Back to basics: The problem of identifying arguments and a potential solution

Volviendo a las bases: el problema de la identificación de argumentos y una potencial solución

Abstract

Argumentation is crucial in science education to fostering critical thinking and decision-making skills. However, despite extensive research aimed at improving students’ argumentative skills, we propose that the problem begins earlier; many individuals struggle to recognize an argument when they encounter one. To test this theory, we first conducted a survey with 1,217 participants aged 16 to 70, which was then replicated with 47 fourth-year secondary school students. Both surveys revealed similar difficulties in distinguishing arguments from non-arguments when the conclusion is false, and in identifying argumentative uses of signal words. To help students overcome these obstacles, a didactic intervention was designed. Following the intervention, students showed significant improvement in identifying all types of arguments, as well as demonstrating greater precision and recall. These findings emphasize the efficacy of targeted interventions in addressing deficiencies in argument recognition.

Keywords:
secondary education; argumentation; critical thinking; pedagogical intervention.

Resumen

La argumentación es fundamental en la enseñanza de las ciencias para fomentar el pensamiento crítico y la capacidad de tomar decisiones. Sin embargo, aunque muchas investigaciones están orientadas a mejorar las habilidades argumentativas de los estudiantes, proponemos que el problema se encuentra antes: existen serias dificultades para reconocer si se está en presencia de un argumento. Se realizó una encuesta entre 1217 participantes de entre 16 y 70 años que se replicó con 47 estudiantes de cuarto año de secundaria. Los resultados de ambos relevamientos revelaron dificultades similares para identificar argumentos cuando las conclusiones son falsas, y para identificar los usos argumentativos de las palabras indicadoras. Para superar estos obstáculos, se diseñó una intervención didáctica. Tras ella, los estudiantes mostraron mejoraron significativamente en la identificación de argumentos, así como en su precisión y exhaustividad. Estos resultados apoyan la eficacia de intervenciones específicas para subsanar deficiencias en el reconocimiento de argumentos.

Palabras clave:
educación secundaria; argumentación; pensamiento crítico; intervención pedagógica.

Introduction: Argumentation in science education, its importance and some challenges

Argumentation has become a central focus in science education, with increasing interest in recent years (Bricker; Bell, 2008; Erduran; Jiménez Aleixandre, 2008; Erduran; Simon; Osborne, 2004; Sampson; Clark, 2008) given its pivotal role in scientific practice (Driver; Newton; Osborne, 2000; Von Aufschnaiter et al., 2008) and its connection to broader educational objectives (Jiménez-Aleixandre; Bugallo Rodriguez; Duschl, 2000). The promotion of argumentative skills is essential for fostering critical thinking (Kuhn; Park, 2005) and enabling students to engage effectively in complex decision-making processes, such as those related to socio-scientific issues (Osborne; Erduran; Simon, 2004). It also facilitates access to cognitive and metacognitive processes characteristic of expert performance - thereby enabling modelling for students -, helps the development of communicative competences, and promotes the advance of reasoning skills, including the ability to select theories or positions based on rational criteria, among others (Erduran; Jiménez-Aleixandre, 2008).

Prior research indicates significant challenges in argumentation skills, particularly in informal reasoning contexts. For example, a study involving 160 participants of different ages, from adolescents to older adults, found that only a minority consistently provided substantial evidence for their theories, proposed alternative theories, presented counterarguments, and offered rebuttals (Kuhn, 1993). Lin (2014) highlights the important disparities between science and non-science undergraduate students concerning the level of sophistication of the arguments they elaborate when asked to present reasons to support or reject an opinion in a report. Also focused on undergraduate students, Cheng and Yang (2022) found differences in their ability to correctly identify argument components, analysed along the lines of Toulmin’s (2003) proposal and a correlation between these differences and the beliefs that students held about scientific knowledge (for instance, about the level of certainty they attribute to it). Similarly, according to Sampson, Enderle e Grooms (2013) students often simply assert that the evidence they have obtained supports their claims but fail to justify that this evidence is adequately linked to the claims they try to defend. As to Lyzterinou and Iordanou (2020), they show that teachers themselves display important differences in their ability to evaluate arguments, which, in turn, correlates with their ability to construct their own arguments.

In view of these difficulties, numerous efforts have been made to promote the use of arguments in science classrooms, with various frameworks ranging from Toulmin's original proposal, widely adopted in science education research, to modifications and extensions designed to address methodological challenges or broaden analytical scope (Duschl; Osborne, 2002; Kelly; Chen, 1999; Kelly; Takao, 2002; Larrain et al., 2021; Lawson, 2003; Lazarou; Erduran, 2021; Sampson; Clark, 2008; Simon; Erduran; Orborne, 2006b; Soysal, 2023; Von Aufschnaiter et al., 2008; Zembal-Saul et al., 2002; Zohar; Nemet, 2002).

Interestingly, however, these studies in the field of science education do not inquire into the question whether the problem with students’ argumentative skills may lie one step earlier - namely, in their ability to recognize the presence of an argument, to differentiate when a passage is argumentative and when it is not. In fact, if students are unable to recognize that they are in the presence of an argument, they will forcefully be unable to notice that their specific argumentative skills are required. Conversely, their appeal to critical thinking will be misplaced if, as Salmon comments, students “accuse someone of presenting a poor argument for a case” in a situation in which “no argument at all is offered” (Salmon, 2012, p. 14).

Now, this is no trivial affair. As we will see below, argument identification by nonexperts in general and students in particular is misled by factors which, from the point of view of the definition of ‘argument’, should not be considered. In our research, we attempted to identify some crucial factors underlying argument identification, first, by means of a survey among general population and, later, by implementing a brief didactic intervention focused on these factors to see whether they were relevant.

To understand the problem, we must first establish a more precise definition of what we understand by argument, particularly within the context of science education.

Theoretical framework

Beyond Toulmin: Argumentation theory and its reception in science education research

A widely used framework in science education research is Toulmin’s Argument Pattern (TAP) (Toulmin, 2003), which defines argumentation structures through interconnected components such as claims, data, warrants, backings, rebuttals, and qualifiers. TAP has served as an analytical lens for studying student and teacher discourse (Erduran; Jiménez Aleixandre, 2008; Kelly; Chen, 1999). However, its application in classroom settings has presented methodological difficulties, particularly in distinguishing between argument components and setting clear boundary markers for coding (Erduran; Simon; Osborne, 2004; Kelly; Chen, 1999). Some researchers argue that these challenges stem from underspecification rather than inherent limitations of the framework itself (Duschl, 2008).

Beyond TAP, alternative argumentation models have emerged, including Walton’s (1996) schemes for presumptive reasoning and the approaches which, as pragma-dialectical analysis (Eemeren; Grootendorst, 2004) and Bronckart’s (1997) socio-discursive interactionism, approach argumentation from the point of view in the context of dialogue, therefore as a way to discursively solve a disagreement or controversy. These approaches offer different perspectives on the contextualized use of language and reasoning structures in science education. Nevertheless, their methodological contributions remain relatively underexplored compared to TAP (Reiser et al., 2007).

In any case, there is no single theoretical framework which encompasses all the approaches to argumentation in the field of science education - a situation which is further complicated by the problem of the multimodal nature of argumentation, which may go beyond linguistic components in a strict sense¹ (Larrain et al., 2021; Mikeska; Howell; Kinsey, 2022; Simon; Erduran; Osborne, 2006; Soysal, 2023; Zembal-Saul et al., 2002), and that of the relationship between argumentation and evidence² (Erduran; Simon; Osborne, 2004; Felton et al., 2022; Kelly; Takao, 2002; Simon; Erduran; Osborne, 2006; Zembal-Saul et al., 2002). Nevertheless, without attempting to merge all the different approaches to the phenomenon of argument, we will focus on two traits.

