Wednesday, April 1, 2009

Summary Findings: Argument Mapping (3 Stars Out Of 5)

Note: This post represents the synthesis of the thoughts, procedures and experiences of others as represented in the 12 articles read in advance of (see previous posts) and the discussion among the students and instructor during the Advanced Analytic Techniques class at Mercyhurst College on 1 APR 2009 regarding Argument Mapping
specifically. This technique was evaluated based on its overall validity, simplicity, flexibility and its ability to effectively use unstructured data.


Definition:
Argument mapping (AM) is an analytic modifier that can be used to examine the logic behind the development of a particular conclusion and/or hypothesis. The product of AM is a visual representation (typically a box-and-line diagram) of the reasons that support and oppose the claim. Constructing a visual depiction of a complex argument reduces the level of abstraction in evaluating a decision.

Strengths:
Provides Audit Trail
Breaks down prose arguments and visualizes arguments for decision makers.
Shows strengths and weaknesses of argument
Improves critical thinking
Reduces cognitive bias and blindspots

Weaknesses:
Not estimative in nature.
The method does not explicitly seek out contradictory evidence.
The language specific to argument mapping is not uniform from scholar to scholar.
Dependent on the quality of presentation of original argument
The technique does not consider the impact of deliberately deceptive information

How-To:
* Must locate a central claim for a position (this is the conclusion of an argument)
--conclusion indicators include: therefore, thus, so, hence
* extrapolate true and logistical reasons and objections which support or refute the central claim.
--rewrite statements as individual sentences
--Reasons must answer the question: "How do we know that [insert claim] is true/warranted?"
* include all premises for reasons - leave nothing to be implied
--premise indicators include: since, because, for, given that
* With objections, list any possible refutations underneath the objection
* Form a 'warrant' - a statement that justifies the step from a reason to a claim
--warrants reveals an arguer's core implicit beliefs
--used to scrutinize the soundness of an argument
--used to accept or reject controversial arguments
*Use arrows to indicate which premises or co-premises support claims; or premises refute oppositions.
*indicate premises that need to be combined in order to support a conclusion, and premises that are each seperate reasons to believe a conclusion
*Use a color-coding scheme (if possible) to help visualize the arguments
--Supporting claims - green
--opposing claims - red
--nuetral claims - gray or no color

Experience:
We attempted to argument map a one page NY times editorial. The comments below were generated as a result of that exercise.

Argument mapping is difficult to synthesize from prose that is not written in a BLUF (Bottom Line Up Front) format.

The ability to effectively argument map is too dependent on quality of the articles given.

People will disagree on structure of argument and levels of relevance of data.

Argument mapping is not appropriate to use to develop an analytic estimate, rather it can be best implemented as a modifier to validate an already determined estimate.

Argument mapping would be an excellent modifier to test the validity of a conclusion/forecast. Would be an excellent supplement to ACH - suggest conducting an ACH to find a claim that is most likely, then test that claim for validity using Argument Mapping. This will help to reduce cognitive blindspots and biases, as well as test the reasoning behind the conclusion.

Extrapolating the main claim of the argument was easier once all of the premises supporting it were laid out.

The original format of the 'argument" really matters when it come to the ease or difficulty of applying argument mapping to the text.

1 comment:

  1. I’m not sure that the blog was really expecting outside commentary, but here is a comment on the stated weaknesses of argument maps, from a proponent.

    1. If by estimative you mean quantitative, then argument maps are not necessarily estimative. However, doing them right involves evaluating them, and the evaluation phase requires at least qualitative estimates of strength. Yanna Rider argues that a great deal of the value of mapping comes from the evaluation phase. Also, Gheorghe Tecuci has developed a detailed argument mapping system (Disciple) that does use numerical estimates, and will propagate uncertainty. It currently uses intervals and simple combination rules (like average, min, max), but it does make argument maps quantitative.
    2. Although the method does not seek out contradictory evidence, it does require you to show that the conclusion follows from the premises. A partisan proponent is free to assume strong premises to help ensure the conclusion, but they will be visible.
    3. Granted, the language is not uniform.
    4. The fourth point is incorrect. Argument mapping is also used to generate arguments, not just to represent existing prose.
    5. Since argument mapping is not automated, granted the “technique” does not consider deliberately deceptive information. The technique does not consider anything, but analysts can use it to do so. Where deception is a distinct possibility, it that should be represented as a branch in the tree: “Joe might have been lying,” or “The plans could be a diversion.” (Tecuci’s system uses Schum-style schemas to automatically consider deception.)
    6. Argument maps are indeed difficult to extract from prose, because most prose is sloppy, as is the reasoning behind it. However, it is possible that some of the demands of mapping are unnecessary – apparently decision maps are easier to extract from prose, or create fresh.
    People do disagree about argument structure. Training alleviates some of that, but some remains. I would argue that the remaining uncertainty is not an artifact of argument mapping, but a lack of reasoning clarity that mapping has revealed.

    I do not understand the claim that argmapping is inappropriate for analytic estimates. Estimates are now given in prose, and they are argued by reasoning from evidence. Why can that reasoning not be captured in an argument map?

    ReplyDelete