To align or not to align: the enactment of accountability and data-use in disadvantaged school contexts

Despite the growing number of researches about performance-based accountability (PBA) in education, there is still scarce evidence on the mediating role of subjective variables (e.g., perceived pressure and alignment to PBA mandates) in the enactment of PBA in socially disadvantaged contexts. This is paradoxical because marginalized schools are usually those that are on probation and have to cope with the threat of sanctions more frequently. Existing investigations on PBA enactment have put increasing attention to the role of situated and material contexts, but there is still limited knowledge on how subjective variables can mediate policy enactment processes and enable the adoption of different school responses. To address these gaps, the article aims to explore how the perceived accountability pressure, the school performative culture, and meaning-making processes at the school level are mediating the enactment of PBA policies in disadvantaged schools. At the theoretical level, the study is informed by sense-making and policy enactment frameworks. Methodologically speaking, the investigation uses a comparative case study approach based on two extreme cases, which have been selected on the basis of a factorial analysis that combines both survey and secondary data. The extreme cases represent two different scenarios, which, despite operating in similar situated contexts, are characterized by having opposite levels of perceived pressure and alignment with the performative culture. The case studies combine survey data (n = 39) with documentary analysis and semi-structured interviews with the management team and teachers (n = 7). The findings show that subjective variables, in interaction with other contextual factors, can exacerbate or inhibit PBA regulatory pressures and trigger diverging school responses.


Introduction
In recent years, performance-based accountability (PBA) policies have been disseminated globally as a "best practice" by different international organizations such as the World Bank and the Organization for Economic Co-operation and Development (Lingard et al. 2015;Verger and Parcerisa 2018). PBA instruments are being adopted in numerous educational settings around the world with the intention of modernizing governance, strengthening school effectiveness, and reducing learning gaps among different social groups. In Chile, PBA instruments were implemented in the context of a profound market-driven educational reform, introduced in the 1980s (Diaz Rios 2018;Gysling 2016;Verger et al. 2019) but currently PBA generates political consensus among right-wing and center-left parties (Author(s) 2017). Recent educational reforms adopted by both right-wing and center-left governments have only reinforced the PBA logic to educational governance and intensified the number of national assessments, the areas covered by these assessments and the consequences of assessment results for school actors (Carrasco et al. 2015;Falabella 2020). Nonetheless, despite the bipartisan support for PBA, important educational stakeholders are very critical of the current PBA approach and advocate the construction of a new assessment system that is more context-sensitive and formative in nature (Campos-Martínez and Fernández 2014; Flórez Petour and Rozas Assael 2019).
PBA policies follow a principal-agent logic and are not only expected to trigger school improvement through incentives and sanctions, but by promoting the use of learning-achievement data among school actors. Nonetheless, despite the expectations of PBA policies being high, especially in the context of poverty, where most underperforming schools tend to be found (Mintrop 2004), international evidence on the enactment of PBA policies and datafication shows a wide range of school responses (Falabella 2014;Koretz 2017;Landri 2018;Maroy and Pons 2019) that ranges from policy appropriation to rejection (Candido 2019) or opting out (Landri 2018). Existing research suggests that while some schools react to PBA by adopting substantive changes to pedagogical approaches and practices, others tend to adopt cosmetic or instrumental behaviors to improve their performance in national assessments (Mintrop andTrujillo 2007a, 2007b;Mittleman and Jennings 2018). For example, in Australia, Hardy (2015) identifies negotiations and ambivalent responses to PBA and datafication. Specifically, this investigation shows that while teachers are involved in data production, "they also seek to challenge the more reductive effects of such data generation" (Hardy 2015, p. 483).
In this sense, investigations carried out in the context of the USA indicate that educators made sense of PBA policies in complex ways, which may be complementary or even contradictory to the policy. These investigations show the relevance of cognition and meaning-making processes for understanding both school and individual responses and practices (Coburn 2005;Russell and Bray 2013;Spillane 2004). Principals' and teachers' perceptions in relation to PBA might be influenced by their local context, as well as "collective sense-making processes" (Diamond 2012, p. 173). In this regard, an investigation carried out by Mintrop (2004) in schools on probation shows that negative opinions and beliefs relating to the fairness of the accountability system, predominate markedly among teachers. At school and classroom level, PBA instruments can lead to both constructive (e.g., changes in the curriculum and teaching methods, see Koretz 2008;Hamilton et al. 2002) and perverse responses (Koretz 2008) such as cream-skimming (Jennings 2010), teaching to the test (Au 2007) curriculum narrowing (Berliner 2011) and cheating and exclusion (Hofflinger and von Hippel 2020). In turn, perverse responses can undermine educational inclusion and quality (Falabella 2014;Lipman and Haines 2007;Thiel et al. 2017). According to Diamond (2012), in the most challenging contexts, shortterm and instrumental strategies are more likely to be adopted than a substantive transformation of instruction. However, based on research carried out in England, Lupton (2004) points out a variation in school responses and approaches, even in schools that work in apparently similar poverty contexts.
