils" (Daily Nation, 2015), demonstrates how many of the articles examining Uwezo's results were presented: Four out of 100 children in Standard Eight cannot read a Standard Two Kiswahili or English story, a study has revealed.
[…] Learning outcomes are low nationally, with only three out of 10 pupils in Standard Three being able to do Standard Two work. The programme tested more than 135,000 children aged six to 16 in 2013. About four out of 10 Standard Three pupils can read a Standard Two Kiswahili story while in the same class, two out of 10 cannot read a Kiswahili word. In English, three out of 10 Standard Three pupils can read a Standard Two English story while two out of 10 pupils in the same class cannot read an English word. It also indicates that three out of 10 Standard Three children can solve a Standard Two division problem while in the same class, one in 10 children cannot recognize numbers between 11 and 99.
This example, as well as the other articles which present Uwezo's findings, provide the reader with a strong understanding of the gravity of the education quality issue in Kenya. These findings are consistent with Eberhart and Burnett's (2015) report about the successful dissemination of assessment results using the print media. While there is national coverage of these results, more recent Uwezo reports and data on low learning levels have not had the same shock value in the print media. These articles have demonstrated a sense of normalcy and despair with the inadequate learning results. For instance, the article "Church, not school, is where kids should study religious education" (Daily Nation, 2016) states that "our children are bad at these [learning English and math], as Uwezo keeps reminding us." Another article, "Some KCPE exam pupils cannot handle class 2 homework" (Standard, 2016) explains that Uwezo "has once again returned a damning verdict on Kenya's education system." Other post-2015 articles describe the results as consistent, and do not demonstrate the same surprise in Uwezo's findings. In addition, the proportion of articles which include data or statements on low learning outcomes has decreased since 2015, as observed in Figure 3. The decrease of media coverage on learning outcomes, which have remained relatively stagnant since 2009, may signal that the public is well aware of the lack of quality education in Kenya, and that Uwezo has succeeded at its initial goal of widespread awareness.

Figure 3
Articles including and not including information on low learning levels

Lots of Comparisons, Little Context
Since Uwezo's Kenya reports and data are disaggregated by county (or province prior to 2013), comparisons are inevitable. The dataset includes 20 (26%) articles which offer a comparison between specific counties in Kenya. There are 16 articles which describe the release of a Uwezo report which include county comparisons, demonstrating the pervasiveness of comparisons when results are first disseminated. Although showing data disaggregated by county is included in Uwezo reports, these reports also state that differences in county results are strongly connected to geographic (urban vs rural) and socioeconomic disparities and that there is correlation between disparities and results. However, of the articles which include county comparisons, only four mention socioeconomic disparities and seven include statements on urban vs rural issues. County comparisons which do not include contextual information, most notably geographic and socioeconomic, may be misleading to the reader and promote competition instead of debate or solutions. Furthermore, closer examination of contextual factors within county comparisons may spark greater interest in why some counties perform better than others, ultimately leading to debate and action on plans or ideas which improve education quality.
Uwezo is the only citizen-led assessment which encompasses more than one country 5 . While Uwezo is split into three country offices, assessments and surveys are conducted in the same manner throughout Kenya, Tanzania and Uganda, although curriculum expectations differ. Therefore, country comparisons, to a lesser extent, have also been reported by the print media on seven occasions. There are seven articles (9%) which compare Uwezo results across East Africa. Overall, Kenya's results are the best in the region, a fact that is reported in all seven articles. County comparisons across the region are also included to demonstrate that most high-ranking counties are located in Kenya, while few Kenyan counties are in the lowest ranks. Yet, all East African comparison articles do mention low levels of learning in the region.
While there are only a few articles which include a comparison of Uwezo results across East Africa, some of these articles downplay low learning outcomes in Kenya. For instance, the article "Report shows gap in East Africa education standards" (Daily Nation, 2015) states "the [Uwezo] report notes that learning outcomes are high in Kenya and low in Uganda." The article "Kenyan pupils ranked top in the region" (Daily Nation, 2011), whose title already puts a positive spin on 5 India, Pakistan and Nepal's assessments are all called ASER, however they are conducted separately and there are no shared reports or reports are not shared simultaneously. Uwezo's report, notes that "Kenyan pupils came out on top in a standard test administered to gauge basic ability to read letters, words, a sentence and finally a paragraph" and that "Kenya scooped the first 24 slots in a ranking of 135 selected districts across the East African community." Statements like these promote rankings and celebrate Kenya's position in relation to Tanzania and Uganda, while opting to neglect the substance of Uwezo's findings. Even though articles that downplay the negative findings of Uwezo's assessment are not excessive, the regional comparisons tend to overemphasize Kenya's education success at the expense of Tanzania and Uganda (see the 2011 East African op-ed "Why being Kenyan makes you smarter, more literate.") Both county and regional comparisons may shift public discourse in a way that is not conducive to Uwezo's bottom-up philosophy. In sum, while comparisons are inevitable, a lack of context disables the reader from understanding the realities which affect education quality.