Two traits of arguments: Discursive role and relational character

In her seminal article about “science as argument”, Kuhn (1993, p. 323) highlights how, when two interlocutors argue to solve a disagreement, they must “connect supporting and refuting evidence to each of the assertions”. As to Jiménez-Aleixandre (1998, p. 210, own translation), when she tackles “the criteria we use to say that students are ‘talking science’, that they are communicating in the 'language of science'", the focus is on the distinction between “a form of argumentation” and “dogmatic opinions”: an “important part of the adoption of scientific literacy by students” consists in “explicit discussion in the classroom of the criteria by which one hypothesis is preferable to another, the relationship between experimental testing, the data and the proposed solution to a problem”. Here again, we offer arguments when we need to solve some kind of opposition (one hypothesis needs to be shown as preferable to another). Similarly, when Driver, Newton and Osborne (2000, p. 291) turn to the definition of argument, the second approach they consider, the multivoiced one, highlights the fact that an argument takes place “when different perspectives are being examined, and the purpose is to reach agreement on acceptable claims or courses of action”: here again, some kind of disagreement or opposition is required. As is well known, the development of this dialogical approach in the theory of argumentation is due to the work of Eemeren and Grootendorst (2004) and Bronckart (1997), among others.

This second trait is that arguments have a relational character: they connect, relate, a conclusion with one or more premises - and not in any way: premises are offered as, specifically, support for the conclusion. In Kuhn’s passage, this is the relationship between an assertion and the evidence supporting it. According to Jiménez Aleixandre’s (1998) text, unlike a dogmatic opinion, in an argument we support a hypothesis by giving reasons for it, by virtue of which the hypothesis in question is preferable - again, there is a relationship between linguistic elements at play. In the rhetorical view which Driver, Newton and Osborne (2000) consider along with the multivoiced one, it is a relationship between a proposition and a reason (they cite the Oxford English Dictionary: an argument involves “advancing a reason for or against a proposition or course of action”). Finally, in Klein, Olson, and Stanovich (1997, p. 38), insofar as they point out that “evaluating arguments” requires “consideration of the evidence they contain”, and such evidence “can be understood as ‘reasons for belief’”, this involves the idea of a relationship or connection between a conclusion (presented here as “belief”) and premises (or “reasons”).

Some opposing views, and a necessary distinction: Arguments and explanations

Now, although we will understand arguments in a way which attributes to them the two traits which we have identified, and which, we believe, does justice to the proposals of influential researchers such as Kuhn (1993) and Driver, Newton and Osborne (2000), this elucidation of argument is not beyond controversy; in fact, the second trait has explicitly or implicitly been challenged in a way that can illuminate an important difficulty in argument identification: that of distinguishing between arguments (characterized by a specific relationship: that of justification or reasons for a belief) and explanations (in which the key relationship is that of making understandable, intelligible). In fact, the relational character of arguments is challenged even more radically by the authors who do not view them as having a relational structure of any sort, because, in their view, even isolated claims can count as arguments. Zeidler et al. (2003, p. 107) write: “the simplest arguments are those consisting of a claim” - i.e., a claim not supported by any premises, or, in Toulmin’s words, by “warrants, backing or data”, the presence of which these researchers identify as typical of higher-level arguments. This notion of a level 1 reappeared later (Dawson; Venville, 2009, p. 1430), again without questioning whether an isolated claim is an argument at all; was taken up by authors who class under this label a “Claim without a justification” (Lytzerinou; Iordanou, 2020, p. 6), or “an argument composed of a claim” (Lin, 2014, p. 1028), or simply changed its number (Karişan; Tüzün; Zeidler, 2017) to level 0 argument - again, instead of viewing the corresponding passages as not argumentative as a consequence of the lack of the relational character we have identified as a necessary condition for the presence of an argument.

But, more importantly, the specific kind of relationship involved has sometimes been assimilated with one which is not of support or justification. In this vein, Revel Chion (2010, p. 166, own translation) writes that “in sciences, argumentations are explanations about phenomena”, that “scientific school argumentation” is “the production of a text in which a natural phenomenon is subsumed under a theoretical model by means of a mechanism of an analogical nature” (Revel Chion et al., 2005, p. 2), similar remarks can be found in Ruiz Ortega, Márquez Bargalló and Tamayo Alzate (2014), in Ruiz Ortega, Tamayo Alzate and Márquez Bargalló (2015) and in Agudelo Escobar (2017). However, the relationship between an explanandum and the corresponding explanans is not that which holds between a conclusion and the supporting premises. As Eemeren, Grotendoorst and Henkemans (2002, p. 43), the creators of the pragma-dialectical approach, point out, in a sentence such as “[…] the pudding didn’t stiffen because I didn’t put enough gelatin in it”, what we are dealing with, in spite of the presence of the word “because”, is not an argument; the utterer is here giving us “causes rather than reasons”. If something “is being explained”, they comment, it is “Something that is already accepted”, whereas arguments “are always brought to bear on a standpoint that has not yet been accepted” (Eemeren; Grotendoorst; Henkemans, 2002, p. 43). We explain why the pudding did not stiffen because we take for granted that it did not stiffen, and this is certainly different from an argument, in which we want to show, not why something is the case, but that it is in fact the case. Similarly, Grize (1981, p. 8, own translation, own emphasis), whose proposal was in turn taken up by Bronckart (1997, p. 237), distinguishes explanatory relationships such as the one in the train is late because it’s snowing from justificatory ones, as in “‘The whole dignity of man lies in thought,’ said Pascal. All our dignity consists,

therefore, in thought”. But it is important to notice, at this point, that the distinction these authors make in order to class some uses of language as argumentative and some others as explanatory appeals not only to the idea that these uses involve different kinds of relationships (A justifies B is not the same as A explains B) but also to the notion of a subjective intention to employ language one way or the other (Salmon, 2012, p. 13-14). That is, calling something an argument is claiming that a subject has the intention to present some information as his or her grounds for supporting a conclusion. And this, in turn, is not as trivial as it may seem. To which we now turn.

A second difficulty: a possible normative component in the concept of argument: Intention and achievement

According to the minimal approach to arguments which we have adopted, for some piece of language to count as an argument it is not necessary that the reasons provided in it as support for its conclusion be good enough. An approach in which something is an argument insofar as someone offers some reasons as support for a conclusion, in order to solve an actual or potential disagreement, does not require that we evaluate the reasons and find that they do in fact provide some support for the conclusion in question. In this view, we can classify something as an argument even if, in our view, the reasons provided are so hopelessly bad that the amount of support they provide is zero. All we need to do is to check for indicators that the utterer of a piece of language intends to provide reasons for a conclusion. Under this approach, a text such as grass is green, so cows surely feed on it should count as an argument, because of the presence of the indicator word so, which cannot be understood in any other way, and the epistemic marker surely. While this would constitute a very poor argument, it would still be considered an argument.

Interestingly, however, our study had to test responses to arguments which provide zero support to a conclusion (as well as arguments which provide support other than zero, but still very bad ones) because, from a strictly empirical point of view, our anecdotal evidence (for which we attempted to substitute robust, systematic evidence) pointed in the direction of the hypothesis that people tend to employ the word argument in a normative, commendatory sense: they class something as an argument more often when they somehow consider it good enough (a tendency which we took into account in devising our tools; cf. infra, Hypotheses). As to the theoretical background, the claim that we should work with a normative concept of argument, such that someone may have the intention to provide an argument but actually fail to do it (so their utterances end up not being argumentative at all) is not absent from the literature: it was explicitly proposed by Blair (2012, p. 143-144, 147) and later defended by McKeon (2024). If we compare this normative approach to its mainstream alternative, the pattern which emerges is not that the latter posits a strictly non-normative concept of argument, which does not involve any definitional relationship with the idea of good reasons, but that it posits a concept in which the presence of good reasons is a subjective affair: we class someone else’s utterances as an argument if and only if we attribute to this other person the belief that some reasons are good enough to provide at least some support to a conclusion. By offering these reasons, this person thus displays their intention to support a conclusion. On the contrary, if we understand arguments in the way Blair proposes, the intention is not enough. This normative approach, though it is interesting and (as we will see below when we introduce the results of our research) seems to coincide at least partially with laypeople’s approach to the notion of argument, remains nonetheless a minority view, so what we will take into account when classifying the responses as correct or incorrect will be the subjectivistic criteria held by most authors, according to whom argument identification is identification of an intention, not of some kind of success. At this point, however, let us focus on a difficulty these subjectivistic criteria create for our research purposes.

