Apples Ergo Oranges: The Argumentative Use of Comparisons

The main purpose of this article is to gain a better understanding of analogical arguments. To do so, I will adopt a contrastivist approach and analyse them as part of a broad family of ways of arguing that rely on a comparison between two elements to support a claim about one of them. I will call this ‘comparison argumentation’ and propose some criteria to highlight the distinctive features of analogical arguments. This approach aims to address two recurrent issues in the literature on the subject. The first one is what we can call ‘the definition problem.’ In argumentation theory there seems to be agreement that analogical arguments are based on some kind of comparison. Trudy Govier, for example, claims that "arguments by analogy draw a conclusion about one thing on the basis of a comparison of that thing and another" (Govier 1985, p. 350). Douglas Walton, in the same vein, writes that "an argument that proceeds on the basis of a comparison of two similar cases is called an argument from analogy" (Walton 1989, p. 307). And Frans H. van Eemeren, Rob Grootendorst and Francisca Snoeck Henkemans propose the following scheme for argumentation based on analogy:

Y is true of X,

because Y is true of Z,

and Z is comparable to X (van Eemeren, Grootendorst and Snoeck Henkemans 2002, p. 99).

These approaches, while not wrong, are too broad. They encompass two traditionally opposed accounts of analogies. On the one hand, there are those that place emphasis on the similarity of properties. The more aspects two objects have in common, the more likely it is that they also have other aspects in common.

[Case 1]

"The moon resembles [395] the earth in being a solid, opaque, nearly spherical substance, appearing to contain, or to have contained, active volcanoes; receiving heat and light from the sun, in about the same quantity as our earth; revolving on its axis; composed of materials which gravitate, and obeying all the various laws resulting from that property. And I think no one will deny that if this were all that was known of the moon, the existence of inhabitants in that luminary would derive from these various resemblances to the earth, a greater degree of probability than it would otherwise have" (Mill 2009, p. 394).

In this case, the likelihood of life on the moon is derived from the "various resemblances" between the moon and the earth.^{Footnote 1} The problem is that this account confuses "being analogous with being similar" (Juthe 2005, p. 7). Analogy, according to other authors, has to do with a proportion or parallelism of relations:

"Analogy ought to be distinguished from direct resemblance, with which it is often confounded [...]. Analogy being a "resemblance of ratios," that should strictly be called an argument from analogy in which the two things (viz., the one from which, and the one to which, we argue) are not, necessarily, themselves alike, but stand in similar relations to some other things" (Whately 1963, p. 91).

Following this idea, elsewhere (Alhambra 2022) I have argued that arguments by analogy can be seen as based on a comparison of argumentative relationships. Let’s consider an example:

[Case 2]

"When people tell him they don’t fear surveillance because they have nothing to hide, [Edward] Snowden says he tells them: "Arguing that you don’t care about privacy because you have nothing to hide is like arguing that you don’t care about free speech because you have nothing to say."" (Paul Schrodt, "Edward Snowden just made an impassioned argument for why privacy is the most important right," Insider, 15/09/2016)^{Footnote 2}

Unlike the arguer in case [1], Edward Snowden does not compare the properties of two objects, but rather the relationship between two sentences, marked by the connectors ‘because.’ In other words, what he compares are arguments –specifically the relationship between their premises and conclusions.^{Footnote 3} I will call them "target argument" (Fig. 1) and "source argument" (Fig. 2).

Thus, Snowden’s point is that the target argument is incorrect, because it parallels the source argument –which is assumed to be so. In this sense, we have an argument whose premise and conclusion are, in turn, other arguments, i.e. a meta-argument (see Woods and Hudak 1989, p. 127). As we shall see, an analogy thus understood may be used not only to attack, but also to defend the target argument –which in turn may be much more complex, involving not only arguments, but also combinations of them.

Plausible though it may seem, this account has a significant drawback: it seems too restrictive.^{Footnote 4} Some scholars have argued that there are arguments which do not involve a comparison of relations –argumentative or otherwise– and still deserve the name of ‘arguments by analogy’ (e.g., Mill). So, the question arises: Should we restrict the definition of arguments by analogy to arguments based on a parallelism of relations or should we extend it to cover arguments based on any kind of comparison? My aim is to advocate a middle ground position. On the one hand, I will defend a restrictive definition and use the term only to refer to arguments based on a comparison of relations –in the sense outlined above. But on the other hand, I will favour a more inclusive approach by showing that analogical arguments are part of a wider family of ways of arguing that rely on a comparison to pose a reason.

This move is intended to be useful with respect to a second issue, ‘the evaluation problem.’ In argumentation theory there are at least two general views on this matter. On the one hand, some scholars have argued that analogical arguments presuppose a general principle or rule that accounts for the similarities of the things compared and guarantees the conclusion. If this were so, the assessment of an argument by analogy would come down to checking whether that principle is acceptable and applicable to the cases at hand. "One might dispute this argument in two ways: by disagreeing with the principle [...]; or by claiming that the two cases don’t fit under the same principle and thus that the analogy is flawed" (Waller 2001, p. 202). This approach, however, has a major issue: it makes the analogy irrelevant. The conclusion of the argument can be deduced from the principle and the information about the target.^{Footnote 5} Based on this, others have contended that analogical arguments do not presuppose any principles or rules, but operate on a case-to-case basis. And so does their assessment:

"We can evaluate the argument without raising the issue, just by sticking to the cases at hand. We can point out relevant differences between A and B and show how they undermine the conclusion without addressing the U[niversal]-claim as such" (Govier 1989, p. 148).^{Footnote 6}

Elsewhere (Alhambra 2023), I called these approaches ‘generalist’ and ‘particularist,’ respectively, and argued for a particularist one. The challenge with this approach was to specify what those "relevant differences" consist of. To do so, I argued, one first has to specify what kind of comparison the arguer is appealing to. And this is where the article enters the picture. The criteria for distinguishing types of comparisons that I propose here could be useful in this matter, as they specify what arguments by analogy are based on, thus helping to identify relevant differences. In the last section, I will show how this might be the case. But first, I will explain what these criteria are (Sect. 2) and illustrate them by analysing cases of comparison argumentation in the press (Sect. 3).