In Latin American countries, there is also a growing corpus of investigations that explore the enactment of datafication and PBA in disadvantaged contexts. For example, in the case of Brazil, Candido (2019) found that datafication has penetrated "the imaginaries of school actors" and that school actors' engagement with PBA is higher and becomes more meaningful "when connected to problems individuals aim to solve" (p.14). Nevertheless, this investigation also illustrates the contentious nature of policy enactment and shows how school principals and teachers negotiate, decouple and even resist PBA policies. In Chilean disadvantaged schools, Campos-Martínez and Guerrero (2016) note that PBA policies contribute to increasing stress and trigger feelings of frustration among teachers. In terms of schools' logics of action, Falabella and Opazo (2014) identified four different school responses that include expressive, tactical, and instrumental strategies. Not surprisingly, school staff from "on probation" schools set against a background of poverty demonstrate higher levels of pressure and stress and are more likely to adopt short-term strategies such as teaching to the test. Similarly, in an ethnographic investigation carried out in a marginalized school context, Armijo (2019) shows that even in schools with alternative pedagogical projects, PBA pressure alters school practices and promotes the emergence of tensions between standardization and a school's ethical and pedagogical principles of inclusion.
The present investigation is situated in the social and academic debate that PBA generates in Chile (see Acuña et al. 2014;Assaél et al. 2014;Falabella 2016Falabella , 2020Montecinos et al. 2014;) and aims to provide new knowledge about how Chilean school actors understand and respond to accountability policies, with a focus on contexts of disadvantages. Specifically, the objective of this study is twofold. It aims to understand, firstly, how principals and teachers interpret and experience the PBA system and secondly, how these interpretations, together with other variables of a subjective nature, mediate the enactment of accountability policies and the use of performance data among school actors.
To achieve these objectives at a theoretical level, this research draws on sensemaking (Spillane 2004;Coburn 2005) and enactment theories (Ball et al. 2012), and, at a methodological level, the research adopts a mixed-methods design approach through which two extreme school cases (i.e., schools that perceive very different levels of performative pressure and have very different perceptions about PBA) are compared. The article is structured as follows. Firstly, I present the theoretical framework. Secondly, I describe the main features of the current PBA system in Chile. Thirdly, I present and justify both the methods and the sample of schools. Fourthly, the two school cases are presented, and finally, I discuss the main results.
2 Theorizing schools' responses to high-stakes accountability: Meaning-making and policy translations in context As shown above, the PBA policies can trigger a wide range of responses and effects. To capture schools' responses to PBA, and to what extent schools align or not to PBA expectations, this article combines the policy enactment framework with sense-making theory. The framework of policy enactment has been developed as a critique of conventional approaches to implementation research (Ball et al. 2011(Ball et al. , 2012. According to these scholars, policy implementation literature does not tend to pay enough attention to the mediating role of local contingencies and the agency of policy implementers. Approaches that draw from a school effectiveness tradition tend to conceive schools as dematerialized contexts and assume that implementation is a linear process where inputs are too directly translated into outcomes (Ball et al. 2012;Thrupp and Easter 2012). More recently, some investigations have pointed out the limitations of these linear and oversimplified perspectives, recognizing the complexity of the policy process (Eppel et al. 2011). For these scholars, the concept of enactment is more useful as a heuristic tool, to investigate how education policies are implemented in real school settings (Braun et al. 2011a, b). Sense-making theory makes a similar point to enactment theory by stating that during policy implementation "actors are making meaning, being influential, contesting, constructing responses, dealing with contradictions, attempting representations of policy" (Ball 1994, p.21).
One of the most prominent contributions of this framework refers to the analytical distinction between two interrelated stages or moments: interpretation and policy translation. Interpretation refers to meaning-making processes and the decodification and analysis of external policy mandates and messages (Ball et al. 2012;Coburn 2005;Spillane 2004), whereas policy translation refers to the moment at which "schools transform their institutional agenda into specific practices, concepts, and materials as a means of responding to the policy" (Zancajo 2020, p.49). In other words, policy translation refers to those processes that entail the local adaptation or the modification of a given policy (Steiner-Khamsi 2014). Policy translation "can be selective, picking only those components of a policy that are deemed compatible with the local environment, beneficial in the light of local conditions, or sellable to the local community" (Schulte 2018, p. 629).
Finally, policy enactment is a process that is contingent on different contextual circumstances. These include (a) situated contexts, which refers mainly to the local history of the school and to the socio-economic characteristics of the school population; (b) professional contexts, including the ethos, values , and beliefs of the management team and the teaching staff, together with the leadership style; (c) the material contexts, referring to economic resources, staffing, school facilities, and technological resources; and (d) external contexts related to expectations, support, and pressures that come from the account holder or from the policy context at the local or state level (Braun et al. 2011b). As noted by Braun et al. (2011b), Candido (2019), Carrasco and Fromm (2016), Keddie (2014), and Thrupp and Lupton (2006), contextual circumstances of a different nature are crucial in explaining differences in the way schools put new policy mandates into practice. School contexts can influence not only the degree of policy appropriation but can also trigger differences in terms of strategies, practices, and policy outcomes.
