Argumentation: Parts to Whole, Comparisons, and Correlations

Argumentation: The Study of Effective Reasoning

Commentary: This is a series on effective reasoning as it applies to project management. Using proper argumentation in a project while vetting risk, options, objectives, strategies, and workaround solutions can strengthen a project's performance, improve communications, and develop a sense of unity. Effective argumentations comes down to building the strongest case for a claim. In this series I will be summarizing points made by David Zarefsky in his Teaching Company coursework as well as drawing on other resources.   This series of posts may be reviewed at the Argumentation Series Posts link.

The next six sessions focus on inferences. I'll be compressing them into two postings and try to keep them down in length.  Inferences are the most complex part of an argument and determine the scheme that will be used.  There are six inference patterns that we will consider. 

Reasoning From Parts to the Whole

When reasoning from parts to the whole inferences are made which is a leap in knowledge when information is missing. Because information is missing there is a degree of certainty of a claim given its   evidence. Inferences affect arguments as they are often organized by their patterns of influence. The six inference patterns are example, analogy, sign, cause, commonplaces, and form with some hybrid combinations. This allows us to explore major argument schemes asking some fundamental questions:

• What is the inference and does it work?
• What are the test to determine the strength of an inference?

A warrant from example, that is arguing from parts to the whole, is illustrated in President's Roosevelt's war message on December 8, 1941. he began by listing many examples of Japan's simultaneous attacks on different nations. Then made a claim that Japan has undertaken a surprise offensive throughout the Pacific inferring that what is true of the part applies to the whole. Finally, Roosevelt inferred that Japan had intentions of war.

If the enumeration were complete then the argument would be deductive with 100% certainty. Instead, generalizations are used to make inferences from an example to the whole or general. The inference is that what is true of the part is true of the whole. These generalizations are drawn from one of two patterns; statistical or anecdotal. The statistical is probabilistic and is drawn from a sample of the whole population. Whereas the anecdotal is representative of tests, examples, representativeness, and counterexamples in volume or even the lack of these things. All generalizations should be tested for fallacy of composition.

Classification uses the inference from examples to derive a specific application based on general principles.  This approach is reversed from generalizations. Classifications go from what is true of the whole must be true of the part.  Once again there is a degree of uncertainty in order to make the inference. The effect of classification is to increase the salience of an abstract or general claim by making it more concrete.  The tests for classification are:

• Does the particular member really belong in the general class?
• Are there reasons to think that the particular member is an exceptional case, so that the general principle might not apply?

Any classification also should be tested for the fallacy of division.

Reasoning With Comparisons

One common form of inference is that like things should be treated alike. This is the reasoning behind analogy which can be literal or figurative. Inferences from analogies are based on comparisons and resemblances. Thus, with respect to the topic under discussion the inference will make one thing like the other through induction.

Literal analogies are direct comparisons of objects, events, situations, places, and other items in the same sphere of reality. The inference warrants that if they are alike in the most basic respects then they are alike in the respects under discussion. literal analogies identify parallel cases then derive guidelines for action. Most often literal analogies reason from similar historical situations.

There are two special cases of literal analogies. The judicial analogy is a special case literal analogy. This is used in legal cases arguing that preceding cases have decided the current case. It is also used as a rule of conduct arguing that previous conduct applies to the current situation. The other special case literal analogy is a fortiori. This special case suggests that what is true of the lesser is even more true of the greater and likewise what is false of the greater is even more false of the lesser. This is known 'super-analogy' because it compares things that are largely alike but that differ in magnitude.

Figurative analogies are comparisons between relationships of objects, events, places, and situations rather than comparisons of the things themselves. Typically, the items compared are in different spheres of reality arguing A is to B (known as the theme) as C is to D (known as the phoros). The theme relates to terms of the claim. Whereas, the phoros contains the better known terms of the analogy. The warrant then links the theme to the phoros relating the terms. figurative analogies attempt to clarify or make concrete ideas.

Analogies require careful testing since resemblances are not identities and analogies are never certain. The test is simple. Do the essential similarities outweigh the essential differences between the items being compared? Thus, a false analogy is one that fails this test.

Establishing Correlations

Correlations focus on sign inferences which establish a relationship between two factors in such a way that one con be determined from the other. Sign inferences are not causal and them in their self do not constitute the relationship. These inferences are fallible and depend on probability. They are used to infer the known from the unknown by predicting outcomes and relying on expert judgments.

Sign inferences involve correlations, patterns, occurrences, or changes that vary in relation to each other. The basics are that something can be predicted from the occurrence of another thing. Aristotle distinguished between fallible and the infallible stating that if something were infallible then it would be certain and deductive. Thus, fallible signs inferences are probabilistic having a possibility of being mistaken. The warrant implies a predictable relationship between the variables. Sign inferences assert a predictable relationship between variables but do not account for the relationship. Thus a sign is less powerful than a causal inference. the prototype case or model for a sign relationship is that the surface observations is regarded as a sign of something deeper.

The purpose of a sign segment serves a variety of uses. For example, they infer the known from the unknown. They predict outcomes when it is not necessary to reason the outcomes. They also rely on the judgments of experts or authorities. The test for signs are a series of questions.

• Does the sign and the subject generally appear together?
• Are there any counter signs?
• Can a sign signify other things; opposite or different?
• Is there a basis for the relationship other than mere coincidence?
• Has the relationship been mistakenly regarded as causal?

Commentary: As a project manager, understanding how to reason and communicate is essential to avoiding problems especially in contractual and purchase order agreements. A challenge is to overcome innuendo, ambiguity, and uncertainty ensuring clear and concise communications. In complex projects or where progress elaboration is present justifications, reasoning, and defining needs may be uncertain to some degree. During the initiation phase of a project this problem is often more prevalent.  It is not uncommon for project managers to pull project histories from the PERL database or other databases to give a basis for inferring requirements, objectives, and other project details.  Educated guesses are made and contracts must be written in ways permissive of the range of possible outcomes when certainty is elusive. 

Next week I will condense three more sessions into a single post. Moving this series towards the conclusion.

References:

Zarefsky, D. (2005) Argumentation: the study of effective reasoning. 2nd Ed. the Teaching Company. Chantilly, VA