2.1 Background^{Footnote 7}

The thesis, then, is that arguments by analogy are part of a family of ways of arguing that rely on a comparison between two elements to support a claim about one of them. The aim is to provide criteria that both highlight the distinctive features of analogical arguments and show the feasibility of other ways of arguing based on comparisons. This approach is not new though. One of the first to put the issue in these terms was André Juthe. In (Juthe 2005), he claims that what characterises analogy, as opposed to mere resemblance, is that it involves a parallelism or correspondence of relations. "Just having the same properties would make the object similar but not analogous" (Juthe 2005, p. 7). The following example is considered,

[Case 3]

The Porsche and the Chevrolet are both in the 40,000ドル price range, and the Porsche is of excellent quality. Therefore, the Chevrolet is probably also of excellent quality,

Here, as in case [1], the properties of two objects are compared in order to support a claim about one of them, but there is no parallelism of relations, so it cannot be classified as an argument by analogy.^{Footnote 8} Juthe, however, does not deny that case [1] and case [3] are arguments; what he rejects is that they are arguments by analogy. In fact, he considers them as part of a broader category to which the latter also belong: "Arguments that refer to property similarity, relational similarity (analogy) and structural similarity would then be different subtypes of the generic type: argument by similarity" (Ibid., p. 7). The problem is that this suggestion is not developed. Juthe does propose a relational definition of arguments by analogy (see Ibid, pp. 10–15) but does not explore their kindship with other similarity arguments.^{Footnote 9}

Another precedent is the pragmadialectical classification of argumentation schemes. As is well known, pragmadialecticians distinguish three types of argumentation: symptomatic argumentation, causal argumentation, and comparison argumentation (see van Eemeren, Houtlosser and Snoeck Henkemans 2007, pp. 138–154). In the last one,

"the argumentation is presented as if there were a resemblance, an agreement, a likeness, a parallel, a correspondence or some other kind of similarity between that which is stated in the argument and that which is stated in the standpoint" (van Eemeren and Grootendorst 1992: 97).

Although ‘comparison’ and ‘analogy’ sometimes seem to be used interchangeably, this definition makes it clear that there are different "kinds of similarity," including parallelisms. In a recent work, van Eemeren and Garssen have also acknowledged that there may be cases where the comparison operates on arguments, i.e., cases where the argumentation is presented as if there were a parallel between an argument which is stated in the reason and an argument which is stated in the standpoint (van Eemeren and Garssen 2020, p. 273). The question is, then, what is the place of these arguments within the general category of comparison argumentation?

Van Eemeren and Garssen distinguish two subtypes: descriptive comparison argumentation and normative comparison argumentation. In the first one, the arguer extrapolates a property from the source to the target based on a list of commonalities. "Because the two things, persons, groups, institutions, events or situations that are compared have a series of properties in common, they are assumed to share also another property claimed in the standpoint" (van Eemeren and Garssen 2020, p. 264). As can be seen, this is quite similar to Mill’s example in case [1] and Juthe’s argument in case [3]. In normative comparison argumentation, on the other hand, "it is claimed that a something mentioned in the standpoint belongs to the same category as something mentioned in the argumentation advanced in its support, and that the former should be treated in the same or a similar way as the latter" (Ibid., p. 268). This is where we find arguments by parallel reasoning, along with arguments that appeal to the rule of justice or to the principle of reciprocity (see Ibid. pp. 272–273).

The problem with this account is that it seems incompatible with the second objective that I mentioned in the Introduction. According to van Eemeren and Garssen, varieties of comparison argumentation differ in terms of both the standpoint and the comparison. Whereas in the first variety the standpoint is a factual statement, usually a prediction, in the second it is a claim about how the target should be treated. As for the comparison, in the first variety the standpoint is based on a series of similarities between the cases considered, but in the second it is because they "belong to the same category." The problem is that this could foster a generalist account of normative comparison argumentation and, hence, of arguments by parallel reasoning. If both cases belong to the same category, one might argue, the transfer of acceptability from the argumentation to the standpoint would be based, not on the comparison, but on the category itself. This, as we have seen, turns arguments by analogy into deductions in disguise, something that has been highly contested (see Govier 1989; Guarini 2004; Juthe 2019; Bermejo-Luque 2014; van Laar 2014; Stevens 2018 or Alhambra 2023). To avoid this problem, I will adopt van Eemeren and Garssen’s notion of ‘comparison argumentation’ but use other criteria to distinguish some varieties.^{Footnote 10}

Another precedent for my approach is Hubert Marraud’s treatment of so-called ‘arguments of comparison.’ According to this author, the general scheme of these arguments is A is P, so B is P, because B is comparable to A with respect to C. As can be seen, this is quite similar to the pragmadialectical scheme for ‘analogy argumentation’ (see above p. 2). The differences lie in how varieties are distinguished.

"[argument of comparison] are classified according to the generality of the comparison: overall or in a certain aspect; and according to the proportion: equal (A is P so B is P, because B is as C as A), more (A is P so B is P, because B is even more C than A), and less (A is P so B is P, because B is slightly less C than A)" (Marraud 2021, p. 156 –translation is mine).