The market reform introduced key elements that would contribute to the development and consolidation of an "Evaluative state" (Neave 1998) designed to govern "at a distance" (Rose and Miller 2008, p. 205) a decentralized network of autonomous educational providers, both public and private (Parcerisa and Falabella 2017). The reform expected that both administrative and market accountability pressures would favor competition between schools, which in turn would improve educational quality (see Chubb and Moe 1990). Nevertheless, the implementation of market reforms triggered multiple unexpected alterations such as high levels of school segregation and undesired behavior among schools (Corvalán et al. 2016;Hofflinger and von Hippel 2020;Valenzuela et al. 2014;Verger et al. 2016a).
Far from reversing the market model during the 1990s, once democracy was reinstituted, the center-left coalition maintained the model's core components and enacted a co-payment decree (Law N°19,247), through which the government made it possible for privatesubsidized schools to charge fees and make a profit (Carrasco et al. 2017;Villalobos and Quaresma 2015). Additionally, during the first mandate of the center-left coalition, the governmental action focused on increasing fiscal expenditure and improving educational supplies (e.g., teaching materials, pedagogical support) and teachers' working conditions (Burton 2012), especially among those schools receiving the most vulnerable students. In this period, new material and reputational consequences were added to the SIMCE test. Firstly, in 1995, SIMCE's results were disseminated for the first time with the aim of informing parents' choice and intensifying market pressures. Secondly, in the same year, the National System of School Performance Evaluation was enacted, and, as a result, SIMCE became a fundamental part of a merit-based pay system for teachers, in which collective salary bonuses became attached to school performance (Mizala and Schneider 2014).
During the early 2000s, after a decade of steady growth in public spending on education, policymakers identified certain areas of stagnation in student learning outcomes (Bellei and Vanni 2015). For this reason, the government implemented various compensatory programs directed at low-performing schools to promote data use and school improvement (Falabella 2020). In 2008, this policy model crystallized in the Preferential School Subsidy Law (Weinstein and Villalobos 2016), through which the state gives schools an additional subsidy for each student that they enroll from a disadvantaged context (Valenzuela and Montecinos 2017). Nevertheless, to access this subsidy, schools are classified according to their performance in standardized tests and must be even more accountable to the state. In cases of poor performance, the state can impose sanctions, which include the possibility of school closure (Elacqua et al. 2019). Subsequently, the evaluative state was improved through the adoption of new data-intensive policy instruments and tools. Specifically, in 2011, the right-wing government enacted the Education Quality Assurance System (Law n°20,529), which introduced a new institutional matrix through which the state inspects, evaluates, and, if necessary, sanctions both public-and private-subsidized schools (Parcerisa and Falabella 2017).
In recent years, large-scale assessments and accountability have been hot topics in both political and educational debates in Chile. Since 2000, the Ministry of Education has mandated two different "SIMCE Commissions" (in 2003 and 2015) with the purpose of assessing its design and effects (Falabella 2020). In addition, together with student protests, a new social movement called Stop SIMCE has emerged to oppose standardization and test-based accountability policies (Campos-Martínez and Guerrero 2016; Pino et al. 2016). In short, Chile has a long trajectory of accountability reforms linked to standardized tests and can be considered a unique case to study the combined effects of market and administrative pressures in the education sector.

Methods
Methodologically, this article follows a comparative case study approach (Yin 2003) based on the comparison of two extreme cases (cf. Danermark et al. 2002;Gerring 2007), which have been selected on the basis of a quantitative analysis that combines administrative and survey data. 2

Sample, data, and units of analysis
In the quantitative stage of the research, two different surveys were administered to principals and teachers of a representative sample of 79 elementary schools (ISCED 1 and ISCED 2) located in the three largest urban areas of Chile: the metropolitan regions of Santiago, Valparaíso, and Concepción. The sampling strategy used to choose the schools was based on the systematic probability to size sampling method and combined explicit and implicit stratification criteria (Ferrer-Esteban 2021, forthcoming).
The principals' survey (n = 200) was administered to the principal, the pedagogical coordinator (also known as Jefe de UTP, in Chile), and the general inspector in each school. The teachers' survey (n = 1130) was administered to 20 teachers in each school (10 teachers who at the time of implementing the survey taught grades and subjects evaluated in the SIMCE and 10 teachers who taught other grades and/or subjects 3 ). The questionnaires 4 included different modules related to personal and school characteristics, school demand, 2 The data analyzed in this research comes from the REFORMED project This is an international project that analyzes the adoption and enactment of school autonomy with accountability policies in countries like Norway, the Netherlands, Spain and Chile. For more information about the project, see www. reformedproject.eu 3 The following grades (4th, 6th and 8th) and subjects (language, mathematics, natural sciences, social sciences) are assessed in the census-based SIMCE test. For more information, see Ministerio de Educación (2016) Plan de evaluaciones nacionales e internacionales 2016-2020. Santiago: MINEDUC. Available at: https://curriculumnacional.mineduc.cl/614/articles-34980_recurso_1.pdf 4 Further information about the rationale and the content of the questionnaire can be found in the questionnaire methodological note: Levatino, A. (2021, forthcoming). Surveying Principals and Teachers in the framework of the REFORMED Project: Methodological Insights into the Design of the Questionnaires. REFORMED Methodological Papers No. 2. www.reformedproject.eu perception of market and administrative pressure, opinions, perceptions, and beliefs about PBA, data use, commercialization, job satisfaction, etc. (Levatino 2021, forthcoming). The questionnaires were administered online through the specialized software Qualtrics. During the implementation process, a researcher supervised and supported the participants.
