Claims, Evidence, Arguments and Warrants

In The Craft of Research Booth, Colomb and Williams offer much useful help about how researchers construct arguments that are convincing to the reader. They state that an argument has to have two basic elements, a claim and evidence.

They illustrate this with the sentence

Claim: It must have rained last night

the evidence that might be given is:

Evidence The streets are wet

But as they point out, the streets could be wet even if it did not rain; street cleaners could have made the street wet, or a hydrant could have opened and flooded the street, etc. The claim that it has rained is ordinarily accepted with the evidence that the streets are wet, because this is the most common reason for streets being wet. In other words, someone is claiming a cause and effect relationship, and relying on our general knowledge of the world for acceptance of their claim.

They also say that a claim must be substantive and contestable. By substantive they mean it must be important enough to warrant our consideration, otherwise we consider it trivial; by contestable, they mean that there must be a way to (con)test it. And to support it, evidence must be reliable and relevant. Reliable means of course that we can rely on the evidence; or we can replicate the experiment ourselves and see that nothing has been fudged; relevant means that we can see how it is related to the claim.

An example of a non-substantive argument might be:

  1. Claim: Cows give milk

    with the irrelevant evidence that:

  2. Evidence: Grass is green

In research of all sorts, researchers look for causative effects: they want to say that X causes Y in all kinds of areas where explanations are necessary, such as:

In most research, readers are not content to accept another researcher's assertion that a certain thing is the cause; they want to know that the researcher has considered all options, has examined all variables, and has excluded or rebutted all possible other causes or objections. Otherwise, the reader will say that the claim is unwarranted in other words, not sufficiently provided with warrants that we have come to expect as part of the reasoning of an argument. B,C and W show the warrants as part of the superstructure of an argument:

Warrants, they say, "let[..] you connect a particular claim to particular evidence validly. They are like explanations that we know to be true (or think we know to be true) and generally seem sound, given the way things generally work. But often they are not true, or have counterexamples. As B, C and W say:

Even experienced researchers can take their warrants too much for granted, because they are hidden in the theories that guide their research, in the definitions of their words, even in the metaphors they use.
In chapter 9, they show how not to take warrants for granted, how to decide whether one is true, whether one is valid, and when you should make them explicit. (Particularly beginning in sec. 9.2 ff.)

But even when warrants are valid, researchers follow certain other procedures to make sure they are conducting valid research. In many fields of research, it is customary to anticipate objections to their arguments, and rebut them (with evidence etc.) before making their own case, or stating their final conclusion. They refer to these as Qualifications and indicate that there are four types: as shown in the next figure:

The four types are:

  1. Rebut mistaken objections to your evidence or warrants. (Offer counterevidence and counterproofs to the 'mistaken' objections that others have raised or might raised: "The research on the advantages or disadvantages of bilingual education conducted in the early 1920's tested only children from economically disadvantaged backgrounds, and never tested their intelligence in their own native languages.")

  2. Concede objections that you cannot rebut. ("I recognize that X's claim may have some validity, but it does not appear to have more validity, or be more strongly supported, than mine, nor does it invalidate my final conclusions.")

  3. Stipulate the conditions that qualify your evidence or limit the application of your warrant: ("I did not obtain evidence for my claim at temperatures below zero, or under (X) conditions, nor am I convinced that it would work at very high temperatures.")

  4. Stipulate the degree of certainty of your evidence, warrant, or claim: ("The results I obtained are totally reliable within the age group 16 to 29; in higher age groups, the incidence of the variable declines, and is totally absent in populations born before World War II, or who are not native to the five boroughs of New York."

A close reading of their chapter on warrants is probably warranted, and following their suggestions in this area will probably greatly strengthen the arguments you may be making in your research.

For practice, look at this document and analyze the press reports as to the validity of their claims, evidents, and the warrants they provide.,
last modified 10/29/01