The Blame Game
As the organization that put the spotlight on the education quality crisis in Kenya, Uwezo's findings have created a nationwide debate on how to improve the education system. While policies and plans to fix the system are proposed at times in the print media, blame is more often associated with Uwezo's assessment results. There were three main actors and issues which received most of the criticism for Uwezo's results: teachers, national assessments and the curriculum. It must be noted that Uwezo's reports do not place overt blame on any actor or initiative but ponder solutions to issues that affect learning outcomes.

Teachers
At the forefront of the blame for poor learning outcomes are teachers. The claim of poor quality of teachers and the elevated rate of teacher absenteeism are presented as key reasons for low student performance. In the dataset, there are 20 articles (26%) which mention inadequacies of teachers' ability countrywide and connect the low abilities and intelligence of teachers to low learning outcomes. For example, articles include statements about "ill-prepared teachers" (Daily Nation, 2016) and teacher's obsession with "cramming and reproducing" (Daily Nation, 2017). Observations like "most of the teachers and trainers are hardly qualified or make little effort to improve on their teacher skills" (Standard, 2010) and "methods used by teachers […] concentrate on making children pass exams rather than helping them actually learn" (Standard, 2010) were common. An article even questions the intelligence of teachers by proposing that "colleges and universities should go for bright students to train as teachers" (Daily Nation, 2015).
In addition to the small jabs at the quality of teachers, two articles focused solely on the low abilities of teachers and its impact on students. The article "Performance of teachers in public schools below par" (Standard, 2013) presents the issue as follows: Despite taking a huge share of the National budget, too many teachers are asleep on the job, and unless shaken out of their slumber, the road to Vision 2030 could be just another illusion.
[…] The teachers' failure to master even elementary mathematics is putting at risk Kenya's ambitious plan to turn its economy into one driven by science and technology.
[…] Gaps in teachers' knowledge and the little time spent teaching means that pupils are leaving school half-baked. A report by Uwezo Kenya reveals that 50 out of 100 children in Classes Four and Five can't understand stories written for Class Two pupils. The study found that school-aged children are not acquiring basic competencies in literacy and numeracy at the right age or grade.
This article, which uses a World Bank report on teachers as well as anecdotal evidence, makes a direct connection between teacher quality and student performance and does not hold back on the need to be tough on teachers. Another article, bluntly titled "Standard Six pupils brighter than their teachers, new report shows" (Standard, 2014) also puts most of the criticism for low learning levels on teachers. The article, which uses a UNESCO study about Kenyan teachers, states no teacher has a complete mastery of any subject, including the ones they taught.
[…] The yet-to-be released report points out the incompetence of teaching staff are among the key factors that have contributed to illiteracy among primary school graduates, of whom 40 percent are unable to read a single sentence.
Uwezo's results are used to vilify teachers and to demand higher quality teachers, although few concrete plans on how to do so are referenced in these articles.
In similar fashion to teacher quality, evidence of high rates of teacher absenteeism were cited in 15 articles (19%). Statistics on teacher absenteeism are collected by Uwezo in their school survey and are used as the primary evidence in articles which reference teacher absenteeism. According to Uwezo Kenya, teacher absenteeism has remained between 10% and 13% throughout their assessments. Articles include comments like "[teacher] absenteeism has contributed greatly to this problem [of low learning]" (Standard, 2013), "teacher absenteeism is still very high and could be a reason why learning levels in numeracy, literacy and general knowledge is worsening" (Standard, 2013), "[low learning levels are] largely an outcome of absenteeism by teachers" (Standard, 2013) and "absenteeism by teachers is contributing to poor performance" (Standard, 2013). This blame on teachers, both overtly and clandestinely, for the reality of subpar learning outcomes is unmistakable in the print media.
In response to Uwezo's results and subsequent public criticism of teachers, the Kenya National Union of Teachers (KNUT) has occasionally denounced the assessment and its findings. An op-ed published by the chairman of KNUT titled "Uwezo's questionable report on pupil ability a reckless insult to teachers" (Daily Nation, 2013) condemns Uwezo's claims about teacher absenteeism, questions its methods and attempts to deflect blame onto the government. The following year, KNUT again criticized Uwezo's report, demonstrated in the article "KNUT trashes literacy report" (Standard, 2014) claiming it lacks authenticity and that Kenyan students placed ahead of their peers in Uganda and Tanzania. These reactions by KNUT demonstrates the tenacity of the public's blame of teachers from Uwezo's findings.

