WEBVTT

00:00:02.313 --> 00:00:08.247
In addition to gaining insights from observing people,

00:00:08.247 --> 00:00:13.296
it’s also valuable to interview them — ask them about their experiences directly.

00:00:13.297 --> 00:00:20.298
The first step in setting up an interview is deciding who it is that you’re going to interview.

00:00:20.298 --> 00:00:22.345
It may seem obvious

00:00:22.345 --> 00:00:29.838
but the most important thing is to get people who are representative of the target users of your system —

00:00:29.838 --> 00:00:32.912
who are the people who are going to be using your system.

00:00:32.912 --> 00:00:34.488
Talk to them!

00:00:34.488 --> 00:00:41.657
Now they may be current users of a similar system if you’re creating a better something,

00:00:41.657 --> 00:00:47.137
you’ll likely find current users and you’ll want to learn what they care about,

00:00:47.137 --> 00:00:49.236
what problems they see.

00:00:49.236 --> 00:00:51.724
They might also be non-users.

00:00:51.724 --> 00:00:56.575
So, one reason that you might be building your technology

00:00:56.575 --> 00:01:01.427
is to broaden the set of people that can do a certain task

00:01:01.427 --> 00:01:04.765
and in that case you’ll want to talk to the non-users.

00:01:04.765 --> 00:01:09.812
Learn about what barriers you see, learn about what their goals are.

00:01:09.812 --> 00:01:13.762
Say, for example, you’re designing a lecture support system.

00:01:13.762 --> 00:01:16.735
Who would you interview in this case?

00:01:17.381 --> 00:01:23.379
There’s a lot of stakeholders: You might pick the teacher.

00:01:23.379 --> 00:01:27.667
You might pick the students. You might also pick the teaching staff.

00:01:27.667 --> 00:01:35.959
Maybe the department administrators that are responsible for making sure the grading is handled properly.

00:01:35.959 --> 00:01:37.452
Maybe even the parents.

00:01:37.452 --> 00:01:39.580
It’s important to get different types of users.

00:01:39.580 --> 00:01:44.635
So, you might have both the freshman and a PhD student.

00:01:44.635 --> 00:01:49.160
You might have a domestic student and an international student.

00:01:49.160 --> 00:01:51.784
Somebody that’s in major and out of major.

00:01:51.784 --> 00:01:53.963
Stronger and weaker students.

00:01:53.963 --> 00:01:57.041
There are a lot of different kinds of users.

00:01:57.041 --> 00:02:02.928
You can’t get them all, but you’ll want to find several different people

00:02:02.928 --> 00:02:07.630
who are representative of major user groups, and talk to them.

00:02:08.691 --> 00:02:10.688
How can you find these people?

00:02:10.688 --> 00:02:17.356
Well, one great strategy, if you’re in the US, is to use Craigslist.

00:02:17.371 --> 00:02:20.529
You can also reach out through friends and family.

00:02:20.529 --> 00:02:24.558
Your social network is a powerful tool for finding people to interview.

00:02:24.558 --> 00:02:30.387
That doesn’t just mean or even primarily mean “interview your sister”;

00:02:30.387 --> 00:02:33.569
it means “ask your sister whom she might know”

00:02:33.569 --> 00:02:39.029
or “ask your friend to put you in touch with people they might know.”

00:02:39.029 --> 00:02:42.547
It’s quite likely that you’ll need to pay people.

00:02:43.055 --> 00:02:45.558
In the Bay Area, on Craigslist,

00:02:45.558 --> 00:02:51.477
somewhere in the land of 50 to 100 bucks for an interview seems to be the going rate.

00:02:51.477 --> 00:02:57.088
This gets cheaper if you can have less specialized users.

00:02:57.088 --> 00:03:03.350
Also, the amount that you’ll need to pay people depends on what they believe is going to be used for.

00:03:03.350 --> 00:03:07.362
If you can convey that in some way you’re making the world a better place,

00:03:07.362 --> 00:03:11.721
people may be willing to offer their time for free, even.

00:03:11.721 --> 00:03:15.716
On the other hand, if they believe that something will be used for a profit,

00:03:15.716 --> 00:03:19.711
they’ll likely want a fair wage for their time.

00:03:19.711 --> 00:03:25.486
If you’re a skunkworks project and don’t have enough money to really pay people,

00:03:25.501 --> 00:03:29.413
one thing that you can do is offer a token of appreciation.

00:03:29.413 --> 00:03:36.537
People often appreciate a small thoughtful gift that they can actually use, like a gift certificate or some sort,

00:03:36.537 --> 00:03:38.931
than they do cash.

00:03:38.931 --> 00:03:42.547
For example, a couple of years ago I’ve spoken at a local high school

00:03:42.547 --> 00:03:47.157
and the students gave me a gift certificate to a coffee shop.