Arguments, explanations and the problem of determining intention

The problem is the following: even if, contrary to the approaches which tend to obliterate the distinction between explanations and arguments in general, we insist that these are in fact different, as the mainstream approach claims (other sources in which this distinction is proposed and defended are Bassham et al. (2010), Boss (2012), Copi, Cohen and MacHahon (2014), Eemeren, Grootendorst and Henkemans (2002), Walton (1996), it may still be complicated to determine, in some particular cases, whether we are dealing with an explanation or with an argument, because this determination depends on an intention, and intentions may be hard to find - especially if we work with short isolated texts, as those which pragma-dialecticians and Grize (1981) offer in the examples we have just seen. There is one notorious example, the risks for managers passage, originally presented by Thomas (1981, p. 13) and later taken up by Govier (2018, p. 246) and Walton (1996, p. 68), in which it is not clear whether the authors are trying to justify the claim that managers are exposed to developing insecurity and fear or to explain why this potential development exists³ (Massie; Douglas, 1973, p. 10). The words thus and since which appear in the passage might be functioning just like the word because in the example concerning gelatin, that is, not as markers of an argument but of an explanation; however, unlike the gelatin case, things are not unequivocal here. In view of all of this, even though arguments are usually identified on the basis of the presence of “indicators of argumentation”, such as “therefore, thus, so, consequently, of course, because, since, given that” (Eemeren; Grootendorst; Henkemans, 2002, p. 39), they can be misleading, and in these cases we can only appeal to context in order to find an answer for the question “where did this particular passage occur as part of whatever it was the author was trying to do?” (Walton, 1996, p. 96). The presence of these indicators is thus not sufficient for the presence of an argument, and it is not necessary for arguments either: we can (and usually do) articulate an argument by simply juxtaposing the sentence which expresses our opinion with those which express our reasons supporting it - but this, in turn, generates a problem: the possibility of arguments without indicator words makes it difficult, in some cases, to determine whether we are dealing with an argument that omits such indicators or simply with another type of discourse, such as a narrative account, which lacks justificatory relations. We would have, in that case, simply different “items of information meant to fill the reader in on what is happening” (Fawkes, 1996a, 1996b, 2003; Walton, 1996, p. 95). For there to be an argument we need different propositions mutually connected by the speaker’s perceived intention of grounding one on the other, in the face of a possible disagreement with it, and this is something that on some occasions can be determined only on the basis of context. Consequently, both the diagnostic task of deciding whether a given subject performs well in argument identification and the didactic task of providing them with tools to improve this performance may appear as tasks which go beyond what is achievable by the usual tools. Nevertheless, we can sketch a brief defence of our approach (which is, we insist, the same as in many of the authors who have written on the subject).

A defence of testing responses to isolated sentences

Our point is that, while we in some cases need to reposition a sentence within a lengthy text to determine whether it presents an argument, this does not mean that such a need always arises. In a sense, the problem with classifying texts as argumentative or non-argumentative is similar to other classifications in that it must be taught and evaluated, and which include grey, difficult cases. Teaching students to classify between mammals, amphibians, and reptiles, for example, also involves grey cases, such as the platypus: how can a mammal lay eggs? But the existence of platypuses, and the difficulties it introduces, does not prohibit us, as teachers, to appeal first to clear-cut, paradigm cases of mammals, such as the cow or the dog - which is not a trivial affair given that it is a scientific distinction which does not necessarily emerge in a spontaneous way and which has to compete with other possible ways of classifying animals. Similarly, teaching what kind of thing an argument is, and how it differs from non-argumentative uses of language, is something that we can do based on clear-cut examples, whose classification is not, nevertheless, trivial, and this does not exclude the possibility that grey cases may appear later. Let us put this in very concrete terms, returning to the pragma-dialecticians’ example: in “The pudding didn’t stiffen because I didn’t put enough gelatin in it” (Eemeren; Grotendoorst; Henkemans, 2002, p. 43), it is pretty clear for the trained reader, even in the absence of further context, that the because in question does not function as an introduction of the reasons, the evidence, on the basis of which we affirm that the pudding didn’t stiffen, but as an introduction of the explanation of why the pudding didn’t stiffen. First, there is a very plausible causal connection between the two phenomena mentioned (and causes are the key for at least some kinds of explanations). Second, if we were in the presence of an argument, we would find some epistemic marker, something like The pudding surely didn’t stiffen, or The pudding can’t have stiffened. Even if our epistemic markers highlight certainty, the thing is (perhaps paradoxically) that this is not something we do when we regard something as obvious. On the contrary, when a position is controversial and argued for, we refer to our level of certainty for it.

So let us consider that we can obtain interesting results by analysing responses to isolated sentences as those which can be found in the literature. In particular, this analysis attempted to identify which particular factors in the sentences in question conditioned argument identification.

In search of the crucial factors

In the early 1980s, Eemeren, Grootendorst and Meuffels (1984, p. 299) highlighted the roles of four such factors: (i) the place of premises in the argument (before or after the conclusion); (ii) the presence of argumentative indicators; (iii) the marking of the point of view (with expressions such as in my opinion), and (iv) the loaded character of a topic.

Regarding (i) and (iii) we agreed with this classical approach and decided not to gather evidence about them again.

Regarding (ii) and (iv), we thought that the researchers did not consider two aspects which, as we anticipated, we deemed to be crucial in argument identification.

First, on (ii), by only distinguishing between sentences with or without “argumentative indicators”, they do not consider the problem whether some such indicators (as is the case with because in English and porque in Spanish) may have uses which are not argumentative but explanatory and thus lead to confusion.

Second, on (iv), whereas these authors made the valuable discovery that the difference between loaded and not loaded topics influences our ability to identify arguments about them, they did not further distinguish between cases in which a true, or at least plausible, position about those topics is defended, and those in which the position defended is viewed as clearly false. But let us delve into this a little deeper.

With the terms loaded and not loaded, the authors contrast topics on which “(almost) everyone has an opinion” and which “are regularly under discussion at any given time” and topics on which subjects do not presumably “have an unqualified opinion” - the difference between, in their examples, “the existence of nuclear weapons”, on the one hand, and “the way my neighbour pastes his stamps or on the introduction of a new boating regulation”, on the other hand. “For loaded topics”, they go on, “the discussion character (discussiekarakter) is in principle immediately clear, for non-loaded topics it is often only apparent from the context that a position is being taken on the topic whose acceptability is at stake” (Eemeren; Grootendorst; Meuffels, 1984, p. 301). From this point of view, what is relevant in acknowledging that some use of language is argumentative is whether it refers to a loaded topic, irrespective of whether the position defended appears as plausible or not. So, in a contemporary context, if we focus on the fact that some bizarre positions, such as the belief that the Earth is a flat, motionless disc, are regularly under discussion in social networks, we should expect the public to easily identify as such the arguments proposed by flat-Earthers: the fact that the positions being defended are utterly implausible should not be an obstacle to argument identification. Our conjecture, however, was that the plausibility of the claims being defended is a crucial factor: insofar as people tend to use argument in a normative way, they will, caeteris paribus, find it more difficult to detect the argumentative character of a use of language when the position being defended is deemed implausible (the Earth doesn’t move or all mammals have four legs).

In light of these reasons, we designed a survey and a didactic intervention on the basis of the following specific hypotheses:

1. Individuals find it more difficult to acknowledge the argumentative character of a text with a false conclusion than that of a similar text with a true conclusion; that is, they tend to assimilate argument to good argument, in the specific sense that its conclusion is true.

2. Individuals find it difficult to distinguish between argumentative and explanatory uses of the word because.

Objective

The general aim of this study was to identify some of the crucial factors that underlie argument recognition.

Participants and Methods

This research was conducted in two phases to test our hypotheses. First came the diagnostic phase, during which we developed and administered an online survey aimed at the general public and a written survey to a group of fourth-year students. Then came the interventional phase, in which we implemented a didactic intervention with the group of fourth-year students based on the information gathered during the first phase.