Assessment Culture
While teachers received the greatest amount of criticism for low learning outcomes, public discourse also puts blame on Kenya's assessment culture. A total of 19 articles (25%) place blame on national assessments for poor learning levels. For instance, articles include mention that lessons are "concentrated on making children pass exams rather than helping them actually learn" (Standard, 2010), ultimately making "anything that is not examinable no longer taught" (Daily Nation, 2015). The importance, and near obsession, of examination results is seen as a detriment to a well-rounded learning environment. The article "Emphasis on grades blamed for mess" (Standard, 2012), states that "education in Kenyan has ceased to be utilitarian and become a fierce rate race to accumulate impressive grades and papers at the expense of quality." The emphasis on examination results has enabled the classroom to become a place that "just judges them […] but does not help the candidate improve themselves." Assessments in Kenya, in which preparation begins when students first enter school, are presented as inhibitors of real learning and focus solely on memorization and examination preparation, ultimately hurting the quality of education.

Curriculum
Kenya's national curriculum as a reason for low learning outcomes is mentioned in 15 articles (19%). In similar fashion to assessments, public discourse on the curriculum revolves around its focus on assessment preparation and its lack of learning-centered approaches. The dominant thinking concerning the curriculum is that it is "overloaded, […] children barely have time to develop critical foundation skills such as reading and comprehension" (Daily Nation, 2015) and that "it promotes and glorifies mastery of content at the expense of everything else" (Daily Nation, 2015). The idea that excessive homework, a lack of creativity and memorization are central to Kenya's curriculum is observed by teachers, citizens and education experts alike. A more recent article titled "Bags and books bogging us down, say school children" (Standard, 2016), which examined a curriculum review conference which Uwezo representatives took part in, reported on a group of primary students who complained about the number of books they need to carry home in order to study for examinations. While there are few proposals on how to fix the curriculum, and promote foundational learning, the curriculum's laser-like focus on mastering content for formative assessments has been deemed a reason for Uwezo's findings.