Marraud also considers cases where the elements compared are themselves arguments: "according to the things compared, we distinguish arguments of comparison from meta-arguments of comparison" (Marraud 2016, p. 45 –translation is mine). We therefore have three criteria: specificity, degree, and subject of comparison. The first one is related to how the comparison is presented, so I will leave it aside here. The latter is already in Juthe’s proposal and in van Eemeren and Garssen’s. What is new with respect to them is the degree of the comparison. This idea allows Marraud to consider more subtle comparisons and include in the category, among others, a fortiori arguments. The drawback is that Marraud does not consider other uses of comparisons, such as explanations or illustrations. Given that, as (Waller 2001, p. 200) points out, it is quite common to confuse explanations and justifications by analogy, this can be an issue. To solve it, I will add a third criterion: the purpose of the comparison.

2.2 The Classification

The first criterion, then, is the subject of the comparison. This aims to capture the idea that the reason posed by an analogical argument relies on a parallelism of relations. I will distinguish two varieties based on what the subject of the comparison is. On the one hand, a comparison may be about attributes of objects (e.g., "Maria is smart" or "Javier is Galician") and relations between objects (e.g., "Maria has a better CV than Andres" or "Javier is taller than Robert"). In cases [1] and [3], for example, the arguer compares attributes of objects, while the next case it is about relations between objects:

[Case 4]

John is taller than Andrew and weighs more than Charles; Peter is taller than Andrew, so he is likely to weigh more than Charles.

On the other hand, a comparison may also operate on relationships between statements, which I have called ‘argumentative relationships.’ In (Alhambra 2024), I distinguished two types: intra-argumentative relationships and inter-argumentative relationships. The former are those between the reason put forward by an arguer and that for which that is a reason (i.e., the relation between the premises and the conclusion of a given argument). This is usually marked by expressions such as ‘thus,’ ‘therefore,’ ‘consequently,’ ‘because,’ ‘so,’ and the like. Inter-argumentative relationships, on the other hand, are those between two or more reasons. These are usually marked by conjunctive expressions such as ‘and,’ ‘in addition to,’ ‘on the other hand,’ etc., or adversative expressions such as ‘but,’ ‘although,’ ‘nevertheless,’ ‘it is true that..., but it is no less true that...,’ and so on. Case [2] involve relations of the first type; cases [10] and [11] in Sect. 3.2.2 are based on relations of the second type.

The second criterion refers to the relative position of the elements in the comparison. When we say that two things are alike, we can present them in at least two ways. First, we can say that they are on a par with respect to a common feature or scale. This can be expressed using phrases such as A is like B with respect to x, A is as x as B, or A resembles B in x. The comparison between the moon and the earth in case [1] or the parallel between surveillance and free speech arguments in case [2] are examples of it. And second, the elements compared may be presented in different levels, saying, for example, A is more/less than B with respect to x, A is almost as/just a little more x than B, A is much more/less x than B, and so on. "Ana is smarter than Juan", "my argument is better than yours" or "if it is forbidden to step on the grass, it is even more so to pull it up" are examples of this.

Finally, the purpose of the comparison is meant to highlight the distinction between comparison arguments, on the one hand, and explanation or illustration by comparison, on the other. This is based on those accounts that conceive of the distinction between explanations and justifications as a pragmatic distinction. According to them, to tell an explanation from a justification we have to consider the attitudes of the agents towards the subject of the discussion (i.e. the claim or the explanandum), and the agent’s purpose in giving reasons (see, e.g., Govier 2017 [1987], pp. 242–244; Snoeck Henkemans 2001, pp. 231–243; Mayes 2010, pp. 92–95, or Marraud 2021, pp. 10–13). In a justification, the arguer presents something as acceptable (reason) in defence of something that has been put into question (claim), whereas in an explanation something more or less acceptable (explanans) is presented to account for something about which there is no doubt (explanandum). The main purpose in both cases is to show someone that there are reasons for something else, while the secondary purpose is to persuade someone of something by means of reasons or to make them understand something, respectively (see Ibid., p. 12).

2.3 Some Varieties

Based on these criteria, I will distinguish six varieties of comparison argumentation: arguments by resemblance, argumentation by analogy (which I divide into argumentation by parity of reasons and argumentation by parity of structures), degree arguments, a fortiori argumentation, and explanations and illustrations by comparison (Table 1).

In this section, I will analyse some cases of comparison argumentation. Almost all of them are excerpts taken from newspaper articles. The search has been guided by expressions typically linked to comparisons, such as ‘this is analogous to,’ ‘analogy,’ ‘it is as if,’ ‘in the same way as,’ ‘by that logic,’ ‘it would be like saying/arguing/claiming,’ ‘for the same reasons,’ ‘all the more reason,’ ‘how much more so,’ or ‘it is almost as.’ The methodology consisted of entering these expressions into search engines such as Google, Bing or Yahoo, and collecting the information. Following this procedure, 80 cases were gathered. From this corpus, I selected a small group of cases according to the criteria of text type, clarity, length, and need for contextual information. Short excerpts from newspaper articles that required little or no additional information were the priority.

3.1 Arguments by Resemblance

Arguments by resemblance rely on a comparison between attributes of objects or relations between objects to favour the claim that one of them (target) have a property that the other (source) is claimed to have. So, they are arguments (criterion 3) in which the elements compared are presented at the same level (criterion 2) with respect to one or more of their properties (criterion 1). The scheme is C because A, given that A is similar to A’ and if A’, then C’. Let’s take a look at some cases:

[Case 5]

"The big question is whether people like Kerner are just one-time Republican voters, or if they’ll become permanent Republicans. Whether people like Tory Williams can be convinced to keep voting for a party that, until a few months ago, she never imagined voting for. [...] In other words: Are there DeSantis Democrats elsewhere in America waiting to vote for the man in 2024?