The sample for the qualitative stage used factorial analysis to select schools with different levels of perceived accountability pressure and alignment to PBA (see Verger et al. 2021, forthcoming). The factorial analysis combined data from the survey (accountability pressure index) with secondary data from the school vulnerability index database (MINEDUC 2019). Additionally, we also used the database of schools' performance categories elaborated by the Education Quality Agency (Agencia de la Calidad 2019). In total, we obtained four quadrants from the factorial analysis, which were used to select the primary education schools in which to carry out qualitative fieldwork. The two qualitative case studies were conducted in urban schools located in popular municipalities of the Metropolitan Region of Santiago.
For the purpose of this article, I rely on data from two extreme cases selected on the basis of the factorial analysis. 5 According to Gerring (2007), p. 101), "extreme-case method selects a case because of its extreme value on an independent or dependent variable of interest." In this regard, Danermark et al. (2002) point out that researching extreme and/or pathological cases are an appropriate strategy to explore mechanisms that are normally hidden or difficult to observe in most common cases and situations. As can be seen in Table 1, although both cases have a disadvantaged socio-economic composition (quartile 2, see column "socioeconomic status"), they differ in terms of their perception of pressure as well as their alignment regarding performative instruments and goals. Thus, the two schools are located in scenarios that are opposite in terms of pressure and ideational disposition towards PBA (low alignment with PBA/low pressure and high alignment with PBA/high pressure). School A shows a low alignment with PBA and, despite being categorized as "low-performing", perceives low levels of accountability pressure. In contrast, School B has a medium performance but perceives high accountability pressure and shows high alignment with PBA instruments and goals.
Following the selection criteria mentioned above, in each case, surveys were conducted with the management team (school A, n = 3 and school B, n = 3) and the teachers (school A, n = 13 6 and School B, n = 20). In total, the qualitative fieldwork involved two semi-structured interviews with purposely selected teachers 7 and one principal from each school (and in the case of School B there was also an interview with the pedagogical coordinator). The length of each interview was between 30 and 90 min. Complementarily, we selected key documents from each school (the institutional project and the reports from the Quality Education Agency). 0 Further information about the rationale and the content of the questionnaire can be found in the questionnaire methodological note: Levatino, A. (2021, forthcoming). Surveying Principals and Teachers in the framework of the REFORMED Project: Methodological Insights into the Design of the Questionnaires. REFORMED Methodological Papers No. 2. www.reformedproject.eu 5 There is a detailed explanation of how quadrants were obtained in Ferrer-Esteban, G. (2021, forthcoming). Sampling strategy. REFORMED Methodological Papers No. 3. Bellaterra: Universitat Autònoma de Barcelona. Available at: www.reformedproject.eu. For further information, see Verger et al. (2021, forthcoming) 6 Since School A has fewer than 20 teachers, the sample of participants included all members of the teaching staff 7 Teachers who taught in grades and subjects evaluated in the SIMCE test, and were available at the time of the visit to the school, were selected for the qualitative fieldwork

Data analysis
To analyze the quantitative data, a descriptive analysis of the responses from each school was used. The quantitative analysis enables us to illustrate the context and the general patterns of response identified in each school. Regarding the qualitative data together with the institutional documents, all the interviews were transcribed and analyzed with the ATLAS.ti version 8 software. To examine the qualitative data, qualitative content analysis was carried out (Mayring 2004). Concretely, a codebook was developed, and all data were coded by combining pre-established (theory-driven) and emerging codes that covered a wide range of key themes, such as school contexts and culture, the perception of pressure, the interpretation, and translation of PBA policies, data use, etc.

School A: low reactivity to PBA regulations
School A is a public school with more than 50 years of history. In terms of situated context, it has an approximate enrollment of 250 students and a vulnerability index of about 90% and welcomes students who have been expelled from other schools from the neighborhood. The school's material context is characterized by a small and very poor infrastructure. Regarding the external context, the survey results show that the reputation of the school is slightly below average and school managers describe being pressured to maintain or increase enrollment (6.3 points out of 7). In terms of performance, the school has been classified as low performing for two consecutive years. Finally, concerning the professional context, the survey data show higher levels of cooperation between teachers (see Table 2, in the appendix) and trust (see Table 3) in the headteacher than observed in School B. Its pedagogical project emphasizes safety and the promotion of community participation. The school's ethos is characterized by the promotion of care and respect for the environment, human rights and dignity, and interculturality.