00:03:47.157 --> 00:03:53.386
Now, the amount of money there was not much; but I meant that every time I travelled,

00:03:53.386 --> 00:03:56.665
I could go in the coffee shop in the airport and get a cup of coffee

00:03:56.681 --> 00:04:01.157
and it reminded me of that group of students and their thoughtfulness.

00:04:04.572 --> 00:04:08.769
Sometimes, especially early on the design project,

00:04:08.769 --> 00:04:15.544
it can be difficult to get a hold of exactly the users that you believe will eventually be the target.

00:04:15.544 --> 00:04:20.051
In this case don’t give up completely and interview nobody.

00:04:20.051 --> 00:04:24.953
Do the best you can; approximate if necessary.

00:04:24.953 --> 00:04:28.592
For example, if you are designing a system for doctors,

00:04:28.592 --> 00:04:36.838
early on the project, you might need to satisfice and use medical students instead of doctors.

00:04:36.838 --> 00:04:42.785
Or if you are working on something for software engineers, you might get computer science students.


00:04:42.785 --> 00:04:46.675
Not ideal; better than nothing.

00:04:48.675 --> 00:04:54.972
I’ve been teaching and doing research in human-computer interaction for about 15 years now

00:04:54.972 --> 00:05:05.273
and I can’t tell you how many times I’ve talked to students or others who say that there’s nothing to be found,


00:05:05.273 --> 00:05:11.741
that, for the problem that they’re tackling, it’s either possible or obvious.

00:05:11.741 --> 00:05:14.713
It’s very rarely the case that there is nothing new.

00:05:14.713 --> 00:05:18.812
The more mundane the domain, the more creative you need to be.

00:05:18.812 --> 00:05:22.029
If you don’t believe me, Malcolm Gladwell the writer

00:05:22.029 --> 00:05:28.930
has a wonderful explanation in the introduction to his new book of collective stories « What the Dog [Saw] ».

00:05:28.930 --> 00:05:35.590
“The trick to finding ideas is to convince yourself that everyone and everything has a story to tell.

00:05:35.590 --> 00:05:40.909
I say trick, but what I really mean is challenge, because it’s a very hard thing to do.

00:05:40.909 --> 00:05:45.101
Shampoo doesn’t seem interesting? Well, dammit, it must be, and if it isn’t,

00:05:45.101 --> 00:05:48.893
I have to believe that it will ultimately lead me [to something] that is.”

00:05:48.893 --> 00:05:54.506
“The other trick to finding ideas is figuring out the difference between power and knowledge.

00:05:54.506 --> 00:05:58.217
You don’t start at the top if you want to find the story.

00:05:58.217 --> 00:06:04.206
You start in the middle, because it’s the people in the middle who do the actual work in the world.

00:06:04.206 --> 00:06:07.718
My friend Dave, who taught me about ketchup, is a middle guy.

00:06:07.718 --> 00:06:11.120
He’s worked on ketchup. That’s how he knows about it.

00:06:11.120 --> 00:06:15.731
People at the top are self-conscious about what they say (and rightfully so)

00:06:15.731 --> 00:06:18.584
because they have position and privilege to protect —

00:06:18.599 --> 00:06:22.545
and self consciousness is the enemy of ‘interestingness.’”

00:06:22.545 --> 00:06:25.668
“In ‘The Pitchman’ you’ll meet Arnold Morris, who gave me the pitch

00:06:25.668 --> 00:06:30.600
for the ‘Dial-O-Matic’ vegetable slicer one summer day in his kitchen on the Jersey Shore:

00:06:30.600 --> 00:06:32.772
‘Come on over, folks.

00:06:32.772 --> 00:06:36.882
I’m going to show you the most amazing slicing machine you have ever seen in your life,’ he began.

00:06:36.882 --> 00:06:42.647
He picked up a package of barbecue spices and used it as a prop.

00:06:42.647 --> 00:06:48.304
‘Take a look at this!’ He held it in the air as if he were holding up a Tiffany vase.

00:06:48.304 --> 00:06:53.070
That’s where you find stories, in someone’s kitchen on the Jersey Shore.”

00:06:54.793 --> 00:06:58.003
What makes a good question when you’re interviewing?

00:06:58.003 --> 00:07:04.489
Let’s look an example and you can decide whether you think this is a good question.

00:07:04.489 --> 00:07:11.763
Our question is: “Is the daily update an important feature to you?”

00:07:11.763 --> 00:07:16.605
My guess when asked a question like this is that most users would say “Yes.”

00:07:16.605 --> 00:07:18.968
It’s kind of a leading question.

00:07:18.968 --> 00:07:22.618
Most people, when asked if something’s important, “Sure, why not?”

00:07:22.618 --> 00:07:24.591
What’s at stake? Why would you say no?