A survey among the general public

We developed an online instrument in which each participant was presented with a series of exercises. The first of these involved deciding which out of 14 sentences, of which we will be analysing here 11, were arguments and which were not (table 2). In each case, we find only one sentence (in the grammatical sense of a unit which begins with a capital letter and ends with a period) but, of course, in some cases we can distinguish two or more propositions (in the logical sense of a unit which is susceptible of a truth-value) and, therefore, possible justificatory relationships between them. The instrument was shared through various social networks, and participation was voluntary and consensual. Each participant was asked for additional information regarding their education, age and gender to delineate the demographic profile. A total of 1217 participants completed the survey: regarding their gender, 62.7% were female, 35.4% were male, 1.4% selected ‘other’ and 0.7% chose not to answer. The age distribution is shown in table 1.

Table 1
Age distribution among 1217 participants

Age range	%
16-20	1,1
20-29	23,8
30-39	33,3
40-49	26,0
50-59	10,8
60-69	4,5
> 70	0,5

Table 2
Percentage of correct answers for each sentence for general public (GP) and 4th-year high-school students (pre-intervention)

Are these sentences arguments?	GP % correct answers	4th year % correct answers	p-value
Unlike planets, stars shine with their own light.	83.1	78.7	0.49
My grandmother always told me: mate4 without sugar and sweet tortas fritas.	92.9	91.5	0.76
The issue of anti-scientific rhetoric is an urgent concern internationally since studies in various countries have shown that more and more people are sceptical of vaccines.	75.5	70.2	0.46
Since Messi was born in Rome and Rome is in Italy, Messi was born in Italy.	74.7	70.2	0.55
All metals expand with heat. Iron expands with heat; therefore, iron is a metal.	69.5	70.2	0.85
In Brazil people speak Portuguese, so it was probably a Portuguese colony.	64.0	53.2	0.17
We don't perceive any movement; therefore, the Earth doesn't move.	56.0	53.2	0.81
The dog, the cat, the cow, the mice are mammals and have four legs; therefore, all mammals have four legs.	46.6	51.1	0.46
Dinosaurs became extinct because a meteorite fell to the Earth.	40.8	23.4	<.05
George is saving money because he wants to go on vacation.	43.5	25.5	<.05
We should not get vaccinated against COVID-19 because Lucía Langer, an immunologist at the Italian Hospital, said that the vaccines contain graphene hydroxide and graphene, which is used to control people.	58.3	55.3	0.78
Average % correct answers (SD)	64.1 (16.1)	58.4 (19.9)	0.47

As for their education, 94.4% of the participants reported being currently enrolled in or having completed tertiary or university studies (that is, all of them at least have completed high school). This bias towards higher education is due to the authors distributing the survey on their social networks. Despite having a significant number of followers, the majority are pro-science and have continued their studies beyond high school. This bias, which might seem like an obstacle, is actually not for this study: we will see that even individuals from the general public with completed or incomplete tertiary or university studies show serious difficulties in distinguishing whether they are or not in the presence of an argument.

Upon analysing these findings, we resolved to initiate the second phase of the research, focusing specifically on an educational environment. In particular, first we wanted to ascertain if the results obtained with the general public were replicated in high school students, and - if this was the case - to explore the feasibility of developing an effective intervention strategy to address some obstacles.

The didactic approach: an intervention with high-school students

For the second part of the research, 47 fourth-year students (aged 15-16) from a private secondary school in the city of La Plata, Argentina, took part. Authorization was obtained from the school principals and, through them, from the students' families. Participation was voluntary.

To evaluate the effectiveness of our intervention, we employed a One-Group Pretest-Post-test Design which is, as is known, a paradigmatic example of “quasi-experimental designs” (Campbell; Stanley, 1963; Shadish; Luellen, 2006).

This kind of designs risk celebrating the success of an intervention which may have had no effect: “lack of a pretest makes it difficult to know if any change at all occurred in those who received treatment - maybe people did not change at all after treatment” (Shadish; Luellen, 2006, p. 543). To avoid this difficulty, a pretest is necessary, and our research design took this into account.

However, as long as the comparison is made only between the results of the pretest and those of the post-test (and not also between a treatment group and a control group), the possibility remains that the “assumption” “that people would not have changed at all over time without treatment”, and which lies behind the conclusion that “any change that is observed over time” indicates a treatment effect, may be on occasion “very dubious”, in particular in an educational context: given that “students were developing intellectually anyway”, perhaps “their scores would have improved even without treatment” (Shadish; Luellen, 2006, p. 544). Nevertheless, there are considerations which exclude this possibility in a rather unequivocal way. Firstly, our pre-test and post-test were only a few days apart, so any differences in results among students cannot be attributed to general intellectual maturation. It is not plausible that a global improvement in such a specific competence as argument identification could be attributed to changes other than our intervention in such a short time. This is reinforced by a second consideration: improvement in argument recognition cannot be attributed to general intellectual maturation because it is a task that generates similar difficulties for both high school students and adults, including those with university degrees, as our research showed. Therefore, our work aimed to test the effectiveness of an intervention, and it was carried out with a design appropriate to this purpose.

Pre-intervention: for the pre-intervention, we applied the same instrument that we used for the general public survey but, in this case, we conducted the data recollection in person -with printed sheets - instead of using an online form.

Didactic intervention: it was intentionally designed to be concise, allowing completion within a two-class-hour period (80 minutes), addressing the prevalent issue of time constraints frequently faced by educators. It comprised the following activities.

Activity 1: What is an argument? (15 minutes). Students were initially engaged by discussing contemporary music preferences, prompting them to articulate their opinions on the best singer or band (Who is currently the best singer or band?). As divergent views emerged, we asked participants to justify their choices (for example, their lyrics are the best or she has sold a lot of albums). This lively debate served as a springboard to introduce the concept of argumentation. We told them that what they were doing was precisely arguing; that humans argue all the time when we talk about music, soccer, cooking, and also about science. By defining an argument as a form of discourse wherein a claim is substantiated by reasons, students grasped the distinction between conclusions (the opinions defended) and premises (the supporting reasons). We wrote these definitions on the blackboard.

At this point, we emphasized greatly that an argument can be very bad, but that does not mean it ceases to be an argument. We also emphasized that the truth or falsehood of the conclusion does not define whether it is an argument or not. We used as examples the sentences that had been part of the pre-intervention instrument, in particular #7, #8, #9 and #10 (table 2).

Subsequent control exercises were carried out: we presented them with some sentences, and they had to identify for each case if it was an argument and, if so, which was the premise and which was the conclusion.

Activity 2: exploring indicator words (10 minutes). Attention then shifted to indicator words - words that bridge premises and conclusions in arguments. Specifically, the significance of because was underscored, elucidating the distinction between explanations and arguments. Whereas arguments entail disagreement, explanations elucidate agreed-upon sentences of facts and their potential causes. Through further exercises, students practiced identifying explanations and arguments, honing their ability to discern between the two.

Activity 3: types of reasoning (40 minutes). Subsequently, we introduced deductive and inductive reasoning approaches to the students. We did not emphasize the deductive approach because preliminary results showed that the greatest difficulties were not found there, although we briefly presented it using examples from the instrument used for the pre-test, in particular, sentences #4 and #5 (table 2). While we briefly touched upon the deductive approach, our primary focus was on the inductive reasoning. Specifically, we highlighted three of the most common types used in science: inferences to the best explanation, generalizations from cases, and recourse to expert testimony. For each type, we elucidated their main characteristics. Then, we recounted brief narratives from science, prompting students to identify the type of arguments employed in each case.

Activity 4: evaluation of inductive arguments (15 minutes). As the final activity, we delved into the critical factors for evaluating the quality of these three types of inductive arguments. We emphasized that generalizations from cases are deemed legitimate when based on representative samples. Additionally, inferences to the best explanation must consider alternative hypotheses, ensuring that the chosen explanation is not only probable itself but also allows us to determine the probability of the phenomenon being explained.