Principled resistance to PBA and standardized tests
In School A, teachers and principals predominantly consider the PBA system to be unfair (see Table 4) and question its validity and usefulness (see Table 5). Specifically, Likewise, teachers in School A are critical of the market use of SIMCE results (i.e., publication of school scores to promote school choice). Specifically, 70% of the participants express negative opinions regarding this item. When asked about the reasons for this negative opinion, a language teacher, Violeta, points out that the SIMCE test is a "discriminatory" instrument. By publishing results based on a limited set of indicators, the test contributes to the stigmatization of low-performing schools from marginalized contexts.
(Interviewer): What do you think about the publication of the SIMCE's results or the classification [of schools]? (Violeta): I don't like it, I don't agree with it because, for example, it puts this school in a disadvantaged position because parents see the SIMCE results to enroll their children. Other topics such as those that I told you about [emotional support to the students] are not published. [...]Then, of course, in that sense it puts us at a disadvantage because they say: "oh look at the SIMCE's results". Then there is a kind of discrimination and ignorance of what is the vision or mission of the school, what is the [singular] characteristic that we have, which is quite particular in relation to the other 17 schools in the municipality. (Violeta, language teacher, School A.) 8 The names of interviewees are preserved for privacy purposes. For this reason, pseudonyms are used to maintain the anonymity of the participants. The research follows the European code of conduct for research integrity and complies with the ethics review procedures from the European Research Council. 9 Also known as profesor jefe, in Spanish. In Chile, the lead teacher is in charge of a grade. The lead teacher can teach various subjects beyond their own specialism. In the case of Joaquín, he taught languages and mathematics in a grade evaluated in the SIMCE test.
In other words, according to Violeta, by putting the main focus on cognitive skills, the publication of SIMCE results does not sufficiently capture the value of a school's pedagogical projects such as in School A, whose main focus is to prevent early school leaving in a context where low levels of student engagement are observed.
Similarly, teachers in School A are critical of the validity and usefulness of the SIMCE test. Many of the teachers (80%) consider that the SIMCE test does not provide useful information about student learning and that the results of SIMCE do not adequately represent what students have learned and what they can do. Violeta points out that since she does not believe in the validity and usefulness of the SIMCE test, she does not see it as very relevant. To her, a qualitative assessment system would be more meaningful to guide the teaching practice. Teachers disagree (50% disagree, 50% strongly disagree) with the statement that a good teacher can be recognized by his/ her students' results in SIMCE. Similarly, high levels of disagreement are observed regarding the statement that the content of SIMCE shows what the school priorities are (40% disagree, 30% strongly disagree). In this regard, the headteacher expresses doubts about the compatibility between standardized tests and the mission and vision of the school. To him, the test is the same for all the schools, while students' access to knowledge and experiences are very different (Víctor, headteacher, School A).
Another reason that School A teachers dislike SIMCE is that they have the feeling that SIMCE data makes teachers compete with each other (Violeta, language teacher, School A). According to 40% of the respondents, SIMCE data generates competition among teachers (see Table 11). Nonetheless, as mentioned above, despite working in an "on probation" and underperforming school, teachers and managers in School A experience low levels of performance pressure coming from the public authorities. This perception of low pressure is explained by the fact that the school staff, at both the management and teaching levels, have opted out of participating in the "competition game" promoted by the PBA system due to both professional and situated contexts. To them, "school improvement" is conceived as a long-term process in which SIMCE data are not their primary proxy for success.

Data use in context: Coping with cultural and material contingencies
In School A, half of the participants use SIMCE data to identify the needs of students (50%), to report teachers' results among the teaching staff (50%), or to inform parents about the school results (50%). A relatively small percentage of teachers also use SIMCE to make decisions about professional development activities for teachers (10%) (see Table 6). Nonetheless, it is in the interviews with both the headteacher and teachers that it becomes clearer that there is a rather superficial approach to SIMCE data in School A.