These voters are not all that dissimilar to the Reagan Democrats who fuelled the Republican’s 1980 White House victory. Like the Reagan Democrats, the DeSantis Democrats feel condescended to, abandoned by the progressive elites who bankroll Democratic candidates and shape the party’s agenda. Then, like now, inflation was out of control. Then, like now, the leadership in Washington seemed tired, out of ideas. Then, like now, the country seemed adrift." (Olivia Reingold, "The Rise of the DeSantis Democrats," The Free Press, 12/31/2022).^{Footnote 11}

In the first part of the excerpt, journalist Olivia Reingold raises the question of whether the trend among some Democrats in Florida to vote Republican might spread to the rest of the country and lead to the victory of the then-favourite Republican candidate, Ron DeSantis. After raising the issue, she goes on to compare the political situation in 2022 to that of 1980, when many Democrats across the country voted for the Republican candidate. Although Reingold does not state it explicitly, the fact that she presents this comparison right after raising the issue suggests that she takes a positive stance on it. The question is then, why would Democrats across the country vote for DeSantis? And the answer seems to be that many Democrats across the country feel condescended to and abandoned by the progressive elites; inflation is out of control; leadership in Washington is out of ideas, and the country seems adrift (Fig. 3).

The question now is: Why is that alleged fact a reason to think that many Democrats will vote for DeSantis in 2024? And here is where comparison enters to the picture: because something similar happened in 1980 with Ronald Reagan’s presidential race. What Reingold seems to be doing here is to extrapolate the alleged causal relationship between the political situation and the outcome of the 1980 elections (source) to the present situation (target), and to do so she relies on the similarity between them. This idea can be represented in the diagram by adding a second argument which conclusion account for the step from the premises to the conclusion of the former argument (Fig. 4).

The same idea can be seen in the following example.

[Case 6]

""The outer solar system is probably replete with moons that could have liquid water oceans on them, and a subset could have geothermal and water-rock interactions on the bottom," says Chris German, an oceanographer at the Woods Hole Oceanographic Institution, who is co-leading a NASA-funded initiative called Network for Ocean Worlds (NOW). Why do those characteristics matter? "Everywhere that has those on our planet gets colonized by microbial life," German says." (Rebeca Boyle and Juan Velasco, "The Six Moons Most Likely to Host Life in Our Solar System", Scientific American, 01/05/2023)^{Footnote 12}

The hypothesis appears in the title of the article: there are six moons in our Solar System that most likely host life. But why is so? Well, because those moons that could have liquid water oceans on them and a subset could have geothermal and water-rock interactions on the bottom. But why is that a reason to believe that there is probably life in them (or "Why do those characteristics matter?")? And the answer is a comparison with what happens on earth. What makes those features support the claim for the existence of life in those moos is that they are related to the existence of life on Earth, as the arguer makes clear in the last sentence.

As can be seen, case [5] and case [6] are very similar. Both favour a prediction about a case by relating it to another case with which it shares certain features. Let us now look at a somewhat different case. In the following excerpt, journalist Bob Peterson responds to those who defend that voucher schools are public schools.

[Case 7]

"Voucher schools, often referred to as "choice" schools, are private schools that receive taxpayer money that pays for tuition. To argue that a private school is "public" merely because it receives public tax dollars is like arguing that Metro Mart is a public grocery store because it accepts food stamps" (Bob Peterson "Wisconsin Public Schools Held to Higher Standards than Choice or Voucher Ones," Milwaukee Journal Sentinel, 27/11/2023)^{Footnote 13}

Here, too, the argument is based on a comparison, but in a different way. First, no prediction is made, but it is claimed that two ways of arguing deserve the same evaluation. Second, the justification works differently. What voucher schools and Metro Mart have in common is that the considerations put forward to justify the claim that they are public institutions are similar. In other words, the reasons in favour of those positions are alike. Thus, the argument rests on a parallelism of argumentative relations: A is (or is not) a reason for B, because C is (or is not) a reason for D, and A is to B what C is to D. This is precisely an argumentation by analogy.

3.2 Argumentation by Analogy

What sets argumentation by analogy apart from arguments by resemblance is, therefore, that it relies on a parallelism of argumentative relationships (criterion 1). As mentioned, these relationships have to do with the reasons given for a certain position and may be of two types, intraargumentative and interargumentative. This results in two varieties of argumentation by analogy: argumentation by parity of reasons and argumentation by parity of structures.

3.2.1 Argumentation by Parity of Reasons

Argumentation by parity of reasons relies on the comparison of intraargumentative relationships, i.e., the relation between what is presented as a reason by an arguer and that for which that is a reason. This is precisely what happened in cases [2] and [7].

[Case 8]

"The Biden administration then filed an emergency appeal at the Supreme Court, urging the court to reinstate the rule. Lawyers for the government argued that unassembled "ghost guns" should be treated like guns that come fully assembled.

"Every speaker of English would recognize that a tax on sales of ‘bookshelves’ applies to IKEA when it sells boxes of parts and the tools and instructions for assembling them into bookshelves," Solicitor General Elizabeth Prelogar wrote in the government’s emergency appeal" (Betsy Woodruff Swan, "Divided Supreme Court allows ‘ghost gun’ rule to take effect," Politico, 08/08/2023)^{Footnote 14}

This is an excerpt from an article about the Biden administration’s attempts to regulate so-called ‘ghost guns,’ kits of unassembled parts used to make homemade firearms. The lawyers argument is that since regular, assembled guns are strictly regulated by federal law, unassembled ‘ghost guns’ should be regulated by federal law too (Fig. 5).

But Solicitor General Elizabeth Prelogar goes on and makes a comparison: It is reasonable to think ("every speaker of English would recognise") that IKEA kits for assembling bookshelves should be taxed, just as fully assembled bookshelves are. This seems a parallel case (Fig. 6).