Furthermore, in arguments relying on expert testimony, we stressed the importance of analysing the expertise of the speaker, identifying potential conflicts of interest, and assessing whether the expert aligns with the scientific consensus. At this juncture, we underscored that while some inductive arguments may lack legitimacy, such as those based on false expertise, they nonetheless retain their status as arguments. To illustrate these concepts, we employed examples from the instrument, namely sentences #3 and #8 (table 2).

Post-intervention: we designed another instrument comprising 14 sentences, 11 of which we will analyse here. These sentences were different from the pre-intervention sentences, but centered around the same categories (non-arguments, arguments with false conclusions, arguments with varied indicator words, and so on). As with the pre-intervention assessment, this phase was conducted face-to-face, one week after the intervention.

Statistical analysis

We ran t-tests of statistical significance with a 95% confidence level for each pair of sentences in table 2. We also ran t-tests of statistical significance with a 95% confidence level for each pair of sentences in the case of the general public since we wanted to see whether there were significant differences between the percentages of correct answers for some specific pairs of sentences. In all cases p-values were less than .05 except for the pairs shown in table 3.

Table 3
Percentage of correct answers and p-values for pairs of sentences that did not differ significantly out of all possible combinations

Are these sentences arguments?	% correct answers	p-value
The issue of anti-scientific rhetoric is an urgent concern…	75.5	.64
Since Messi was born in Rome and Rome is in Italy, Messi was born in Italy.	74.7
We don't perceive any movement; therefore, the Earth doesn't move.	56.0	.37
When taking a penalty kick, we must always pay close attention…	54.1
We should not get vaccinated against COVID-19 because Lucía Langer…	58.3	.25
We don't perceive any movement; therefore, the Earth doesn't move.	56.0
George is saving money because he wants to go on vacation.	43.5	.18
Dinosaurs became extinct because a meteorite fell to the Earth.	40.8
The dog, the cat, the cow, the mice are mammals and have four legs…	46.6	.12
George is saving money because he wants to go on vacation.	43.5

In the case of high school students, our aim was not to compare the percentages of correct answers for these sentences amongst themselves, but rather to contrast them with those obtained from the general public. That is why we did not perform this test with them.

Precision and recall

While comparing the percentages of correct answers offers a broad perspective on the general public's ability to identify arguments, it does not shed light on a more specific aspect: the types of errors participants are making most frequently. To address this, we can employ additional metrics: precision and recall. Precision aims to determine the number of correct positive identifications; recall, on the other hand, assesses the number of true positives correctly identified. To calculate these parameters, it is necessary to define what constitutes positive and negative. In this context, since we consider it more serious to not identify an argument when it is there (as this would imply, among other issues, the risk of taking the conclusion as the statement of a fact rather than as an opinion that the speaker is defending with reasons), identifying arguments has been designated as ‘positive’. Therefore, when evaluating a response, there are four possible outcomes:

• - true positive claims: Arguments correctly identified by the participant;

• - true negative claims: Non-arguments correctly identified by the participant;

• - false negative claims: Arguments incorrectly identified as non-arguments (type I errors);

• - false positive claims: Non-arguments incorrectly identified as arguments (type II errors).

Since precision represents the number of correctly identified ‘positives,’ in this case, it indicates how many sentences identified as arguments were indeed so. Recall, on the other hand, denotes the proportion of true ‘positives’ correctly identified, which in this case refers to the total number of arguments correctly identified by the participants. Both precision and recall values range from 0 to 1, with a participant achieving a precision of 1 if they do not provide any false positives, and a recall of 1 if they do not have any false negatives.

Results

Tables 2 and 3 show the percentage of correct answers for each sentence, broken down by the general public (GP) and 4th-year high school students (pre-intervention). Arguments are highlighted in italics, and the right-hand column lists the p-values. for clarity, we have chosen to present them in this order, but participants received them in a random order.

Table 4 shows the percentage of correct answers to each of the eleven sentences presented following the didactic intervention.

Table 4
Percentage of correct answers for each of the 11 sentences presented after the didactic intervention

Are these sentences arguments?	% correct answers
Pizza definitely does not have pineapple as an ingredient.	91.3
Argentina has completely free top-notch universities, in contrast to most countries.	89.1
María Becerra is Argentinean and Argentinean women sing well, so María Becerra sings well.	93.5
Marie Curie was a talented scientist and was a woman; Alice Hamilton was a talented scientist and was a woman; Cecilia Payne was a talented scientist and was a woman; therefore, all talented scientists are women.	91.3
Chagas disease is endemic in Chaco. Chaco is a province in Argentina. Therefore, Chagas disease is endemic in an Argentinean province.	89.1
Since frogs are mammals and mammals do not lay eggs, frogs do not lay eggs.	87.0
We don't see any curvature, so surely the Earth is flat.	60.9
Writing surely originated in present-day Iraq, because there we found the oldest written tablets.	91.3
The attitude of teens towards anti-scientific discourses is surely worrying, since various studies show that many are unable to distinguish astronomy from astrology.	91.3
Columbus arrived in America because he was looking for a passage to the Indies.	69.6
Ezequiel is worried because he could lose his job.	67.4

In the following section, we will discuss in depth these results and what we can infer from them.

Discussion

Initially, let us examine the first 11 sentences and the choices made by each group, general public and high school students.

General public

Sentences #1 and #2: components of an argument

As we saw, an argument consists of two main parts: the conclusion, which is the main point that the arguer is trying to prove, and the premises, which are the reasons provided to support the conclusion. An important issue that emerges from this presentation is that when we identify something as an argument, what we are doing is to attribute to someone else the intention of supporting an opinion (conclusion) on the basis of some evidence (premise). We are not claiming ourselves that the evidence does support the opinion - which is why, as we will see, we can identify something as an argument and realize that it is a very bad one.

What is implicit in the notion of a support for the conclusion provided by the premises is the concept of truth. An argument is a connection between sentences in which believing that the premises are true functions as a reason for believing that the conclusion is true (Salmon, 2012). This is why there can only be an argument insofar as we have at least two pieces of information each one capable of being true or false. In the sentence Unlike planets, stars shine with their own light, this condition is not met. This sentence can be divided in a variety of ways, but none of these will yield two propositions - two pieces of information that can be true or false. In particular, unlike planets cannot have a truth-value. Therefore, we cannot use this sentence to support a conclusion provided by a premise.

Additionally, there is no argument when, even if there are two pieces of information which can have a truth-value, the relationship between them cannot be interpreted as being one of justification or support. This is exemplified by the sentence My grandmother always told me: mate without sugar and sweet tortas fritas: there is no indication of an argument; instead, it simply quotes the grandmother's advice.

In both instances, most participants successfully recognized that they were not dealing with arguments: 83.1% for the planets sentence and 92.9% for the grandmother's sentence (table 5).

Table 5
Sentences that are not arguments because they lack a premise and/or conclusion

Sentences	% correct answers
1. Unlike planets, stars shine with their own light.	83.1
2. My grandmother always told me: mate without sugar and sweet tortas fritas.	92.9

Our hypothesis for the statistically significant difference between the percentages lies in the perception of the participants regarding the content of the sentences: in the case of the planets, 16.9% might have interpreted that this was an argument due to the true nature of the description: it is indeed the case that unlike planets, stars emit their own light. Conversely, in the grandmother's case, given its clear status as her personal opinion, participants could have been less inclined to accept it as inherently true, thus not considering it as an argument. We will see that these differences in percentages arising from a poor association between the truth value of a sentence or part of a sentence (whether it is true, false or indeterminate) and its identification as an argument reappear in the following analysis.

Sentences #3, #4, #5, #6, #7 and #8: indicator words as a clue

Certain words like so, consequently, therefore, and since are indicator words that serve as indicators of an argument's presence. Sentences #3, #4, #5, #6, #7 and #8 contain these words and should have been readily identified as arguments by the participants. However, there were noticeable differences in the percentages of correct responses (table 6).