Here at the school everyone, from the management team to the last teacher, got together to [look at SIMCE data]. Not with the importance that we should give to it, in the sense that it is not, […], what the SIMCE score shows is not relevant for us. I insist, I see it from the point of view that this test is misdirected, to me to say: "Do you know that a school in Las Condes, Providencia, Vitacura [which are rich areas in Santiago] obtained 10 points more than us?", [ ...] that is, they are lightyears away from [us] in comparison. (Joaquín, lead teacher, school A.) Joaquín adds that he does not use SIMCE data in his daily teaching practices, because he finds the information that he receives from the management team, his colleagues, and his own experience much more useful and meaningful: I believe more in the management team or in the same advice from my colleagues and the experience that we have with children every day. I think that for me that is much more important than external information, [...] (Joaquín, lead teacher, School A.) Regarding the ability to use data, 60% of the respondents point out that they have little ability to do so, while 40% of the participants point out that they have some or full ability to use data (see Table 10). In addition, almost 60% of the participants in School A consider that they do not have sufficient time to analyze SIMCE data (see Table 6). However, once again, they emphasize that if they do not do so it is because they have other priorities. To them, school improvement aspirations are still too much conditioned by the material and situated reality in which their school is positioned: I think that much more could be done concerning what SIMCE is looking for, which is to have excellent scores, but beyond SIMCE, we could have better results. Because for example, here we do not have access to a projector, for example, we have a room where the sun comes in, that I cannot project [the slides] well, that I do not have a blackboard, rooms where it rains. (Violeta, language teacher, School A.) The high proportion of student transfers between schools that prevails in contexts of disadvantage is due to precarious labor and housing conditions and is another factor that challenges the fulfillment of PBA expectations for data use:

Test preparation as a symbolic practice
Test preparation is a common practice among schools that want to improve their scores in national assessments. Nonetheless, in School A, 60% of the participants (see Table 7) indicate that the principal or the management team has never recommended or instructed that teaching should be more directed towards the achievement of the learning standards that are evaluated. The same percentage of teachers indicates that they never carry out "teaching to the test" practices (see Table 8 In short, School A responds superficially to PBA pressures. Despite the fact that this school is on probation, the level of data use and teaching to the test is very low. This type of response is explained, in part, by the school's disadvantaged position in the education market, but also by a prevailing organizational culture that is critical of and resistant to PBA policies and standardized tests. Thus, the case of School A challenges some of the premises of the high-stakes accountability theory of action (see Mintrop 2004), according to which threats and sanctions would contribute to change school actors' behavior and promote performative action.

School B: Aligning to PBA expectations with fidelity
School B is a private subsidized school with nearly 30 years history. In terms of situated context, it has more than 800 students and a vulnerability index of about 90%, although according to the principal, the pupils have high levels of behavioral and cognitive engagement. Thus, despite having similar rates of vulnerability to School A, the students' disposition towards the school is completely different. Regarding the material context, the school is more spacious and the infrastructure more modern than in School A. The headteacher's office is large and well-equipped and has a mirror that generates a panopticon effect. With regard to the external context, this school is categorized as a medium-performance school and has a reputation slightly above the average (only 17.6% of the staff consider it below the average). In relation to the professional context, survey data show lower levels of cooperation between teachers (see Table 2) and lower levels of trust (see Table 3) in the headteacher than those observed in School A. This could be explained partly by the employment of a more hierarchical and managerial leadership style as well as by the performative pressure that permeates the school's organizational culture. As previous investigations have noted, in (Verger et al. 2021, forthcoming;Zancajo 2017). It is important to note that the owner of School B is also the proprietor of other schools. In this regard, the headteacher pointed out that the schools belonging to the same owner compete with each other for performance results. One of the key points in the institutional project is the promotion of students' social mobility. In the case of School B, the school's ethos is much more oriented towards learning metrics, performativity, and datafication. Thus, its main educational theme refers to promoting social mobility in poverty contexts. The school's mission and vision highlight the focus on the achievement of quality learning for all, the evaluation of learning outcomes, and educational inclusion.

Trust in numbers: a game of love and hate towards PBA instruments
In contrast to School A, in School B I found more positive opinions, perceptions and beliefs regarding the fairness, validity and usefulness of the PBA system (see Tables 4  and 5). In School B, the use of standardized tests for market accountability purposes has a higher acceptance among school actors than in School A (see Table 4). Half of the respondents affirm that it is fair to publicly disseminate SIMCE's results through the media and/or Internet. Following this line of thought, the pedagogical coordinator of the school highlights the importance of publishing school scores to facilitate parental choice. Critical views of some aspects of PBA are also present in this school, especially with the threat of sanctions for low-performing schools that the PBA system in Chile is in place. According to María, the headteacher of School B, punitive uses of test scores distort the potential benefits of evaluation and might contribute to exacerbating inequalities between schools, among other undesired effects.
But in others, when it [the standardized test] is used with another purpose, and with a more punitive purpose, which is what has happened in Chile that has to do with the issue of mandating in a law that if you lower this or that categorization, your school is closed, I think that it loses the focus, the end, of that validated, standardized test and ... obviously, having a function that can be noble to improve or equalize the opportunities for the different schools, on the contrary yields more tribute to this great gap that exists at least in our country, and basically, [...] it becomes a threat and from there generates fear in [school] communities, stress, pressure, and [the focus on learning] is lost within the classroom […] and many times the students begin to be trained on how they should respond to that test. (María, headteacher, School B.) In spite of these criticisms, both the organizational and pedagogical practices in School B are clearly aligned with the PBA mandate. To a large extent, this is related to the fact that, unlike what we observed in School A, teachers in School B consider the SIMCE test to be both valid and useful. For example, almost half of the participants disagree with the statement that the results of SIMCE do not provide useful information regarding students' learning. Also, in this school, the percentage of teachers who agree with the statement that the results of SIMCE do not adequately represent what students learn is 30 points lower than in School A. Overall, what stands out in this school is that both managers and teachers emphasize the value of SIMCE data for decision-making at the school level: (Interviewer): What do you think about standardized tests? (Marcelo): I think they are a very good element to measure certain issues, I think it is essential that an educational system has this type of instrument because it requires a lot of information that schools could not access to otherwise. (Marcelo, pedagogical coordinator, School B.) Despite the fact that School B's performance is good, both its management team and its teachers experience high levels of performance pressure. The headteacher considers that this pressure is related to the importance of maintaining the good results, but also to her improvement ambitions. In her own words, the school needs to "do well in terms of maintaining the category" and "from there go up" (María, headteacher, School B). Manuel, who is a teacher in mathematics, states that he does not feel very much pressured by the management team, but by the Ministry of Education and by the school owner. Carlos, who is a language teacher, considers that performance pressure is experienced "on a yearly basis" and that this pressure is especially felt by math and language teachers.

