The Wrong Answer

The transcript runs to ninety-one pages. The first page is a header block: case number, date, location, the names of the two investigators, and the name of the person being interviewed, listed as "witness." On page forty the word "witness" is crossed out by hand and "suspect" is written above it. Nobody initialed the change.

What follows is a composite. It is built from patterns documented in published interrogation research and court records, and it does not come from any single case. The real transcripts this chapter relies on are discussed afterward, by name, with their outcomes. Read this one first, the way the file presents it, and notice what you believe by the bottom of the page.

DET. 1: Nobody thinks you planned this. We know you love her.

FATHER: I do. I loved her. I love her.

DET. 1: So help us understand how it happened. Was it an accident?

FATHER: I don't know what happened. I was asleep.

DET. 2: The evidence says somebody was in that house. The evidence says it was you. We're past whether. We're on why.

FATHER: I don't remember.

DET. 1: Maybe you don't remember because it was an accident. People block things out. That happens.

FATHER: Maybe. I don't know. Maybe it was an accident.

Then comes a page of silence, which the stenographer marks as a pause of two minutes and ten seconds. After it, the father begins to describe what he did.

Ask the three questions. The page says a parent was asked how his child died and, after denying it, gave an account. It misses almost everything about the hours before page forty: how long he had been awake, whether he had eaten, whether he had asked for a lawyer, whether he had been told he could leave. It forces one question, and you have probably already answered it. Why would a man describe a killing he did not commit?

Most readers answer that he wouldn't. Juries have answered the same way for a long time. A confession feels like the end of a file, the one document that does not need corroboration because it came from the person with the most to lose. Investigators have said in interviews over the years that a confession is the best evidence there is, and courts have often treated it that way. If you have spent the last five chapters learning to distrust first assumptions, the confession can feel like relief. Here, at last, is a document where the subject speaks for himself.

Stay on that feeling. The phrase "maybe it was an accident" appears in the composite because some form of it shows up again and again in documented cases. It reads as a crack in a lie, the moment a guilty man lets go. Read it that way and the next forty pages make sense. He was asleep, he blocked it out, and then he remembered. The detectives were patient. They gave him room to tell the truth and he took it.

There is also the matter of who he is. In the earlier chapters you saw how often a child's killer is someone the child knew. That pattern is real, and investigators are right to look closely at the household. A father who was home that night, who can't account for the hours, whose grief looks wrong to the people watching him, has every reason to be interviewed. He may be interviewed again, and possibly for a long time.

Look at the header block again. The interview began at 6:15 in the evening, and the stenographer's last time stamp reads 7:40 the next morning. The page count matches that span. The file does not record whether he slept.

You may notice that you have started reading his grief as a symptom. He cried too much on page twelve and not enough on page thirty. He asked about funeral arrangements, which seems strange, and then asked whether the dog had been fed, which seems stranger. None of this is evidence, and each line of it still pushes a little.

Riley Fox was three years old. She lived with her parents and her older brother in Wilmington, Illinois. On a June morning in 2004 her father, Kevin Fox, woke to find her missing from the house, and he and others began searching. She was found later that day in a creek in a forest preserve a few miles away.

That is as much as this chapter needs to say about how she died.

Investigators looked at the household, as they do. The months passed without an arrest, and the public pressure was heavy. In October 2004 Kevin Fox was brought in for questioning. According to the civil case he later brought and the reporting that followed it, the interrogation lasted about fourteen hours. Over that time he was told that the evidence pointed to him. He also faced the threat of what would happen to him in prison as a man accused of harming a child. Finally he was offered a way to describe her death as an accident. He signed a statement saying it had been one. He was charged with her murder and spent roughly eight months in jail.

Then the lab work came back. Biological evidence from the investigation was tested, and it excluded him. The charges were dismissed in 2005. In 2010 DNA linked the case to a man named Scott Eby, who had no connection to the family. Eby pleaded guilty and was sentenced to life in prison. In 2010 a federal jury found that investigators had violated Kevin Fox's rights and awarded the family a large judgment, which was later reduced.

Now go back to the composite and read "maybe it was an accident" again.

The line has not changed. Your reading of it has. That is the admission this chapter asks of you, and it is uncomfortable, because nothing on the page told you to read it the first way. You brought that reading with you.

The research explains how such a sentence gets made. Many American departments train in some version of the Reid technique, an interrogation method that pairs firm accusation with "themes" that minimize the crime by offering the suspect a less shameful story. In the composite, the line about an accident is the theme. U.S. law has allowed police to lie about evidence since Frazier v. Cupp in 1969, so the statement that "the evidence says it was you" can be legal even when no such evidence exists. Saul Kassin, a psychologist who has studied interrogation for decades, has shown in laboratory work that minimizing themes raise the rate of false confessions along with true ones. They do it by suggesting leniency without ever promising it.

The field data is more sobering. In a 2004 study, Steven Drizin and Richard Leo examined 125 proven false confessions. Many came from vulnerable people, including juveniles and people with intellectual disabilities. Many also came from ordinary adults after long interrogations, a large share lasting six hours or more. The Innocence Project reports that in roughly a quarter to a third of the DNA exonerations it has documented, the wrongly convicted person had made a false confession or incriminating statement. The National Registry of Exonerations reports a lower share across all exonerations and a markedly higher one for people who were juveniles when questioned.

Grieving parents fit this research in a particular way. They have usually not slept, and they are often willing to keep talking long after anyone else would stop, because they believe they are helping find their child's killer. They carry guilt that is real but has nothing to do with the crime: the unlocked door, the window, the night they did not check on her. A skilled interrogator can take that guilt and turn it into a confession. "You blame yourself" can become "you did this" in a single theme. And parents are often told that their grief looks wrong, as if there were a correct way to grieve that investigators could recognize.

Something else changes once a confession is signed, and it is the part of the Fox case that should stay with you longest. Tests that might have pointed elsewhere become tests meant only to confirm. Tips about strangers have less to attach to. In the two years after he was ruled out, the man ultimately convicted had been free the entire time. Every hour the file spent on the wrong person was an hour it did not spend on him.

In your margin, find the first assumption you wrote in the left-hand column. Then draw a line from it to the moment a confession would have closed the file. Count the hours between the two. Those are not pencil hours or pen hours. They are hours the file stopped being read.

One more thing about the ninety-one pages. Before an interrogation reaches the word "suspect," it usually gathers the most careful account anyone will ever take of the family's last ordinary night. That includes what time the doors were locked, who came by, and what the child said before bed. Investigators skim those pages once they have decided where the story ends. Some departments now record interviews from start to finish, because a recording keeps those pages in the file and preserves what a typed summary drops. Many still do not.

The reader who started this chapter could trust a confession. The reader finishing it cannot, not on the confession's word alone, and that is a real loss. It means every confession in every file now needs the same scrutiny as a tip from a gas station. It means a parent's statement deserves the protection you would want for your own if you were awake at hour eleven, being told the evidence says it was you.

The next document in the file is a memo dated years later. It was written by an investigator who was not in the room for any of this. She began by pulling the transcript and reading every page before page forty. On page seventeen, the father describes hearing a car door outside sometime after two in the morning. The detective's follow-up question is blank, and the transcript moves on.