Table 6
Sentences that are arguments with indicator words linking the premise(s) and conclusion

Sentences	% correct answers
The issue of anti-scientific rhetoric is an urgent concern internationally since studies in various countries have shown that more and more people are sceptical of vaccines.	75.5
Since Messi was born in Rome and Rome is in Italy, Messi was born in Italy.	74.7
All metals expand with heat. Iron expands with heat; therefore, iron is a metal.	69.5
In Brazil people speak Portuguese, so it was probably a Portuguese colony.	64.0
We don't perceive any movement; therefore, the Earth doesn't move.	56.0
The dog, the cat, the cow, the mice are mammals and have four legs; therefore, all mammals have four legs.	46.6

To explore the reasons behind these discrepancies, we will analyse two aspects: the truth value of the conclusions and the type of reasoning involved.

• - Sentences #3, #6, #7 and #8: true v. false conclusions. One of the reasons that allow us to explain the discrepancy in the number of correct answers is the truth value of the argument's conclusion: we hypothesize that when individuals encounter a conclusion that contradicts their existing knowledge or understanding of the topic, they are more inclined to indicate that the sentence is not an argument; conversely, when they find conclusions that align with their previous beliefs, they may be more inclined to accept them as arguments. This is what we see in the case of the number of legs of animals and the movement of the Earth (false conclusions) compared to Brazil as a Portuguese colony and the growing scepticism regarding vaccines (true conclusions).

• - Sentences #4, #5 and #8: deductive v. inductive arguments. To explain the similarity in the percentages of correct answers between #4 and #5, we cannot rely solely on the analysis of the truth value of their conclusions because this approach fails to account for the observed pattern: despite its false conclusion, people identified more correctly that #4 was an argument. Instead, we must delve into the type of reasoning involved. We may interpret that participant found deductive arguments easier to identify than inductive ones. It is worth noting that #5 represents an invalid form of reasoning, yet this flaw did not seem to deter participants from recognizing it as an argument. Whether this stemmed from an understanding that a fallacy still constitutes an argument or simply a failure to recognize its invalidity remains uncertain. Conversely, a similar analysis may explain the discrepancy between #4 and #8. Both have evidently false conclusions, yet #4 is a deductive argument while #8 is a (bad) induction. The presence of an illegitimate generalization in #8 likely led participants to perceive it as non-argumentative, despite its inherent argumentative nature.

Sentences #9, #10 and #11: the explanatory because v. the argumentative because

Generally, the presence of indicator words such as therefore, so, consequently, strongly suggest that we are dealing with an argument. However, there is one particular indicator word that can cause confusion.

Neither Dinosaurs became extinct because a meteorite fell to the Earth nor George is saving money because he wants to go on vacation constitute arguments. However, we should not get vaccinated against COVID-19 because Lucía Langer, an immunologist, said that the vaccines..., does (table 7).

Table 7
Sentences containing argumentative and explanatory uses of because

Sentences	% correct answers
Dinosaurs became extinct because a meteorite fell to the Earth.	40.8
George is saving money because he wants to go on vacation.	43.5
We should not get vaccinated against COVID-19 because Lucía Langer, an immunologist at the Italian Hospital, said that the vaccines contain graphene hydroxide and graphene, which is used to control people.	58.3

In the cases of the dinosaurs and George, because does not introduce the reasons to defend a sentence, but rather the causes that explain a phenomenon. Nothing in this text indicates that George is saving money is a sentence in dispute that needs to be defended with reasons. We do not want to justify our assertion that he is saving, but to explain the fact that he is doing so. Similarly, in the example of the dinosaurs, we are offering the cause that explains a fact that we take for granted. We are not trying to justify the assertion that (surely) the dinosaurs became extinct. However, in the case of COVID-19 vaccination, the use of the indicator word because indeed takes us from conclusion to premise. It introduces the reasons that justify the opinion We should not get vaccinated, which we certainly do not take for granted.

We had anticipated that the general public might not have a clear understanding of this distinction, and thus, we deliberately chose sentences that would help us ascertain whether this was indeed a prevalent issue. Consequently, we selected a sentence explaining a non-scientific matter (George), another concerning scientific phenomena (Dinosaurs), and an argument regarding a scientific issue (Vaccines) to see whether there were significative differences in how participants identified them. As expected, the data showed that people perceived all three cases as arguments in roughly similar proportions: 59.2% of the participants claimed that Dinosaurs was an argument, 56.5% thought that George was one too and 58.3% said that (correctly) Vaccines was an argument.

While the difficulties the general public encounters in discerning between argumentation and explanation might appear as a minor hurdle, it holds significant implications for scientific literacy in our societies. How can we expect citizens’ meaningful engagement in discussions surrounding socio-scientific dilemmas if they lack the ability to discern between matters under debate and established scientific facts or consensuses? This critical distinction is foundational to fostering informed dialogue and decision-making on complex issues, highlighting the pivotal role of educational initiatives aimed at enhancing public understanding of scientific processes and discourse.

Precision and recall

For the 1217 evaluated cases, a precision value of 0.75 (SD: 0.20) and a recall value of 0.64 (SD: 0.27) were obtained.

a. Fourth-year students (pre-intervention) v. General public

As it can be seen in table 9 from the p-values for each pair and for the comparison between the average percentages of correct answers, the results obtained with the general public were replicated with the population of high school students. The only exceptions were sentences #9 and #10: students were significantly worse at evaluating the sentences that contained an explanatory because. That is, they exhibited comparable capabilities in identifying (or failing to identify) arguments, yet they faced even greater challenges in distinguishing between arguments and explanations compared to the general public (let us remember that this particular general public was biased toward people with tertiary or university complete o incomplete studies).

Table 9
Comparison of average percentage of correct answers, precision and recall between the preand post-intervention with fourth year high-school students

	Pre-intervention (N=47)	Post-intervention (N=46)	p-value
Average % correct answers	58.4 (19.9)	83.8(12.0)	< .001
Precision (SD)	0.69 (0.19)	0.89 (0.11)	< .001
Recall (SD)	0.60 (0.25)	0.84 (0.16)	< .001

b. Fourth-year students: pre-intervention v. post-intervention

Table 8 shows a comparison of students’ capacity to recognize arguments before and after the intervention but, this time, according to categories that stem from the analysis of the results from the general public, in which we identified obstacles such as difficulty in identifying arguments when they have false conclusions or distinguishing between the causal and the argumentative because, among others.

Table 8
Comparison of students' skills of identification of arguments before and after the intervention in terms of average percentage of correct answers

Are these sentences arguments?	Pre-intervention average % correct answers	Post-intervention average %correct answers
Non-arguments (mere juxtaposition of propositions; one proposition alone)	88.0	90.2
Indicator words as a clue	61.4	85.5
Non-arguments (explanatory because)	24.4	68.5

As we can see in table 8, after the intervention, students improved their ability to identify different types of arguments, overcame some of the obstacles described in previous sections and maintained their capacity to effectively identify that if a sentence lacks at least a premise and a conclusion, one of which can support the other, then it cannot constitute an argument. They improved their ability to distinguish between causal because and argumentative because which is essential for differentiating explanations from arguments.

Regarding the specific parameters of precision and recall, the values obtained preand post-intervention are shown in table 9 along with the average percentage of correct answers. Standard deviations for each case are displayed in brackets.

Further precision and recall analyses elucidate the nuances of students' reasoning processes. Before the intervention, high school students demonstrated slightly higher rates of type II errors (false positives) than the general public. More critically, however, they exhibited substantially more type I errors (false negatives), indicating a tendency to overlook arguments even when they were present.

Conclusions

In this study, we aimed to address the fundamental challenge of recognizing arguments as a precursor to improving argumentation skills within science education. Despite the growing emphasis on argumentation in science classrooms, our findings highlight significant difficulties among both the general public and high school students in discerning when they are encountering an argument.

Upon analysing the preliminary findings of our research, and even with our sample biased towards individuals with tertiary and university education, we identified significant challenges in argument recognition among the general public. These challenges included difficulties in discerning between arguments and non-arguments, as well as distinguishing between explanatory and argumentative uses of indicator words like because. As mentioned, this could lead to serious consequences when it comes to exercising responsible, scientifically informed citizenship for daily decision-making.

Motivated by these initial results, we embarked on a second phase of the research focused specifically on high school students. Through this phase, we sought to replicate our findings and explore the feasibility of implementing an effective but simple intervention strategy to address these challenges.