But what is the function this second argument? Although there is no explicit indication, the order of the sentences and the context suggest that Prelogar is trying to justify her first argument by means of the comparison. We can paraphrase her point as follows: just as IKEA bookshelves kits are taxed by the same laws as fully assembled bookshelves, unassembled ghost guns should be regulated by the same laws as fully assembled guns. We can depict this idea using the conditionals associated with both arguments (Fig. 7).

But what is then an ‘analogy’? As such, at least three things may be understood here: (1) the comparison itself, which in the diagram appears next to the ‘so’; (2) the justification of the conditional associated with the target that relies on that comparison (i.e., the whole diagram); and (3) the argumentative operation in which that justification is embedded. In this last sense analogy can be seen as a substitute for Toulminian warrants. The ‘step’ from the premises to the conclusion of a given argument can be explained or justified either by appeal to a general rule (i.e., a warrant) or by comparing it with a similar case. (Fig. 8) represents this idea.

I will call (1) ‘analogy’ and (2) ‘argumentation by analogy.’ I do not have a specific term for (3), so when I want to refer to it, I will say it explicitly.

This distinction makes it easier to accommodate cases in which analogy is used, not to support, but to attack the reason put forward in a given argument. In this cases, the parallelism is the same, but instead of asserting the source in the premises, it is denied. We then speak of arguments by counter-analogy or refutation by parallel argument (see Govier 1985; or Juthe 2009). This is precisely what happened in cases [2] and [7]. Here is another example:

[Case 9]

"J. Brauner (Letters, Jan. 12) attempts to justify the genocide in Chechnya on the grounds that "Chechnya has been part of Russia for very long. It cannot legitimately break away unilaterally." This is analogous to saying that a woman who has been married for a long time cannot leave her husband, and that he has every right to beat her senseless if she as much as declares her intention for a divorce" (Danylo H. Struk, "Letters to Editors," The New York Times, 01/17/1995)^{Footnote 15}

In the first sentence J. Brauner’s argument is presented: since Chechnya has always been part of Russia, it cannot unilaterally declare its independence –if it does, Russia has the right to prevent it by force. Right after that, the arguer, Danylo H. Struk, presents a parallel case using the expression "this is analogous to saying": since a woman has been married for a long time, she cannot leave her husband –if she did, the husband has the right "to beat her senseless." Even though there is no explicit indication, it is quite reasonable to think that Struk presents this argument as blatantly wrong. His point is that Brauner’s argument is wrong (i.e., it does not pose a reason worthy of consideration), because it parallels the abusive husband’s argument, which is –obviously– wrong.

3.2.2 Argumentation by Parity of Structures

Argumentation by parity of structure differs from argumentation by parity of reasons in the type of relations on which the comparison operates.

[Case 10]

"Isabel González does not believe that fires are profitable for anyone. If they activate any kind of sector of the economy, it would be the sector in charge of putting out forest fires. But it is "absurd to demonise them, because at the end of the day they do their job," and she suggests that it would be like saying that companies that manufacture vaccines or masks are profiting from the pandemic. They make profits, yes, but they are necessary in case of an emergency" ("El lucrativo negocio de quemar el monte [The lucrative business of burning the forests]", Canarias te quiero, 28/08/2022).^{Footnote 16}

The extract presents the position of Isabel González, vice-dean of the Official College of Forestry Engineers and coordinator of Together for Forests. The issue under debate is whether the companies in charge of extinguishing forest fires are doing something illegitimate in carrying out their activity. González’s position is that they are not: although these companies make profits by putting out fires, and that would be a reason to criticise them, they carry out a necessary task that someone has to cover, so it makes no sense to demonise them. In support of this argumentation, the journalist appeals to the case of companies that manufacture vaccines or masks: in the same way that it does not make sense to demonise companies that manufacture vaccines or masks, because, although they make profits, they carry out a necessary activity, it does not make sense to demonise companies that put out fires either.

What is interesting here is that the subject of the comparison is not single arguments, as in cases [7], [8] and [9], but weighings of reasons, or more specifically, refutations. A ‘refutation’ is an argumentative operation in which an arguer weighs two or more reasons favouring incompatible conclusions, gives more weight to one of them and asserts its conclusion (see Marraud 2025, pp. 104–108). Refutations are usually marked by connectors such as ‘but,’ ‘however,’ ‘although,’ ‘although it is true that,’ and so on. Using ‘but’ as a standard weighing connector, we can diagram Gonzalez’s point as in (Fig. 9).

The aim of González’s argument is to justify the attribution of weight expressed by the connector ‘but’ by comparing it with the case of companies manufacturing vaccines or masks, which can be depicted as in (Fig. 10).

Isabel González’s argumentation by analogy can be depicted by a diagram with the source refutation (Fig. 9) as the premises and the target refutation (Fig. 9) as the conclusion of the argument (Fig. 11).

The analogy in this case is more complex than in the previous cases. Instead of comparing the relation between what is presented as a reason and that for which that is a reason, the arguer compares the relation between the reasons presented in the source and target cases. In other words, if in argumentation by parity of reasons one tries to justify the ‘so’ in the target argument by comparing it with the ‘so’ in the source argument, in argumentation by parity of structures it is the ‘but’ in the target structure what the arguer tries to justify by comparing it with the ‘but’ in the source structure.

With regard to this variety, two things should be borne in mind. First, although here I am presenting an analogy of refutations, reasons can be combined in other ways, leading to more complex comparisons. In principle there is no theoretical constraints to this complexity, although it is likely that in practice the more complex the less frequently it is used. Second, analogy can also be used to attack a weighing of reasons –or other structure. In the following excerpt Gabriel de Prados responds to a user who, in connection with the murder of Marta del Castillo, had criticised the government for not changing the law that would allegedly allow the suspects to be released if the victim’s body was not found.