An expansive and intensive data use
School B teachers indicate that they use SIMCE data to define and monitor the school improvement plan (93.75%), to make decisions about professional development activities for teachers (62.5%), to adjust the curriculum according to the results (56.25%), to inform parents about the school results (56.25%), and to compare their school performance with that of other schools (56.25%) (see Table 6). Carlos considers SIMCE as a complementary instrument, which is helpful for monitoring the work of the school and making pedagogical decisions: (Interviewer): And do you find the SIMCE results useful for decision-making at school? (Carlos): I think that every result is useful, but it will depend a lot on where you put the focus. I think knowing the result of your school in relation to the different evaluations that were applied is essential to be able to take actions, create a work plan because if [you don't have data], you work on the basis of anything, that is, what [I don't like from] SIMCE is this massive [classification of schools] that is done. (Carlos, language teacher, School B.) In School B, 43.75% of the respondents declare that they are reasonably capable or completely capable of using learning achievement data (see Table 10), even though, similar to School A, about 75% of the respondents consider that they do not have sufficient time to analyze the data (see Table 6). To address this challenge, School B has contracted an external test service that provides them with disaggregated data on student performance at both classroom and individual level. According to the management team and the teachers interviewed, they find these data much richer and more useful than SIMCE data for making pedagogical decisions, as well as to monitor the school improvement plan. Regarding the performative effects of the SIMCE test in School B, a higher level of competition among teachers is observed (see Table 11). In School B, half of the respondents declare that there is a lot of competition (31.25%) or complete competition (18.75%) between teachers because of the test, while 43.75% of teachers state that there is some level of competition between them. This quote from Manuel illustrates the performative pressure that teachers in School B experience due to the PBA system. (Manuel): I believe that pressure does not generate an anxiety that is going to kill me, but there is pressure, there is always pressure, that is what SIMCE generates for me. There is pressure [...] also, because as I was saying before, it is not funny to see the test... that your boss may have confidence in the work that one is doing, but it is not nice to see that the students are at an elementary, adequate and insufficient level. It is not something that is seen with good eyes, of course, there are economic, work and labor consequences, but there are also consequences in the mindset that one would like to have for these things. (Manuel, maths teacher, School B.)

Routinized practices of test preparation
At School B, instrumental practices to boost performance on standardized tests are prevalent. In this school, a high routinization of test preparation practices is observed. For example, 87.5% of the respondents indicate that the principal or the management team has recommended (56.25%) or instructed (31.25%) that teaching should be more directed towards the achievement of the measurable learning standards (see Table 7). Significantly, only 6.25% of respondents state that the headteacher and management team have never recommended or insisted that students should practice for SIMCE. In this regard, 75% of the respondents declare that they have carried out teaching to the test practices. More specifically, 75% of the participants indicate that they carry out this practice during the whole year (50%) or the month before the test (25%) (see Table 8), which shows that these instrumental practices to boost school scores are, to some extent, institutionalized and embedded within the school culture.
I am sure that, if we do a series of practices that are already installed, and that we have tried to consolidate and strengthen, and that we do them as part of our daily routine in a natural way, on a day-to-day basis, etc., the results should be maintained or should improve. (María, headteacher, School B.) With Concerning the influence on pedagogy and curricular narrowing (Table 9), more than half of the respondents declared that SIMCE has led to some (43.75%) or much (25%) redistribution of resources in favor of the subject areas and competences that are tested. Also, 68.75% of the respondents point out that the existence of centralized learning standards has influenced the pedagogical approach of the school, and 56.25% of them declare that SIMCE is considered when making decisions about curricular content. When asked for an example of how test results have helped with school decisions, the pedagogical coordinator of School B states that: Look, the evaluations, for instance, we make decisions regarding curricular coverage, how it is being met or not. So, the curricular coverage that we are usually measuring, we accompanied it with the fact that when that objective was seen, it was seen in this way. (Marcelo, pedagogical coordinator, School B.) In summary, School B follows the PBA mandate almost to the letter. This school shows a very intensive and routinized use of achievement data and resorts to commercial education services to monitor student learning. Despite this, they are aware and critical of some of the uses of PBA policies. In this school, the combination of an organizational culture that is aligned with learning achievement goals and a high perception of accountability pressure (which also comes from the school owner) translates into the intensification of instrumental practices such as teaching to the test and narrowing the curriculum. Nevertheless, this case also shows that a priori instrumental practices can be re-created and translated by teachers who adhere to more expressive logics of action.