Our results indicate that the difficulties observed in argument recognition among the general public were indeed replicated in fourth-year high school students. However, students exhibited even greater challenges in distinguishing between arguments and explanations.

Following the intervention, 4th year high school students showed significant improvements in their ability to identify arguments, especially in cases involving arguments with false conclusions and in distinguishing between causal and argumentative because, as well as enhancements in both recall and precision parameters. This suggests that the intervention successfully addressed deficiencies in their abilities, leading to more accurate recognition of arguments and reduced rates of both false positives and false negatives.

Overall, our study emphasizes the critical role of foundational skills in argument recognition for fostering informed dialogue, critical thinking, and scientific literacy. By addressing these challenges and implementing targeted interventions, even with brief ones, we can better equip individuals to engage meaningfully in discussions surrounding complex socio-scientific issues and contribute to informed decision-making processes. This highlights the importance of incorporating targeted interventions into science education curricula to promote scientific literacy and informed decision-making among students.

• 1
  Recent research has recognized the role of gestures, visual representations, and other non-verbal elements in scientific reasoning. This calls for methodological advancements that can accommodate broader communicative modalities within argumentation analysis. Additionally, researchers have emphasized the need for methodologies that are sensitive to different educational levels, from primary to tertiary education, capturing variations in students' and teachers' argumentative practices.
• 2
  In fact, the role of evidence in argumentation remains an open methodological question. Distinctions between empirical and theoretical evidence in science classrooms pose coding challenges. Some studies suggest that Toulmin’s framework alone may not fully capture the epistemic complexity of scientific argumentation, particularly in socio-scientific discussions. To address these challenges, researchers have proposed hierarchical coding systems that integrate pedagogical strategies into argumentation analysis.
• 3
  The example reads as follows: “Managers who at an early stage showed much promise, career growth, and mobility may find themselves classified as nonpromotable for any of a dozen reasons [...]. Once labeled nonpromotable, a person is frequently put on a shelf and only tolerated within an organization. Thus, there is a great potential for the development of insecurity and fear in a manager in an organization, since he has few legal rights for his protection and must develop himself so that the organization continually views him as a valuable asset”.
• 4
  Mate is a beverage that is usually drunk without sweeteners. Tortas fritas, which can be sweet or salty, are often eaten with mate.

Ethics Statement

At the time this study was performed, ethics committee approval was not required. Nevertheless, we took measurements to ensure anonymity. Participation in the online survey was voluntary, consensual, and open only to people over the age of 16. We did not ask for any sensitive information. For the high school intervention, we obtained authorization from the school board and, through them, from the students’ families. Participation was voluntary. We did not record the classes or ask the students to identify themselves in the preand post-tests.

Data Availability Statement

The full dataset supporting the results of this study is available in the article itself.

