Before the Senate Intelligence community held a public hearing on Russian interference in the 2016 election, Donald Trump tweeted, “James Clapper and others stated there is no evidence Potus colluded with Russia.  This story is FAKE NEWS and everyone knows it!” @realDonaldTrump, 20 March 2017. His tweet highlights several problems. One of them is an imprecise use of language when discussing the nature of evidence. This is a problem that is more widespread than we think.

The interview with Director Clapper to which Trump is referring is below. You will notice Director Clapper was answering a question from Chuck Todd about whether there was any “definitive” evidence of collusion between Russian and the Trump campaign. He answered no. That does not mean there is no evidence; that means he is not aware of any definitive evidence.

The Christopher Steele dossier provides evidence of collusion between the Russians and the Trump campaign. Director Clapper did not include this dossier in his official report because the intelligence community was unable to confirm the information in the dossier. This however does not change the fact that the dossier itself constitutes evidence of collusion. It also does not mean the dossier is not true.

The evidence in the dossier is often described in the press as being “unconfirmed.”  The distinction being used here between confirmed and unconfirmed is often used in the press.  It is based on the idea that a reporter can confirm a story if for example the reporter can get one more top official to say a story is true.  This distinction plays an important role in journalism, but there are areas where it loses its utility.

Many big questions cannot be confirmed with one phone call.  The tapes led to Nixon’s resignation but they did not confirm the truth of the entire Watergate scandal.  The scandal was too big to be confirmed with one phone call or one tape.  The same is true of the Kennedy assassination.  Who would we call to get a confirmation on that story?

The distinction between confirmed and unconfirmed evidence is not generally used in inductive reasoning; evidence is evaluated in terms of the strength of the support it provides an argument: some inductive arguments will be stronger than others based on the evidence they provide.

Deductive arguments, by comparison, seek to prove a conclusion in such a way that it is not possible for the conclusion to be false if the premise or premises are true.  For example, if John is taller than Brian and Brian is taller than Brenda, then John is taller than Brenda.

Inductive arguments can demonstrate that it is highly likely that a conclusion is true. They can for example prove a conclusion beyond all reasonable doubt, a standard used in courtrooms.  But inductive arguments cannot prove a conclusion in such a way that it is not possible for the conclusion to be false if the premise or premises are true.  This level of certainty is what defines deductive arguments.  So in inductive reasoning it is not possible to prove or confirm a conclusion in such a way that it is not possible for it to be doubted. For this reason it can be misleading to describe an inductive argument as being confirmed.

The ranking minority member on the Senate Intelligence Committee, Senator Adam Schiff, has stated we have circumstantial evidence the Trump campaign colluded with the Russians to undermine the U. S. election.  Circumstantial evidence is defined in opposition to direct evidence: direct evidence provides direct evidence to support a conclusion; circumstantial evidence supports a conclusion only with an additional inference.  If you wake up in the morning and see snow on the ground, it provides circumstantial evidence it snowed over the night; if you stayed up the night before and saw snow falling on your yard, it constitutes direct evidence it snowed.

On March 22nd, Schiff said there is now more than circumstantial evidence of collusion between the Russians and the Trump campaign.  Politico reported this discussion:

“I can tell you that the case is more than that,” Schiff told Chuck Todd on MSNBC. “And I can’t go into the particulars, but there is more than circumstantial evidence now.”

When Todd followed up, asking if he had “seen direct evidence of collusion,” Schiff would not say so directly, but insisted that he has seen some “evidence that is not circumstantial” and is worth investigating.

“I don’t want to go into specifics, but I will say that there is evidence that is not circumstantial and is very much worthy of investigation, so that is what we ought to do,” Schiff said.

This distinction between circumstantial and direct evidence can be useful.  But we should acknowledge that we can have direct evidence that does not confirm a conclusion beyond all possible doubt.  Even if we saw snow falling on our yard last night, it could have been because our neighbor was shoveling his roof. This is the nature of inductive reasoning.

In conclusion, it is not accurate to say there is no evidence the Trump campaign colluded with the Russians during the election.  The Steele dossier provides evidence.  It is misleading to describe the Steele dossier as being unconfirmed.  Parts of this dossier have been confirmed, as Senator Schiff describes.  The distinction between confirmed and unconfirmed evidence can also imply a standard of evidence that is beyond inductive reasoning, particularly when it is applied to complex questions.

It is more accurate to say we have evidence the Trump campaign colluded with the Russians to influence the 2016 election.  We continue to gather more.  The argument that can be developed based on this evidence continues to get stronger and stronger.