Promoting Citizen Action Is Not Present in Public Discourse
Uwezo's theory of change argues that informed citizen who are involved, engaged and provide oversight are integral to sustained improvements. However, literature on social and information accountability and development initiatives have demonstrated mixed results. To follow the information accountability causal chain, the dissemination of Uwezo's results in the media would need to promote citizen-led solutions that positively impact education and push newly informed citizens to put pressure on governments to improve education. Since media is one of the main ways that Uwezo disseminates its data, it can both inform citizens about the learning crisis and prompt them to act. Yet, as Uwezo has become the exposer of a learning crisis as well as a well-respected education policy thinktank in the region 6 , there has been minimal engagement with its theory of change and citizen engagement philosophy in the print media.
Instrumental to Uwezo's theory of change and increased citizen action is the increased role of parents and community members within the education environment. 16 articles (21%) included some mention of the role of parents and communities in reference to education. However, parent and community involvement in education is only alluded to six times (8%), with one of these instances blaming parents for not responding adequately to Uwezo's results and stating that citizens "read it [Uwezo's findings] and forget that such scary statistics were ever published." Parent and community engagement is raised primarily as a way to hold teachers, schools and the government accountable for subpar education quality, but there is no mention as to how this could be approached. There is one mention, in the article "Parental responsibility in education put to test" (Standard, 2011) which includes Uwezo's vision about "the need for parents to be involved in education success of their children." Additionally, of the 34 articles which include speech acts by Uwezo experts, only two promote any form of parental engagement: one asks for parents to discuss with teachers while the other advises parents to work with their children at home from an early age.
In addition to specific mentions of the importance of parental and community engagement, few articles present any information on Uwezo's philosophy. Only eight articles (10%) provide information more than that Uwezo is an NGO, lobby or assessment. Articles define Uwezo as a group that "monitors achievement in education" (Daily Nation, 2013), an initiative that "promotes access to information and improved service delivery outcomes" (Standard, 2012) or aims "to improve competencies in literacy and numeracy among children aged six to 16 in Kenya, Tanzania and Uganda" (Standard, 2014). Only one article, "Decline in literacy and numeracy skills in primary schools make learning a pain" (Daily Nation, 2015), presents Uwezo's philosophy, stating that "Uwezo is a five-year initiative that aims to improve competencies in literacy and numeracy among children aged between six and 16 years in Kenya, Uganda and Tanzania by using an innovative approach to social change that is citizen driven and accountable to the public." Overall, there is very little interaction with Uwezo's citizen engagement philosophy or its theory of change in the print media, ensuring that public discourse does not promote greater community responsibility or actions that citizens can take to improve the situation.
While Uwezo's results have been successfully disseminated to the public, Uwezo's portrayal in the print media presents little information or promotion of parent and community engagement. Although the print media is used primarily to disseminate the results of the assessment, Uwezo's growing importance as an expert organization in the education community provides it with a platform to promote its citizen-led, socially accountable philosophy and to speak directly to the need and importance of a more engaged citizenry. However, this paper demonstrates that the promotion of citizen action will not occur naturally in public discourse, even when the learning crisis is clearly articulated to the public. It must be noted that Uwezo as an organization has made a concerted effort to promote activities which engage parents and communities, moving from assessment to action (see PAL Network, 2017b), but public discourse has not referenced the need for local stakeholders to act.

Uwezo as a Leading Policy Thinktank
While Uwezo is predominantly described as an assessment, or a group which conducts tests of learning outcomes, its role as a national player in education policy ideas has grown significantly since its inception. The analysis included an examination of Uwezo speech acts. Voice is a fundamental component of analyzing public discourse as it determines who is given legitimacy concerning a specific topic and who has the power to provide information (Green Saraisky, 2016). Speech acts included direct quotes by Uwezo staff members as well as opinion pieces written by staff members as a representative of the organization. Figure 4 demonstrates the yearly growth in the number of Uwezo speech acts in articles from the dataset.

Figure 4
Articles with Uwezo speech acts, by year Exploring the nature of these speech acts demonstrates a shift in public discourse concerning Uwezo's role in the education sector. Earlier articles focused primarily on Uwezo's findings of low learning outcomes, in which speech acts by Uwezo representatives explain the reports and the results. However, as the spotlight on low learning outcomes decreases in discourse, Uwezo's speech acts grow. Uwezo representatives are tasked with providing insight on a range of education issues linked to quality of learning, but not directly to the assessment. For instance, since 2015, Uwezo representatives are looked upon to comment on reasons for low scores on national tests and how scores can be improved, the positive impact of competency-based learning, teacher training, safety in schools, and the growth of private schools. Overall, while Uwezo's assessment continues to garner attention, the organization's voice pushing for education policies has begun to dominate public discourse in recent years.