Conclusions
In Chile, recent educational reforms have emphasized accountability and datafication as core policy instruments to promote school improvement. The theory of action of PBA assumes that schools will react to performance pressure and will use test data to promote instructional improvement measures and to address learning inequalities between different social groups. Nonetheless, by adopting a policy enactment perspective, this study shows that PBA policies might be experienced and lived very differently by school actors and that schools might respond to PBA pressures in ways that deviate significantly from policy expectations. Even schools that apparently operate in similar contexts activate and rework PBA instruments in diverging ways. In this final section, I would like to highlight three key, interrelated conclusions that derive from this study.
Firstly, the study contributes to gaining a better understanding of under what circumstances and in which way PBA policies work in disadvantaged contexts. The overall findings presented in this article show that schools with similar situated contexts can interpret and translate PBA policies in diverging ways (see Lupton 2004;Lupton and Thrupp 2013;Thrupp and Lupton 2006). Actors' cognition and policy perceptions interact and are co-produced with material, cultural, social and institutional factors. By carrying out an in-depth analysis of two schools that operate within a similar disadvantaged context, this article demonstrates the importance of paying more attention to variables of a subjective nature, to better understand school responses to PBA policies. Specifically, the research shows that both the perceived accountability pressure and alignment with performative culture among school actors play a key mediating role in the enactment of PBA policies.
Secondly, the study shows that there is not necessarily a correspondence between the existence of an external threat of sanctions and the levels of perceived pressure experienced by school actors. Despite the fact that School A is "on probation" and could be closed down due to continuous poor performance, the school staff does not experience pressure and adopts very superficial-even passive-responses to accountability regulations (including low levels of data use and test preparation). This passive response could be explained by three interrelated factors, namely, the lack of trust in public authorities and the PBA system (their capacity to impose sanctions), the conflicting relationship between PBA mandates and the school staff's principles, as well as the school ethos, which interact with socio-economic and material variables. On the contrary, although School B obtains good results in the standardized test, both the teaching staff and the school leaders experience much higher levels of pressure than initially expected and overreact to accountability pressures by intensifying test simulations, resorting to external services to manage additional performance data, and so on. These findings reinforce that, to understand the effects of regulatory pressures on educational actors, it is more important to take note of how these actors experience and perceive pressure rather than considering the "objective" pressure that, according to formal regulations, actors are expected to experience (see Verger et al. 2021).
Thirdly, the study challenges some of the core premises of the action theory of PBA. PBA assumes that, especially in the context of high-stake accountability systems, the existence of external pressures will increase the motivation of headteachers and teachers working in low-performing schools and will promote reflexivity for instructional improvement among them. It is usually assumed that the motivational effects, triggered by PBA pressures will favor organizational and instructional changes that, in turn, will increase the educational quality of the school. However, the case of School A shows that negative interpretations and considerations of the policy translate into superficial policy implementation.
Overall, when educational actors have a negative opinion of a policy, they will tend to dilute this policy, and/or reject or avoid implementing it (Achinstein and Ogawa 2006). These dynamics are reinforced when negative interpretations accompany low levels of perceived pressure to conform with the policy in question. This is the case in School A, where, as just shown, negative perceptions with PBA predominate, at the same time as the PBA system generates a low level of performance pressure. Accordingly, as I have shown in this investigation, School A does the minimum required to comply with the PBA mandate. As Candido (2019) points out, school actors engage with data-based governance instruments to the extent that they perceive them as significant in solving existing perceived problems and improving their teaching practices. Thus, the fact that a policy is regulated and mandated is not a sufficient condition to promote educational change (see Schulte 2018). Policy instruments such as PBA need legitimacy and trust in principal, to be appropriated by school actors and enacted in constructive ways. If school actors do not believe in the validity or the fairness of the policy in question, in this case PBA, they will avoid implementing it or will do the minimum to comply with regulatory requirements. Likewise, in contexts where high perceived pressure is combined with a favorable disposition towards PBA instruments and school competition, as in the case of School B, instrumental behaviors (such as teaching to the test or narrowing the curriculum) might overtake educational practices. In this case, PBA expectations of school improvement could be undermined because an excessive emphasis on test preparation could generate impoverishment of both teaching and learning experiences. This example also shows that certain contexts encourage feelings of stress and instrumental practices (e.g., teaching to the test) to become internalized and made invisible to school actors (see also Falabella 2020;Waldow 2009).
To conclude, and in terms of policy implications, the study suggests the need to design more context-sensitive and robust assessment technologies and related accountability policies. Future research should explore which accountability policy designs can contribute to the adoption of substantive instructional and organizational changes, in order to improve students' learning experiences and teaching methods.