00:07:24.591 --> 00:07:30.827
If you wanted to learn about the daily update, participant observation might be a lot more effective


00:07:30.827 --> 00:07:35.890
because you could see whether people actually use the daily update

00:07:35.890 --> 00:07:39.671
or you might even use log files as a way of driving your questions.

00:07:39.671 --> 00:07:48.423
So, for example, you might ask somebody, “I see from the log that you’ve never used the daily update.


00:07:48.423 --> 00:07:50.761
Why is that? Tell me more.”

00:07:50.761 --> 00:07:58.233
Especially early on, the more open-ended your questions are, the more interesting the answers that you’ll get.


00:08:06.402 --> 00:08:10.785
Remember — back to our Walmart example — they asked a leading question:

00:08:10.785 --> 00:08:13.383
“Would you like stories with less clutter?”

00:08:13.383 --> 00:08:18.176
The answer, of course, is “yes”, but it may not be what people would actually want.

00:08:19.269 --> 00:08:24.348
Here’s another example of a question: “What would you like in a tool?”

00:08:24.348 --> 00:08:29.981
The problem with this question, at least as a beginning or primary question,

00:08:29.981 --> 00:08:33.522
is that users are experts in their own lives.


00:08:33.522 --> 00:08:40.811
They’re not experts in design — that’s your role — so people may not know what to say.

00:08:40.811 --> 00:08:45.290
There’s a famous fable in design attributed to Henry Ford where he says that,

00:08:45.290 --> 00:08:49.737
“if I asked people what they wanted in the age of the horse and buggy,

00:08:49.737 --> 00:08:54.183
what they’d say they want is a faster horse.”


00:08:54.183 --> 00:08:58.863
And of course we know that the designers used the possibility of the car,

00:08:58.863 --> 00:09:03.542
but users may not know about that yet.

00:09:03.542 --> 00:09:11.115
Instead, it’s more valuable to ask users about their own lives and their own goals.

00:09:11.115 --> 00:09:14.008
That’s what people are experts in.

00:09:14.962 --> 00:09:22.924
Here are three other types of questions to avoid because people aren’t very good at self report for them;


00:09:22.924 --> 00:09:29.819
Asking people what they would do or like or want in a hypothetical scenario — too hard to know.

00:09:31.581 --> 00:09:36.443
Asking us how often we do things — We often lie to ourselves.

00:09:36.443 --> 00:09:42.860
For example, “How often do you exercise?” Most people overestimate in their answer to that question.


00:09:44.245 --> 00:09:50.605
Other questions about how are things similar. This works better if you can make things concrete.


00:09:50.605 --> 00:09:56.693
“How much do I exercise in a typical week?” — Not very reliable.

00:09:56.693 --> 00:10:02.738
“How much did I exercise this week?” Well, I’m likely to give more truthful answer to that one.

00:10:02.738 --> 00:10:06.863
It’s more concrete, and more recent in my memory.

00:10:06.863 --> 00:10:11.809
I suggest also avoiding asking people how much they like things on an absolute scale:

00:10:11.809 --> 00:10:15.278
What does “seven” mean?

00:10:15.278 --> 00:10:20.114
Finally, avoid binary questions, things that have a yes or no answer.


00:10:20.114 --> 00:10:24.938
“Do you like grapefruit?” – “Yes.” Not a very interesting interview.

00:10:25.199 --> 00:10:28.240
So what are good questions?

00:10:28.240 --> 00:10:33.978
Especially at the beginning of an interview, I recommend open-ended questions.

00:10:33.978 --> 00:10:38.761
After asking the question, give somebody a chance to respond.

00:10:38.761 --> 00:10:42.351
A little bit of silence is golden.

00:10:46.089 --> 00:10:52.631
When I first started teaching, I made a mistake that a lot of young teachers make:

00:10:52.631 --> 00:10:58.510
I would ask the class for a question, and after 17 milliseconds, when nobody had answered,

00:10:58.510 --> 00:11:01.404
I would jump in and offer the question myself.

00:11:01.404 --> 00:11:09.019
I trained the students that they didn’t need to be part of the class, that I would always answer every question.

00:11:09.019 --> 00:11:11.652
There is no reason to participate.

00:11:11.652 --> 00:11:14.739
One of my colleagues taught me to wait a little longer.

00:11:14.739 --> 00:11:20.227
And it’s amazing, even in a quiet classroom, before students are used to interacting,

00:11:20.227 --> 00:11:23.898
let those few seconds of silence happen and people will start to chime in.

00:11:23.913 --> 00:11:26.434
The same is true in interviews.

00:11:26.434 --> 00:11:31.417
You may get a quick answer at first. Let some silence happen.

00:11:31.417 --> 99:59:59.999
After a few seconds, you’ll hear the second story. And the second story is often a lot more interesting.