Discussion and Conclusion
This analysis of Kenya's two most widely read newspapers provides evidence of Uwezo's portrayal in the media and how its findings influence public discourse on education in the country. In addition, the analysis attempts to understand how Uwezo's philosophy of social accountability and community-led engagement is presented to the public through these newspapers. As a main disseminator of Uwezo's work, newspapers provide an opportunity and platform for Uwezo's assessment results to spark debate amongst citizens and to inform and invigorate citizens to get further engaged; an integral step in the information-accountability causal chain and in the variation and selection mechanisms of political evolution. By examining social accountability development projects and information-accountability frameworks, and by closely examining media as a tool to influence public discourse, I argue that discourse in Kenya concerning learning outcomes is largely connected to Uwezo's assessments, but that there is minimal information energizing citizens to act or promoting skills needed to take actions. These findings mirror the outcomes determined in Lieberman, Posner and Tsai (2014a) and demonstrate that citizen-led assessments should move further up the information-accountability causal chain in hopes of spurring greater citizen-led responsibility and action.
The descriptive aspect of this study finds that there is a sharp and sustained increase in public discourse surrounding Uwezo since 2012. Uwezo's findings are presented as a serious concern in public discourse, especially following the release of its reports, and opinions on education problems in Kenya use Uwezo's findings to demonstrate their arguments. In addition, Uwezo's assessments results and organizational expertise are used in a variety of school-based issues in the media, thus touching upon the variation political mechanism.
In analyzing the articles and headlines, I find that the shock of Uwezo's assessment results is decreasing and that there seems to be a sense of annoyance and complacency with the findings, but not a call to action. The perceived lack of shock found in later articles and the fact that Uwezo has conducted five assessments with very similar results demonstrates that the information is not new, which may be hindering citizen action at early step in the causal chain.
Public discourse focuses on comparisons, both within Kenya and cross-nationally, and places blame on various stakeholders, most notably teachers. While comparisons may create a sense of urgency to act, cross-national comparisons seem to downplay the troubling Kenyan findings, and a general lack of context for comparisons may promote Kenyan student success at the expense of neighboring countries. The blame game in public discourse is also not conducive for community engagement, especially considering the deflection of responsibility from citizens (and governments) and the lack of focus put on citizen-centered solutions to the many education issues brought up in the articles. While blame put on teachers and government may put pressure on these actors to improve, Uwezo's philosophy does not solely call for pressure tactics but for locally born solutions. I find that comparisons and blaming through public discourse on Uwezo is similar to understanding media reactions of international large-scale assessments, which focus on blame, comparisons and a lack of policy insights (see Grey & Morris, 2018;Pizmony-Levy, 2016;Wadlow et al., 2014). For instance, Steiner-Khamsi's (2003) typology for political reactions to international assessment results (scandalization, glorification, indifference) are present in this analysis, especially concerning scandalization.
While the anger and concern in public discourse stemming from the realization of subpar learning levels has led to public debate, its impact on community engagement still seems difficult to gauge. Lieberman, Posner and Tsai's (2014a) work proposed a more forceful and targeted use of information to lead to greater community engagement and ultimately improvements in education quality. Although this study does not examine how citizens react to articles about Uwezo or public opinion, my findings do reveal that public discourse in the media does not follow the informationaccountability causal chain or promote the selection of possible policy solutions. Building on Liberman, Posner and Tsai's (2014b) conclusions and recommendations, I believe that Uwezo's respected portrayal in the media and its growing role as a policy stakeholder provides the organization with an opportunity to use its prominent position in the media to further promote its social accountability philosophy. Uwezo's (2016) latest Kenya report states that "the release of this report therefore renews our call to all citizens to play a role in improving learning. If each of us did one small thing where we are, all children in Kenya could attend school and learn" (p. 5). While Uwezo has grown from an assessment organization to an education organization with greater policy and measurement ventures, there is still significant room to put its theory of change into action. It must continue to inform, but also provide guidance and a push to citizens to take responsibility and action, to ensure people have the knowledge, capacity and confidence to do one small thing to improve education in Kenya. education policy analysis archives editorial board