Taxonomy: Logical Fallacy > Informal Fallacy > Equivocation > Redefinition

Subfallacy: No-True-Scotsman Move


In July 1993, the Commonwealth Fund released the results of a telephone survey of 2,500 women, designed and carried out by Louis Harris and Associates. The Commonwealth and Harris investigators took their questions directly from the Gelles and Straus survey…. [T]he Harris/Commonwealth survey concluded that as many as four million women a year were victims of physical assaults…. But the most interesting finding of all, and one entirely overlooked by the press, for it did not harmonize with the notes of alarm in the Harris/Commonwealth press releases, was the response the poll received…about the most severe forms of violence. Gelles and Straus had estimated that these things happen to fewer than 1 percent of women. According to the survey sample, the percentage of women who had these experiences was virtually zero; all respondents answered 'no' to all the questions on severe violence. This finding does not, of course, mean that no one was brutally attacked. But it does suggest that severe violence is relatively rare.

So where did the four million figure for physical assault come from? … Clearly the interpreters of the Harris/Commonwealth poll data were operating with a much wider conception of 'abuse' than Gelles and Straus. Looking at the "survey instrument", we find that they had indeed opened the door wide to the alarmist conclusions they disseminated. … To arrive at the figure of four million for physical abuse, the survey used the simple expedient of ignoring the distinction between minor and severe violent acts, counting all acts of violence as acts of abuse. Five percent of women they spoke to said they had been "pushed, grabbed, shoved, or slapped"; they were all classified as victims of domestic violence and added in to get a projection of four million victims nationwide. … If a couple has a fight, and she stomps out of the room (or yard), and he grabs her arm, this would count as a violent physical assault on her.

…As for journalists and the newscasters, their interests too often lie in giving a sensational rather than an accurate picture of gender violence, and they tend to credit the advocacy sources. Better four million or five than one or two. … And all the better, too, if the media's readers and viewers get the impression that the inflated figures refer not to slaps, shoves, or pushes but to brutal, terrifying, life-threatening assaults.1



To redefine a term is, of course, to assign it a new meaning. It is not necessarily fallacious to give a term a new meaning, and it is often done to produce technical terms, but there is always a danger of slipping back into using the term in its old meaning out of habit. So, one may start out reasoning with the term using its new meaning in the premisses, then fall back into using it in its familiar meaning in the conclusion. Thus, one may accidentally end up equivocating on the redefined term, which is why Redefinition is a subfallacy of Equivocation―see the Taxonomy, above.

There are two types of redefinition, depending on whether the redefined meaning has a wider or narrower extension than the original meaning: Low Redefinition


High and low redefinition are not mutually exclusive categories, as it is possible for a redefinition to be both high and low. This happens when the redefinition applies to some things that the original definition did not, and fails to apply to other things that the original did. For example, if we redefine "bird" as "flying animal", then bats would be "birds" (low), while flightless birds would not (high).

It is even possible for the extension of a redefined term to be disjoint from the extension of the original meaning. This often happens when an existing word is redefined as a technical term. Fortunately, such technical terms are not likely to result in equivocation because the term's two meanings are too different to be confused. Confusion is most likely to occur when the slippage between meanings is subtle.

Analysis of the Example:

The Harris/Commonwealth poll was based on a low redefinition of "physical abuse" which produced a large number of cases. "Physical abuse" is a vague concept, and any statistical study of it will have to define it in some precise way in order to be able to count instances of physical abuse. Arm-grabbing and pushing may count as "physical abuse" in some situations, so that it makes sense to include questions about them in the poll. However, such a broad definition is a logical boobytrap which could lead to the conclusion that there are more cases of severe violent assaults on women than indicated by the poll.


  1. Christina Hoff Sommers, Who Stole Feminism? (1994), pp. 196-198
  2. Antony Flew, How to Think Straight (1998), Section 3.7