[Case 11]

"A different matter is the legal changes you mention. Although this is off the point, I would like to make it clear that it is not the fairest thing to change laws based on "social alarm." In Almeria we have had an example with the Gran Plaza de Roquetas which, while recognising its manifest illegality, is not being pulled down because of the social alarm of the people who work there; that would be like saying that although the sale of drugs is manifestly illegal, given the social alarm of the number of people who make a living from it, no action should be taken either by the police or the courts." (Gabriel de Prados, "Conciencia frente a intereses [Awareness against interests]," Foro-Ciudad.com, 01/06/09)^{Footnote 17}

In this case, Grabiel de Prados rejects the idea that legal changes should be made on the basis of social alarm. To justify his position, he makes a comparison between the ruling against the demolition of the Gran Plaza shopping centre in Roquetas de Mar (Almería, Spain) and the hypothetical case of the sale of drugs. Unlike González, de Prados argues that if we accept that source weighing is a bad one, we must accept that target weighing is also so, because they are parallel.

3.3 Degree Arguments

Degree arguments, like arguments by resemblance, rely on a comparison between attributes of objects or relation between objects to defend that one of them (target) has a property that the other (source) has. The difference is that in degree arguments the elements compared are presented at different levels with respect to the common property or scale. The general scheme is as follow: A is y, because A is more/less than B with respect to x, and B is y. As Dr Frasier Crane would say: "if less is more, just think how much more more would be!"^{Footnote 18}

[Case 12]

"Devices should have a timetable for use and should preferably be given to adults at bedtime. Their [children’s] judgement and willpower are not yet formed enough to resist the avalanche of information and dopamine that they receive through these devices. If it is sometimes difficult for us adults to draw the line, how much more so for them" ("La generación somnolienta [The drowsy generation]," El Universal, 08/05/2020).^{Footnote 19}

The first sentence contains the main thesis: devices should have a timetable for children’s use and should preferably be given to adults at bedtime. After a full stop the arguer gives her reason: children find it difficult to put a limit on the use of these devices. Here one might ask, "Where do you get that from?" And the answer is a degree argument (Fig. 12).

Thus, we would have a chaining in which an argument of degree justifies the reason given for taking a certain course of action (Fig. 13).

The author of the following excerpt, Germán Martínez Martínez, makes a similar point in relation to the staging of plays originally written in Latin:

[Case 13]

"One thing that is interesting to me about these productions is to observe the reactions of the audience when the curtain comes down: invariably I find some gestures of frustration, which are usually verbalised, when the text has been totally or partially preserved. People of different profiles, native speakers of the language, who discover that they are alienated from what is represented. Sometimes it may be a question of acting deficiencies, but even then, the problem is related to the language of the 16th and 17th centuries. How much more will this be the case with Latin –a language that today has only artificial life, as the official language in a religious community embedded in the Italian capital– and when the authors wrote their verses before the time of Christ or just at the beginning of the Christian era." (Germán Martínez Martínez, "El problema de los clásicos: los latinos de Hernández Busto [The problem of the classics: the Latins of Hernández Busto]," Etcétera, 28/05/21)^{Footnote 20}

In this case the claim is that works written in Latin will not be well received by contemporary audiences. The reason is that Latin is a much more alien language to modern audiences than 16th- and 17th-century English, and plays written in 16th- and 17th-century English are not well received by contemporary English-speaking audiences.

3.4 A Fortiori Argumentation

A fortiori argumentation is like argumentation by analogy, in the sense that in both cases argumentative relations are compared. The difference is that, while in argumentation by analogy the source and the target are on a par, in a fortiori argumentation they are on a different level with respect to the common characteristic or scale. Roughly speaking, the scheme is as follows: argument A puts forward a good reason, because argument A’ puts forward a good reason, and the reason put forward by argument A is even better than the reason put forward by argument A’. Let us consider an example:

[Case 14]

"The mausoleum of Rocío Jurado was "with flowers and clean" on Monday 31 st October, as always, because "it doesn’t need to be a special date" to look like this, but, logically, "all the more reason" on the eve of All Saints’ Day, thanks to Asociación RJ La Más Grande and her fans" (F. J. F., "Fans de Rocío Jurado se topan con el cementerio de Chipiona cerrado [Fans of Rocío Jurado find the Chipiona cemetery closed]", Diario de Cádiz, 02/11/22).^{Footnote 21}

In this excerpt, Rocío Jurado’s fans do not say why the singer’s mausoleum should have flowers and be clean, but they do say that it is not because it is the eve of All Saints’ Day. Let us imagine that the main reason is that she would have wanted it that way (Fig. 14).

But if the fact that it is the eve of All Saint’ Day is not a reason for the conclusion, what role does it play in the argument? This is what in the theory of normative reasons is called a modifier (see Bader 2016 and Marraud 2019): "a consideration that, without being a reason in itself, alters the weight of an argument" (Marraud 2021, p. 34). Modifiers are divided into intensifiers and attenuators. The fact that it is the eve of All Saint’ Day is an intensifier that reinforces the reason that Rocío Jurado would have wanted her mausoleum to be with flowers and clean. We can represent the modified argument by adding the intensifier next to the ‘so’ (Fig. 15).

Rocío Jurado’s fans argue that this argument is good, because the previous one was good and this one makes a stronger case (hence the expression "all the more reason"). One issue to bear in mind is that modifiers are contextual considerations that, although relevant for the evaluation, are not part of the argument. Therefore, a fortiori argumentation scheme is the following: argument A, in situation S, raises a good reason, because in situation S’ the same argument raises a good reason, and in S there is an intensifier that in S’ did not exist (i.e., A is stronger in S than in S’).^{Footnote 22} Here another example:

[Case 15]

"Ethics –from the Greek term ethos, custom, character, conduct– is the branch of philosophy that deals with the study and regulation of human behaviour and actions, in the light of morality. Its knowledge and validity are indispensable in any society, insofar as ethics refers to the way of acting –right or wrong– of those who are part of it, so that relations of all kinds between them obey principles and minimum rules that make coexistence possible.

