Problems With Abductive Reasoning in a Child Abuse Investigation

A Child Protective Services social worker conducting a child abuse investigation relies on abductive reasoning to come to a conclusion once the investigation is complete. Historically, abductive reasoning (or abduction) referred to the reasoning used in generating hypotheses. In our day, abduction more often refers to the explanatory reasoning used in justifying hypotheses. In fact, due to this latter meaning of the word, abduction is often described as: Inference to the best explanation.[i]

When employing abduction, a social worker conducting a child abuse investigation (or a researcher [scientist], diagnostician [doctor], or factfinder [juror]) will conclude the investigation with a set of evidence or facts. The investigator takes for granted the truth that the evidence the investigator has is incomplete—that there exists other evidence, outside the investigator's realm of knowledge, that is relevant to the investigator's inquiry. From the evidence she has, the investigator generates a set of hypotheses that might explain the evidence (which we might call alternative explanations). The investigator then chooses among the hypotheses generated the one which she believes best explains the evidence.[ii]

The troubles with abduction in CACI child abuse investigations and CACI grievance hearings include the following.

First, the investigator conducting a child abuse investigation does not know what the evidence is that she does not have. She does not know what she does not know. (For instance, she may not be aware of the evidence that proves or suggests this was a false allegation of child abuse.) The missing evidence could be of such a character that, were it known, could generate entirely knew hypotheses, could render some hypotheses meaningless, or could alter the strength of an existing hypothesis.

Second, the child abuse investigation relies on the investigator to come up with the hypotheses, and the investigator may have limited knowledge of the relevant evidence. If one admits that the person who knows the most facts should be the person who generates explicatory hypotheses, and one further admits that there may be other individuals who know more about the facts pertinent to the investigation, then it follows that the investigating social worker, then it follows that the investigating social worker may not be the best person to be generating candidate hypotheses.

Third, the child abuse investigation relies on the investigator to generate hypotheses, and the investigator's own unconscious biases may affect which evidence is gathered, what emphasis to place on evidence gathered, and what interpretation to give to the evidence gathered.

Fourth, the child abuse investigation relies on the investigator to generate hypotheses, and the investigator may unconsciously replace the evidence gathered with her own interpretations of the evidence, and thus may end up evaluating her interpretations of the facts rather than the facts themselves. Evidence by itself doesn't mean anything until it is interpreted, and an investigator will interpret it as she pleases. Two investigators can have the same evidence before them and use it to prove the exact opposite conclusion, based on the interpretation they give it.[iii]

Fifth, abduction presupposes the notion of candidate explanations. This may be a false presupposition in cases which allow for no explanation. Worse, this presupposition could force an investigator to generate an unfounded hypothesis where the facts allow for no explanation at all.

Sixth, abduction presupposes the notion that there is a best explanation, i.e., that various hypotheses can be generated and that one of them stands out as the best. In other words, abduction fails to leave place for the scenario where the hypotheses are of equal explanatory value. It may be that the evidence -; any evidence that could ever possibly be available to any investigator -; is unable to favor any one of the hypotheses over any other. In this situation, the investigator would not be warranted in believing any particular one of the hypotheses. (This is the underdetermination argument in the philosophy of science.)

Seventh, the phrases "candidate explanation" and "best explanation" do not have straightforward interpretations. What makes a hypothesis the best one? Do theoretical virtues such as simplicity[iv], generality, and coherence determine the best one? Instead, is the best candidate determined only in probabilistic terms (i.e., how likely it is to be the correct one vis-à-vis [or as compared to] the other candidate explanations)?

Eighth, abduction presupposes that the social worker conducting the child abuse investigation is capable of identifying the best hypothesis that explains the evidence.

Ninth, abduction presupposes an investigator conducting the child abuse investigation whose only goal is to identify the best explanation of the evidence, when in fact that investigator may have an ulterior (even if subconscious) motivation. Such an ulterior motivation might be to protect a child whom the investigator deems vulnerable and for whom the investigator feels a sense of sympathy. Another ulterior motivation might be that the investigator feels an affinity with one of the witnesses (say, an accusing mother who claims to be a victim of spousal abuse) and desires to help or sympathize with the child's mother.

Tenth, abduction, at best warrants an inference only to the probable, or even mere approximate, truth of the best explanation. This runs at odds with the goal of the investigative process and the evidentiary grievance hearing held after the investigation. For both, the goal is, or should be, to identify truth. While the goal of abductive reasoning is merely to identify the best explanation, the goal of a CACI grievance hearing (and, hopefully, the investigation) is to identify the correct explanation.

If one can't identify the correct explanation through abductive reasoning, then the conclusion of the child abuse investigation has no merit and should not be relied upon as a basis for a CACI listing. A conclusion that is only probably approximately true is not a reliable one and should not be the basis for a substantiated finding of child abuse and a CACI listing.

The abductive process only takes an investigator conducting a child abuse investigation to the best explanation as compared to the other set of candidate explanations. However, the conclusion must "also come out as being best in comparison with any other hypotheses that we might have conceived (but for lack of time or ingenuity, or for some other reason, did not conceive)."[v] Otherwise, the hypothesis the investigator relies on is merely "the best of a bad lot." (van Fraassen 1989, 143). From a conclusion that a hypothesis is closer to the truth than any of the other hypotheses generated, it does not follow that the chosen hypothesis itself is more likely to be true than not.

That is the final point: that the conclusion the social worker who conducted the child abuse investigation may not be the correct one. In the CACI grievance hearing, the social worker has to prove that her conclusion is more likely true than not. One way to fight her conclusion, is to point out any gaps or errors in her abductive reasoning, as described above.

[i] The ideas and commentary presented throughout this article come from: Abduction (Stanford Encyclopedia of Philosophy)

[ii] Scientific philosopher Karl Popper, famous for his falsification theory, probably considers abduction pseudoscientific: Abduction does not, for example, employ a null hypothesis. In scientific pursuit, truth cannot be proven or discovered; one can only disprove false theories. In contrast with truth, certainties are subjective. At most, abductive reasoning can provide an investigator with a degree of certainty less than 100%, and that degree of certainty in the conclusion which the investigator has will not likely be shared by another investigator. Thus, pseudoscience is identified by the following: "The non-scientific theories could not be falsified. They were not testable in a legitimate way. There was no possible objection that could be raised which would show the theory to be wrong." [See Karl Popper on The Line Between Science and Pseudoscience - Farnam Street (fs.blog)].

[iii] Think Karl Popper again, who was a proponent of the idea that every investigator interprets facts in light of the knowledge and observations that that investigator already holds: No investigator starts an investigation with a blank slate. An investigator does not interpret with a completely open mind, but while already having presuppositions.

[iv] "Lombrozo (2007) shows that, in some situations, people tend to grossly overrate the probability of simpler explanations compared to more complicated ones." (from the Stanford Encyclopedia, citing Lombrozo, T., 2007. "Simplicity and Probability in Causal Explanation," Cognitive Psychology, 55: 232-;257.

[v]Abduction (Stanford Encyclopedia of Philosophy)