Today I'm going to talk about some serious stuff,

and then hopefully, we'll have a little bit of humor along with it.

I brought some props.

I'm going to need some help from some people in the front row to commit a crime.

(Laughter)

And for the serious part, I'm going to be talking about DNA errors,

forensic DNA errors today.

So imagine

- and I'm only going to ask you to imagine this briefly -

a loved one is hurt or killed in a crime.

Someone's finally arrested, you go through the endless trials,

and someone is incarcerated for the crime.

You think you're done with it.

Twelve years, 20 years later, you hear they got the wrong person.

That's the kind of thing that we are trying to avoid.

And so, in order to do that,

we have to constantly be looking at our techniques

to see how to improve them.

So I'm going to talk about some issues in my own field that I've seen in my work.

I work in all 50 states now,

I have cases overseas: Asia, Africa, Europe.

And some of the things I've seen that I'm concerned about,

that I have raised at professional meetings,

some of it has never been shown before, and I'll show you tonight.

So we're talking about forensic DNA errors.

I don't want to you to think it's all bad,

it's not all bad, it's a great field, it's my life.

And I'm going to break it into two categories:

one is statistical and interpretative errors,

so things where the conclusions are wrong, based on human reasoning and mathematics,

and the other one is contamination, which is kind of easy to understand.

Somebody spills some DNA while it gets transferred,

and then I'll tell you what my lab here in Boise is working on

in the contamination area very briefly.

So we'll have a first slide, I'm going to turn you into the DNA experts.

Get ready for science, here it goes.

This is the easy stuff, this is a single source,

I can teach you guys to do this immediately.

Let's say I decide I'm going to commit a crime with this.

It's clean, it just got out of the dishwasher,

and there was bleach in the dishwasher, so it has no DNA on it.

I touch it, and I go to the front row, and I commit a murder.

The police come, they do the spatter,

they collect some of the blood to identify the victim,

they see my fingerprint, and they --

I always carry a swab.

(Laughter)

I really do always carry --

They take a swab, they swab it, and what's really nifty about these swabs

is you can just knock the ends off of them,

but I'm going to pull it, and put it in here.

I should be wearing gloves etc., etc.

(Laughter)

OK.

Now, the issue is am I the murderer?

Here I am, the suspect,

and all you have to know about DNA is

you get it from your mother and your father.

So that means that every location on the DNA that I'm going to look at

- and I've shown you these are four locations here, in the columns;

sorry that this is a little off -

at every location you've got one number from mom, one number from dad.

I've got one number from my mom, one from dad, I'm the suspect.

Am I a match?

(Audience) Yes.

Greg Hampikian: My students, when I say the word 'match',

have to say 'statistics'.

So, am I a match?

(Audience) Statistics.

GH: At this level,

maybe one in 10,000 people would be included in this match,

maybe one in 2,000, something like that.

Alright, that's a single source, everybody gets it right.

Next.

Here is a mixture.

How do we get a mixture?

Before I commit the crime, a couple of other people I offered--

touch this, maybe three of you.

So now, they come in, and they do the swab, and they get a mixture.

This is what they get: these again are four locations of DNA,

but now, there is a bunch of what we call alleles or sizes in there.

The question becomes, "Am I included in that?"

So, am I still included, or are all of my sizes in there?

(Audience) Yes.

GH: Is it a match? (Audience) Yes.

GH: The statistics is now something like one in 10 or maybe one in 50.

And that's because more people could potentially be included.

Now, you think that this would be an easy concept,

we'd have mathematics that can deal with it,

but when laboratories are given the same data,

they'd sometimes vary in that statistic over ten to the 10.

That's the difference between going to court and say,

"There was 30 dollars of damage to my car,"

and "The damage to my car was greater

than the combined gross domestic product of every country

over the last 100 years" and then sum.

(Laughter)

So the statistics are done differently in different labs,

and they give a very different impression based on how the statistics worked before.

This mixture question becomes very complicated,

and if you give a mixture to ten experts,

you might get ten different interpretations.

I did that.

I took a mixture case from Kerry Robinson,

a man who wrote to me, and who says he's innocent,

and he was convicted in a gang rape in Georgia.

There were three men who raped a woman.

One man was identified,

all of his DNA in the semen was recovered from the victim.

She picked him out of a yearbook.

He's given an option to reduce his sentence:

tell us who one of the other guys is.

He makes up a name, the guy doesn't exist.

They say, "No deal."

Then he says, "Kerry," and he can't remember the last name.

"Robinson," he finally says.

And so they arrest Kerry Robinson, the guy he knew in high school.

Kerry says, "He only gave my name

because he thinks I talked to the cops about him."