References

• AGUDELO ESCOBAR, N. F. Desarrollo de una unidad didáctica en el área de ciencias sobre “el guayacán como flora nativa de mi entorno escolar”, del paisaje cultural cafetero Trabajo de grado (Maestría en Ciencias Ambientales) - Universidad Tecnológica de Pereira, 2017. Retrieved 13 Sep. 2025 from: https://tinyurl.com/yc4smsd5
  » https://tinyurl.com/yc4smsd5
• BASSHAM, G.; IRWIN, W.; NARDONE, H.; WALLACE, J. Critical thinking: a student’s introduction. New York: McGraw-Hill, 2010.
• BLAIR, J. A. Argument and its uses. In: TINDALE, C. W. (ed.). Groundwork in the theory of argumentation Amsterdam: Springer, 2012. p. 185-195.
• BOSS, J. A. Think critical thinking and logic skills for everyday life New York: McGraw-Hill, 2012.
• BRICKER, L. A.; BELL, P. Conceptualizations of argumentation from science studies and the learning sciences and their implications for the practices of science education. Science Education, Hoboken, v. 92, n. 3, p. 473-498, 2008.
• BRONCKART, J.-P. Activité langagière, textes et discours: Pour un interactionisme socio-discursif. Lausanne: Delachaux et Niestlé, 1997.
• CAMPBELL, D. T.; STANLEY, J. Experimental and quasi-experimental designs for research Palo Alto: Houghton Mifflin Company, 1963.
• CHENG, C. H.; YANG, F. Y. Analyzing visual attention during TAP learning and the effect of epistemic beliefs on the understanding of argument components. International Journal of Science Education, Abingdon, v. 44, n. 8, p. 1336-1355, 2022.
• COPI, I. M., COHEN, C.; MACMAHON, K. Introduction to logic London: Pearson Education Limited, 2014.
• DAWSON, V.; VENVILLE, G. J. High-school students’ informal reasoning and argumentation about biotechnology: An indicator of scientific literacy? International Journal of Science Education, Abingdon, v. 31, n. 11, p. 1421-1445, 2009.
• DRIVER, R.; NEWTON, P.; OSBORNE, J. Establishing the norms of scientific argumentation in classrooms. Science Education, Hoboken, v. 84, n. 3, p. 287-312, 2000.
• DUSCHL, R. Science education in three-part harmony: Balancing conceptual, epistemic, and social learning goals. Review of Research in Education, Thousand Oaks, v. 32, p. 268-291, 2008.
• DUSCHL, R. A.; OSBORNE, J. Supporting and promoting argumentation discourse in science education. Studies in Science Education, Abingdon, v. 38, n. 1, p. 39-72, 2002.
• EEMEREN, F. H. V.; GROOTENDORST, R. A systematic theory of argumentation: the pragma-dialectical approach. Cambridge: Cambridge University Press, 2004.
• EEMEREN, F. H. V.; GROOTENDORST, R.; HENKEMANS, A. F. S. Argumentation: analysis, evaluation, presentation. London: Routledge, 2002.
• EEMEREN, F. H. V.; GROOTENDORST, R.; MEUFFELS, B. Het identificeren van enkelvoudige argumentatie. Tijdschrift voor Taalbeheersing, Amsterdam, v. 6, p. 297-310, 1984.
• ERDURAN, S.; JIMÉNEZ ALEIXANDRE, M. P. Argumentation in science education: Perspectives from classroom-based research. New York: Springer, 2008.
• ERDURAN, S.; SIMON, S.; OSBORNE, J. TAPping into argumentation: Developments in the application of Toulmin’s argument pattern for studying science discourse. Science Education, Hoboken, v. 88, n. 6, p. 915-933, 2004.
• FAWKES, D. Critical thinking and its courses. Inquiry: critical thinking across the disciplines, Abingdon, v. 16, n. 1, p. 78-81, 1996a.
• FAWKES, D. On teaching premise/conclusion distinctions. Inquiry: critical thinking across the disciplines, Abingdon, v. 15, n. 3, p. 74-78, 1996b.
• FAWKES, D. Reliance on indicator terms is not critical thinking. Inquiry: critical thinking across the disciplines, Abingdon, v. 21, n. 3, p. 7-14, 2003.
• FELTON, M.; CROWELL, A.; GARCIA-MILA, M.; VILLARROEL, C. Capturing deliberative argument: An analytic coding scheme for studying argumentative dialogue and its benefits for learning. Learning, Culture and Social Interaction, Amsterdam, v. 36, 2022.
• GOVIER, T. Problems in argument analysis and evaluation Windsor: University of Windsor, 2018.
• GRIZE, J.-B. Logique naturelle et explication: introduction. Revue Européenne des Sciences Sociales, Strasbourg, v. 19, n. 56, p. 7-14, 1981.
• JIMENEZ-ALEIXANDRE, M. P. Diseño curricular: indagación y razonamiento con el lenguaje de las ciencias. Enseñanza de las Ciencias, Barcelona, v. 16, n. 2, p. 203-216, 1998.
• JIMENEZ-ALEIXANDRE, M. P.; BUGALLO RODRIGUEZ, A.; DUSCHL, R. A. “Doing the lesson” or “doing science”: Argument in high school genetics. Science Education, Hoboken, v. 84, n. 6, p. 757-792, 2000.
• KARIŞAN, D.; TÜZÜN, Ö. Y.; ZEIDLER, D. L. Quality of preservice teachers’ argumentation in socioscientific issues context. Journal of Human Sciences, Turkey, v. 14, n. 4, p. 3504-3520, 2017.
• KELLY, G. J.; CHEN, C. The sound of music: Constructing science as sociocultural practices through oral and written discourse. Journal of Research in Science Teaching, Hoboken, v. 36, n. 8, p. 883-915, 1999.
• KELLY, G. J.; TAKAO, A. Epistemic levels in argument: An analysis of university oceanography students’ use of evidence in writing. Science Education, Hoboken, v. 86, n. 3, p. 314-342, 2002.
• KLEIN, P.; OLSON, D. R.; STANOVICH, K. Structuring reflection: teaching argument concepts and strategies enhances critical thinking. Canadian Journal of School Psychology, Thousand Oaks, v. 13, n. 1, p. 38-47, 1997.
• KUHN, D. Science as argument: Implications for teaching and learning scientific thinking. Science Education, Hoboken, v. 77, n. 3, p. 319-337, 1993.
• KUHN, D.; PARK, S.-H. Epistemological understanding and the development of intellectual values. International Journal of Educational Research, Oxford, UK, v. 43, n. 3, p. 111-124, 2005.
• LARRAIN, A.; SINGER, V.; STRASSER, K.; HOWE, C.; LOPEZ, P.; PINOCHET, J.; MORAN, C.; SANCHEZ, Á.; SILVA, M.; VILLAVICENCIO, C. Argumentation skills mediate the effect of peer argumentation on content knowledge in middle-school students. Journal of Educational Psychology, Washington, v. 113, n. 4, p. 736-753, 2021.
• LAWSON, A. The nature and development of hypothetic-predictive argumentation with implications for science teaching. International Journal of Science Education, Abingdon, v. 25, n. 11, p. 1387-1408, 2003.
• LAZAROU, D.; ERDURAN, S. “Evaluate what I was taught, not what you expected me to know”: Evaluating students’ arguments based on science teachers’ adaptations to Toulmin’s argument pattern. Journal of Science Teacher Education, Abingdon, v. 32, n. 3, p. 306-324, 2021.
• LIN, S.-S. Science and non-science undergraduate students’ critical thinking and argumentation performance in reading a science news report. International Journal of Science and Mathematics Education, Dordrecht, v. 12, n. 5, p. 1023-1046, 2014.
• LYTZERINOU, E.; IORDANOU, K. Teachers’ ability to construct arguments, but not their perceived self-efficacy of teaching, predicts their ability to evaluate arguments. International Journal of Science Education, Abingdon, v. 42, n. 4, p. 617-634, 2020.
• MCKEON, M. In defense of a normative concept of argument. Argumentation, Dordrecht, v. 38, p. 247-264, 2024. DOI: https://doi.org/10.1007/S10503-024-09629-W
  » https://doi.org/10.1007/S10503-024-09629-W
• MASSIE, J. L.; DOUGLAS, J. Managing: a contemporary introduction. 2nd.ed. New Jersey: Prentice-Hall, 1973.
• MIKESKA, J.; HOWELL, H.; KINSEY, D. Do simulated teaching experiences impact elementary preservice teachers’ ability to facilitate argumentation-focused discussions in mathematics and science? Journal of Teacher Education, Thousand Oaks, v. 74, n. 5, p. 422-436, 2022.
• OSBORNE, J.; ERDURAN, S.; SIMON, S. Enhancing the quality of argumentation in school science. Journal of Research in Science Teaching, Hoboken, v. 41, n. 10, p. 994-1020, 2004.
• REISER, B.; DUSCHL, R. A.; SCHWEINGRUBER, H. A.; SHOUSE, A. W. (ed.). Taking science to school: Learning and teaching science in grades K-8. Washington: The National Academies Press, 2007.
• REVEL CHION, A.; COULÓ, A.; ERDURAN, S.; FURMAN, M.; IGLESIA, P.; ADÚRIZ-BRAVO, A. Estudios sobre la enseñanza de la argumentación científica escolar. Enseñanza de las Ciencias, Barcelona, n. ext, p. 1-5, 2005.
• REVEL CHION, A. Hablar y escribir ciencias. In: MEINARDI, E.; GONZÁLEZ GALLI, L.; REVEL CHION, A.; PLAZA, M. V. (ed.). Educar en ciencias Buenos Aires: Paidós, 2010, p. 163-190.
• RUIZ ORTEGA, F. J.; MÁRQUEZ BARGALLÓ, C.; TAMAYO ALZATE, O. E. Cambio en las concepciones de los docentes sobre la argumentación y su desarrollo en clase de ciencias. Enseñanza de las Ciencias, Barcelona, v. 32, n. 3, p. 53-70, 2014.
• RUIZ ORTEGA, F. J.; TAMAYO ALZATE, O. E.; MÁRQUEZ BARGALLÓ, C. La argumentación en clase de ciencias: un modelo para su enseñanza. Educação e Pesquisa, São Paulo, v. 41, n. 3, p. 629-646, 2015. DOI: https://doi.org/10.1590/S1517-9702201507129480
  » https://doi.org/10.1590/S1517-9702201507129480
• SALMON, M. H. Introduction to logic and critical thinking Boston: Cengage Learning, 2012.
• SAMPSON, V.; CLARK, D. B. Assessment of the ways students generate arguments in science education: Current perspectives and recommendations for future directions. Science Education, Hoboken, v. 92, n. 3, p. 447-472, 2008.
• SAMPSON, V.; ENDERLE, P.; GROOMS, J. Argumentation in science education. The Science Teacher, Arlington, v. 80, n. 5, p. 30-33, 2013.
• SHADISH, W. R.; LUELLEN, J. K. Quasi-experimental design. In: GREEN, J. L.; CAMILLI, G.; ELMORE, P. B. Handbook of complementary methods in education research New York: Routledge, 2006. p. 539-550.
• SIMON, S.; ERDURAN, S.; OSBORNE, J. Learning to teach argumentation: Research and development in the science classroom. International Journal of Science Education, Abingdon, v. 28, n. 2-3, p. 235-260, 2006.
• SOYSAL, Y. An exploration of the determinants of middle school students’ argument quality by classroom discourse analysis. Research in Science & Technological Education, Abingdon, v. 41, n. 1, p. 343-371, 2023.
• THOMAS, S. N. Practical reasoning in natural language New Jersey: Prentice-Hall, 1981.
• TOULMIN, S. E. The uses of argument 2. ed. Cambridge: Cambridge University Press, 2003.
• VON AUFSCHNAITER, C.; ERDURAN, S.; OSBORNE, J.; SIMON, S. Arguing to learn and learning to argue: Case studies of how students’ argumentation relates to their scientific knowledge. Journal of Research in Science Teaching, Hoboken, v. 45, n. 1, p. 101-131, 2008.
• WALTON, D. Argument structure: a pragmatic theory. Toronto: University of Toronto Press, 1996.
• ZEIDLER, D. L.; OSBORNE, J.; ERDURAN, S.; SIMON, S.; MONK, M. The role of argument during discourse about socioscientific issues. In: ZEIDLER, D. L. (ED.). The role of moral reasoning on socioscientific issues and discourse in science education Dordrecht: Springer Netherlands, 2003. p. 97-116.
• ZEMBAL-SAUL, C.; MUNFORD, D.; CRAWFORD, B.; FRIEDRICHSEN, P.; LAND, S. Scaffolding preservice science teachers’ evidence-based arguments during an investigation of natural selection. Research in Science Education, Dordrecht, v. 32, n. 4, p. 437-463, 2002.
• ZOHAR, A.; NEMET, F. Fostering students’ knowledge and argumentation skills through dilemmas in human genetics. Journal of Research in Science Teaching, Hoboken, v. 39, n. 1, p. 35-62, 2002.

• Editor:
  Roberto Nardi

Publication Dates

• Publication in this collection
  17 Nov 2025
• Date of issue
  2025

History

• Received
  14 June 2024
• Accepted
  31 May 2025