[...] Now, if respect for the principles of ethics is demanded in the activity of ordinary citizens, it is even more so in the case of public servants, and it is essential in the highest positions within the State" (José G. Hernández: "Recuperemos la ética [Reclaiming ethics]," El Nuevo Siglo, 15/11/22)^{Footnote 23}

In contrast to the previous case, José G. Hernández does make explicit his reasons for defending the need to recover ethics, namely that ethics has to do with following principles and minimum rules that make coexistence possible. If this is already a good argument with respect to ordinary citizens, "it is even more so" when it comes to public servants and high-ranking officials. Again, it is claimed that an argument is good in one context because it was considered to be good in another context where it was weaker.^{Footnote 24}

3.5 Explanations and Illustrations

As mentioned in Sect. 2.2, the difference between explanations and justifications, insofar as it is based on the attitudes and purposes of the arguer, is pragmatic in nature. In an explanation by comparison the purpose is to make someone understand something that is accepted to be the case. This criterion is transversal to the previous ones, and in principle we could find explanations by resemblance, analogy, degree, and so on.

[Case 16]

"To give a better sense of what is unusual under its appearances of banality, about the overturning that consists in taking a point of view on one’s own point of view and, thereby, on the whole set of points of view in relation to which it defines itself as such, I would like to call to mind the novel by David Garnett, A man in the zoo, of which I have often thought with respect to the approach I adopted in Homo Academicus. It tells the story of a young man who quarrels with his girlfriend during a visit to a zoo and, in despair, writes to the director of the zoo to offer him a mammal missing from his collection, man – himself. He is then put in a cage, next to the chimpanzee, with a sign saying: "Homo sapiens. MAN. This specimen, born in Scotland, was presented to the Society by John Cromartie, Esq. Visitors are requested not to irritate the Man by personal remarks"" (Bourdieu 2003, p. 284).

Pierre Bourdieu refers here to his proposal of a participant objectivation, which consists of including the sociologist’s point of view as the subject of study in the research. To exemplify how paradoxical it is to adopt a point of view about one’s own point of view, Bourdieu mentions A Man in the Zoo. This short novel tells the story of John Cromartie, a 27-year-old man who, after arguing with his girlfriend and being called animal, writes a letter to the zoo director offering to be part of the Great Apes exhibition: "It may seem unimportant at first sight, since the collection is formed for man to look at, and study", Cromartie argues, but exhibiting a man in it "would lead to a thousand interesting comparisons being made by visitors for whose education the Gardens do in a large measure exist" (Garnett 1924, p. 10). Unlike the previous cases, the comparison here is not presented to justify, but to exemplify something that has not been challenged, that is, participant objectivation. Here’s another example:

[Case 17]

"Most people experience some degree of limerence, said Dr. Poerio, but it’s problematic when it’s uncontrollable. Dr. Poerio uses the analogy of a person whose mind has been hijacked. "It interferes with your ability to have meaningful, real-world relationships because you are sustaining a relationship in your mind."" (Amanda McCracken, "Is It a Crush or Have You Fallen into Limerence?" The New York Times, 27/01/24)^{Footnote 25}

In this case the comparison between the subject suffering from ‘limerence’ ("a state of overwhelming and unexpected longing for emotional reciprocation from another human, known as a limerent object (L.O.), who is often perceived as perfect but unavailable" ibid.) and a person who has been hijacked serves to explain how this mental state may interfere with the subject’s social life.

Of course, the purpose of an explanation may appear in combination with other purposes, as in the following example:

[Case 18]

"Like a tender mother, whose weak

son tearfully implores

for some food that she knows

will only serve to make him sick.

But that merciful love prevents her

from seeing the harm that she does

by giving him what he wants –

so she indulges him, feeding his torment.

So, my sick, crazy thinking,

which foolishly yearns for you – I’d like

to cease its fateful seeking

but it begs for you with endless whine,

so much so that I’m yielding,

forgetting its death and even mine."

(Rivers 1991, p. 63 –translation is mine)

Although the comparison serves here to explain an idea very similar to that of the previous example, it is reasonable to think that the poet, Garcilaso de la Vega, seeks not only to make the audience understand the point, but also to create a certain aesthetic pleasure. Sometimes what is sought is amusement:

[Case 19]

"I’m the Michael Jordan of lazy sports analogies" (Conan O’Brien, Twitter, 23/09/19).^{Footnote 26}

Illustrations, on the other hand, are intended to clarify a general rule or category. This term was coined by Perelman and Olbrechts-Tyteca, who distinguish between examples, in which the comparison establishes a rule, and illustrations, which reinforce adherence to an already accepted rule (see Perelman and Olbrechts-Tyteca 1971, pp. 357).

[Case 20]

"The problem begins once the distinction between sovereignty and power is blurred. Sovereignty is about who decides legitimately on behalf of a people – whereas power is the capacity to impose these decisions on the outside world. Iceland is a tiny country. But to claim that Iceland’s sovereignty is illusory because it is too small to have much power is like arguing that a poor person with no political clout might as well give up her right to vote" (Yannis Varoufakis, "The European Union is disintegrating – but leaving is not the answer" The Guardian, 05/04/2016).^{Footnote 27}

"Sovereignty and power must be distinguished" is a principle that, according to Yanis Varoufakis, applies to the case of a country, such as Iceland, and which the poor person’s comparison is intended to illustrate. Thus, in both explanations and illustrations, the purpose of the comparison is to make the interlocutor understand something; the difference is that, while illustrations clarify a general rule or category, explanations operate on particular cases.

