Teachers are the main tutors of academic merit, their evaluations are therefore essential to guarantee equal opportunities. Whether teachers are guided solely by the teaching skills and efforts of their students, or whether they also consider other factors, such as their sociodemographic characteristics, is a crucial question for identifying bias and potential discrimination in the education system.
Due to social desirability, it is not possible to detect these biases from traditional survey data; We need experimental projects, rare until now in analyzes of the Spanish educational field. Normally, to study assessment bias, students with similar performance on standardized tests (e.g. PISA) are compared to determine whether, with equal skills, they have a different grade or probability of repeating depending on their gender, their migratory origin or their level. . socio-economic status of the family.
This article brings together the results of an experiment designed by ourselves and which we recently published in Sociological sciences. The experiment is specifically designed to causally identify whether, beyond academic merit, the characteristics attributed to students (gender, migrant origin, social class) enter into teachers’ equation when assigning grades and grades. predicting the future educational paths of their students.
To design the experiment, we selected a representative sample of 20 education faculties from all Spanish (non-bilingual) education faculties, both public and private, and distributed a questionnaire to their enrollees. online. We focused on future teachers rather than practicing teachers, because we wanted to assess whether, even before entering the education system as professionals, young people reproduce (pre)existing group stereotypes in society and at the school itself. If applicable (spoiler: this is the case), interventions and curricular (re)designs on the cognitive and institutional processes that generate discrimination could be particularly effective during the teacher training period.
In the questionnaire online distributed to the future teachers, a sheet (Figure 1) was presented with the personal and academic data of a fictitious 6th grade student and, subsequently, they were asked to evaluate an essay written by him and evaluate its future school career. . The form, based on real files, provided neutral census information, common to all profiles (center, family address, nationality, date of birth) and variable information (blue parts of the form in Figure 1) on the socio-demographic characteristics of the student: gender, migrant origin (Spanish or Moroccan parents) and father’s profession, camouflaged in the contact details, as a marker of socio-economic status. Thus, we had files for Daniel García González, Lucía García González, Youssef Salhi, Salma Salhi and their parents’ contact emails were, in the first two cases, David.Garcia@Pintores-Express.es or David. Garcia@Notarios- Garcia.es and, in the second two, Mohamed.Salhi@Pintores.Express.es or Mohamed.Salhi@Notarios-Salhi.es. The form also provided information about the student’s previous academic performance, classroom behavior, and compliance with tasks and assignments. These behavioral factors are crucial, since teachers, explicitly or not, also take them into account in their evaluations.
Respondents were then presented with a short essay describing a landscape and the passage of the seasons. This test had two versions according to its objective quality based on the official grids of the 6th year linguistic subject (spelling, vocabulary, syntax) and validated by a pilot study with 250 Andalusian teachers: a good version (7 out of 10) and a Another version of the worst quality (5 out of 10). Additionally, we manipulated two other factors, independent of writing quality. On the one hand, we incorporated a phrase as a signal to inadvertently recall the socio-economic status of the father (notary or painter in the construction sector) and, on the other hand, we slipped in information about cultural capital of the family by referring to two leisure activities. directly in the text: watch the TV show “Temptation Island” or Monet’s impressionist paintings in a museum. Figure 2 presents the good and bad versions of the wording and, in color, the different experimental manipulations that were carried out.
Respondents were then asked to rate the quality of the writing and, in addition, to rate the likelihood of (a) this student repeating 6th grade and (b) reaching high school. Since good or bad writing was attributed in a completely random way, in addition to behavior and previous performance of the file, to the different sociodemographic profiles of the students, we can causally identify the weight that teachers attribute to objective academic ability or demonstrated (i.e., quality of writing, performance, and prior effort) in relation to assigned factors, i.e., sociodemographic factors over which the student has no control (sex biological, the origin of one’s parents or the socio-economic resources and cultural capital of one’s home).
We can highlight several main results of the experiment, which we illustrate in Figure 3. First of all the good news: ability (notably the objective quality of writing and behavior) is what weighs the most in the grade obtained in the task and also in the predictions which the respondents speak of the future academic success of the student. And then the bad news: In all cases, attributed factors creep in that, in principle, given equal academic merit, should not affect grades or expectations for future academic success. Girls receive slightly higher grades and are credited with more future success than merit factors would predict (there are several consistent explanations for this, which are explained in the article). Something similar happens with socio-economic status: the children of notaries receive better predictions about their educational future than the children of painters, with equal abilities and efforts. Additionally, writing an essay that shows “Temptation Island” instead of Monet’s paintings in a museum results in a lower grade, even if both essays are equally good with respect to the assessable objective linguistic criteria. The case of children of Moroccan origin is a little less intuitive. They get more marks than they deserve in the essay. This form of affirmative action may be because teachers attempt to compensate for the additional difficulty that a language skills task involves for students who do not normally speak that language at home. However, they are later assigned a lower probability than a student of Spanish origin of entering high school, which more closely coincides with the stubborn empirical reality.
In the scientific article, we discuss in detail the various theoretical explanations that may be behind these results (statistical discrimination, implicit bias, group stereotypes); We refer interested readers there. We would like to focus here on the substantive conclusion. Consciously or not, teachers take as merit certain characteristics that are not really related to them, such as the gender, social class or country of origin of the students’ family. When we talk about educational inequalities and the factors that explain the intergenerational reproduction of educational advantages and disadvantages, we are generally interested in the role of families and students. However, teachers are the primary judges of educational merit and directly determine who progresses and who does not progress through school. Explicitly or not, they send ongoing signals to students themselves and their families that influence the decisions made, sometimes in the form of self-fulfilling prophecies. In this study we show, with the most relevant evidence so far, that teachers indeed play a fundamental role as guarantors of equal opportunities. Although it is not clear what type of measures could be taken to strengthen this function, empirically establishing the existence of these biases is a first step in this direction.