But, because he gave a name, his sentence is reduced.

That man, the first man is out of prison right now.

Kerry Robinson is in prison.

I looked at the DNA, I say,

"Kerry Robinson is excluded from that mixture."

I took the same data, me and Itiel Dror my co-author,

and we gave it to 17 analysts at the same lab, another lab

- sorry, we all work in one lab,

but not the first lab that did the DNA -

and we asked them, "Is it a match? Is he excluded?"

- you didn't say statistics -

GH: Is it a match? (Audience) Statistics.

GH: Alright, you got a pass.

Is it a match? Is he excluded? Thank you.

Or is it inconclusive?

We gave it to 17 experts,

all trained, in one lab in North America, all part of a state lab.

We got all three answers.

Not only we got all three answers but only one of the people at that lab

agreed with the lab in Georgia's original conclusion

that Kerry was included.

Next, we got a little attention for that, and new scientists picked it up,

and economists picked it up, but it's still a problem.

All right, I was trying to explain this mixture question to a prosecutor

over the summer, so I had to use some scrabble tiles.

And what I said is a mixture

is like if you put my name and your name, Mr. Prosecutor, in a bowl.

We put two people in, but you can pull lots out.

And so I had to do --

I took our two names and these are the letters for our two names.

Now, we'll get into this question of subjectivity and bias.

Who contributed names to this?

I've told you, I did it.

You might not know how to spell my name.

But Al Gore did too, didn't he?

George Will, Larry King, Mary Kay, Porky Pig, Anne Klein, Henry Miller,

Happy Gilmore, Lina Horne, Papa Gino, Pepe Le Pew, Marilyn Monroe,

Willy Wonka, Rainman, King Phillip, King Lear, King Kong, etc.

The point is these mixtures,

even though we go to court, and they say that is the person a match

(Audience) Statistics.

the statistics can widely, widely, widely vary

to the point where they can become meaningless,

and there is no uniformity in my field for that.

Next slide.

This is just some of the attention we got for that article.

And I'll take the next one.

Here is another problem of statistics.

This is a very unfortunate case.

Mr. Denman became a missing person.

This is from the Albuquerque newspaper, this is his obituary,

he was 49 years old when he left us, and donations can be sent, etc.

There is another article about this later, when they interviewed his brother,

because they found the bones near the house,

and they thought it was a murder.

FBI came in to identify, "This is Donnie Denman,"

with DNA they identified it,

so there is another article looking for people who could help solve the murder.

Next slide.

This is now four months later.

Donnie is not dead.

His friend was putting newspapers down to his parents' car in the garage,

saw that there is an obituary for Donnie Denman,

and that there was a funeral, called Donnie Denman and said,

"They think you're dead," and he said, "Who gave the oration?"

He said, "Your pastor."

He called the pastor, he straightened it out.

This shows you that even though there was a DNA match, right,

a FBI DNA match --

People.

(Laughter)

Even though there was a match--

(Audience) Statistics.

the statistic wasn't that high, this is mitochondrial DNA,

a particular type of DNA, we use on old bones, broken hair etc.

And it doesn't give us those powerful statistics.

Next slide please.

This is our last story of statistics, this was my first case in Taiwan.

And this is for the Taiwan Innocence Project.

They heard from a man "Mr Chen".

So I'm going to show you how to read this.

This is now Y-chromosome DNA,

so most of the columns will just have one number.

I'm sorry you can't see it that clearly,

but this is DNA recovered from semen from the underwear of a rape victim.

She was raped by three men, and these are the alleles or DNA sizes

that were found.

This is Mr. Chen.

Now is Mr. Chen's 15 there? Yes.

In fact, every one of Mr. Chen's alleles is there.

Mr. Chen is arrested, Mr. Chen is convicted,

he loses all three appeals, he writes the Taiwan Innocence Project.

They needed me when they came over to stage, show me the state,

and they say, "What can we do, what's the statistic on this?"

And we try to calculate the statistic, it was pretty strong.

I said, "But there are new tests."

So they went to court in Taiwan, and they got a test

that instead of looking at the number of alleles in the first kit,

added some other alleles, some other markers in this kit to look at.

So they looked at more locations, more powerful statistics.

Even though he was included

with the less powerful statistics,

with the more powerful, - here is Mr. Chen -

his 17 is not in the underwear,

his 13 is not in the underwear.

Next slide.

This is Mr. Chen in March of 2014.

So, it's a statistical problem.

It was a match.

(Audience) Statistics.

And now we have a better kit that looks at more sides,

and so gives us more powerful statistics.

We're going to talk about contamination.

Next slide.

This is a Los Angeles case.