In short, regarding the subject of the comparison, cases [1], [3], [5] and [6] (arguments by resemblance) are similar to cases [12] and [13] (degree arguments), and differ from cases [2], [7], [8], [9], [10] and [11] (argumentation by analogy) and cases [14] and [15] (a fortiori argumentation). As for the degree of the comparison, cases [12] and [13] resemble cases [14] and [15] but differ from cases [1], [3], [5] and [6] as well as from cases [2], [7], [8], [9], [10] and [11]. Finally, all of them differ from cases [16], [17], [18], [19] and [20] (explanations and illustrations) in terms of the purpose of the comparison.

This classification aims to better understand arguments by analogy and their role in the practice of arguing. But here understanding might mean at least two things here: to recognise and analyse these arguments when confronted in the wild, as it were, and to being able to evaluate them. In what follows, I will show how the criteria of the subject, the degree, and the purpose of the comparison may be relevant to this task.

As I mention, I conceive of arguments by analogy as a form of metaargumentation: the arguer claims that a target argument is good because it parallels a source argument, which is supposed to be so. I have also contended that to know whether an argument by analogy poses a good reason no general principle, rule or category is needed. With (Govier 1989) I think that analogies can be assessed simply by pointing out relevant differences and showing how they undermine the comparison. This is where the criteria might be useful. To know what a difference is, we must first consider what is being compared in the analogy (criterion 1). It makes no sense to say that two objects are dissimilar in terms of their properties (e.g., colour, shape, texture, and so on) if what is being compared are relationships, and vice versa. The same applies to the degree of the comparison (criterion 2). If the source and target cases are not presented as parallel or equal, and we discover a difference that shows that the source is weaker than the target, that need not mean that the argument as a whole is bad. And finally, whether the analogy is presented for the purpose of justifying, explaining, clarifying, or embellishing something (criterion 3) is relevant, for the simple reason that we do not evaluate those actions in the same way. Arguments by analogy, as I understand them, are based on a comparison of argumentative relationships, which are presented as parallel, and for the purpose of justifying the target argument.

With this in mind, we can outline what the evaluation of an argument by analogy would look like. Let us take case [8] as an example. If my interpretation is correct, and we would like to know whether Elizabeth Prelogar’s argument puts forward a good reason, we have to check if there are relevant differences between the cases compared. Since Prelogar’s point is that the reason posed by the ‘ghost guns’ argument (target) is appropriate because it is parallel the reason posed by the IKEA bookshelves kits’ argument (source), we have to look for consideration that shows that is not the case. In the terminology of argument dialectics, we are looking for ‘rebuttals.’ ^{Footnote 28} A good example of this can be found in the same article:

[Case 21]

"Attorneys for the challengers countered that, since the gun kits don’t always come with all the parts necessary to make guns, the IKEA analogy doesn’t work" (Betsy Woodruff Swan, "Divided Supreme Court allows ‘ghost gun’ rule to take effect," Politico, 08/08/2023).

This consideration, if true, is relevant because it attacks the relation between the premise and the conclusion of the target argument and, hence, shows that both arguments are not really parallel. We can depict this idea by adding the counterargument at the same level as the aspect to be attacked, i.e., the ‘so’ (Fig. 16).

In fact, attorneys for the challengers of ‘ghost gun’ regulation also use an argument by analogy to support their criticism, specifically an argument by counter-analogy:

[Case 22]

"A better analogy would be to a ‘taco kit’ sold as a bundle by a grocery store that includes taco shells, seasoning packets, salsa, and other toppings, along with a slab of raw beef," they argued. "No one would call the taco kit a taco. In addition to ‘assembly,’ turning it into one would require cutting or grinding and cooking the meat — and until that was done, it would be nonsensical to treat it as food and the equivalent of a taco." (Ibid.)

With respect to argumentation by parity of structures, something similar happens. In this case, as we saw, the comparison is between inter-argumentative relations (i.e., relations between reasons) and what is being defended is that the target raises a reason not only worthy of consideration, but relatively strong. Again, a relevant difference will be a consideration that shows that this is not the case, either in the source or in the target. Let us return to case [10] and imagine that someone responds to Isabel González by saying: "right, I see your point. However, what you have not considered is that these companies benefit from state subsidies, something that does not happen with companies that manufacture vaccines or masks." Here the hypothetical critic points out a consideration that supposedly occurs in the target, but not in the source. This consideration is relevant because it affects the weight of the reasons considered: if that were the case, we would have more reasons to defend that companies that put out fires are doing something wrong. We can represent this by adding another refutation (Fig. 17).

In short, the criteria for distinguishing varieties of comparison argumentation may be useful because they highlight features of argumentation by analogy that are relevant for their evaluation. Of course, this is only a hypothesis. The next step would be to collect real examples, such as cases [21] and [22], and check whether arguers themselves evaluate arguments by analogy in this way. Only after this could a proper evaluation procedure be proposed, for example in terms of dialectical profiles.^{Footnote 29}

In this article I have distinguished some varieties of ‘comparison argumentation.’ This way of arguing consists of supporting a claim about one thing on the basis of a comparison of that thing with another. The criteria have been the subject, the degree, and the purpose of the comparison. I have distinguished six varieties: arguments by resemblance, argumentation by parity of reasons, argumentation by parity of structures, degree arguments, a fortiori argumentation, and explanations and illustrations by comparison. This approach has two purposes: to clarify what arguments by analogy are and, on this basis, to contribute to design a process of evaluation. An important aspect has been the analysis of cases. I have selected a relatively large corpus of examples and analysed them based on argumentative connectors and other conventional indicators of comparisons.