perceived problems with personas

16 Nov 2008 - 7:50am
6 years ago
34 replies
1901 reads
Michael Stiso
2006

Hi, all. I was inspired to post this question by the very interesting
ACD/UCD discussion, during which the personas concept has frequently been
mentioned. I've always been a little uncertain of personas, but many people
seem to love them, and so I'm wondering if I'm missing or misunderstanding
something.

Below are my three main problems with the concept. I'm hoping some of you
might be able to tell me whether (and how) I'm on- or off-base with them.

1) *Frankenstein.* As I understand it, the better persona practitioners will
base their constructions on real-world data. Essentially, they use various
methods to gather a bunch of data on behaviors, attitudes, and demographics
from some population, and then reorganize and combine the various data
points into some mock person. If that is correct, then it would seem that
the resulting persona doesn't represent any actual user -- it's just made up
of parts of real users, like a Frankenstein's monster. As James
Page<http://gamma.ixda.org/discuss.php?post=35466#35613>said in a
comment on the ACD/UCD thread, the result is a fiction.

2) *Efficiency.* If personas are made up of pieces of real users, I'm
uncertain about how much benefit is gained from from recombining those
pieces into a narrative that makes sense, as opposed to simply looking at
the dataset and any potential relationships within it. For example, in his
original post in the last ACD/UCD thread, Jared Spool describes the
following:

> "Recently, I had a client show me their persona descriptions that talked
> about the car the family had and the family dog. My first inclination was to
> suggest they take this information out. However, their project was a
> home-improvement information site and providing filters for pet-friendly
> improvement projects and easy-to-bring-home materials was an obvious
> no-brainer out of this simple info."
>
Unless the persona creators got lucky, the car and pet details in their
creations came from some actual data -- a demographics survey, focus groups,
user interviews, or whatever. If so, I wonder whether the effort of creating
those fictional representations of home-improvement customers could have
been saved by just looking at the simple data on which they were based. Put
another way, was it the persona that helped, or was it the simple finding
that some of their customers drive small cars and own pets?

3) *Variability.* Persona creation, and the conclusions that come from them,
seem to be more of an art than a science. (Ultimately, design is an art, but
the recommendations on which those designs are based should perhaps fall
more in the science realm.) My impression (and that's all it is) is that the
process of imagining a persona turns datasets into inkblots, with different
practitioners looking at the same thing and coming up with different
interpretations. If so, that makes me question the utility of the concept.
Going back to the ACD/UCD discussion, I wonder if there would be less
variability in design recommendations if they were based on analyses of the
central activities/tasks that users must perform, along with real data
showing associations between user characteristics and those activities.
That's just speculation, but if there's any truth in it, then the costs and
benefits of personas as a design tool may be an interesting investigation.

Thanks for your time if you've read all of this. I'm hoping it will generate
a good discussion on the topic and give me some insight into the issue.

Mike

-------

Michael Stiso, Ph.D.
HCI Researcher
SINTEF ICT
Oslo, Norway

Comments

16 Nov 2008 - 8:38am
SemanticWill
2007

The myths about the efficacy or not of archtypes in design and design
research tend to be promulgated by those that 1. Can't do them; 2.
Have never done real design or ethnographic research; 3. To justify
the tried and true fact that while most companies and clients say they
want to do (insert silly acronym for process meant to sell services by
people needing to fill a blog post or justify their fees); the truth
is that design research is valuable, and costly. Real practitioners
rely on real people research to inform design and those that can't or
won't then write about how ineffective it is having never done it or
validating the business decision not to do it because of cost. I
heartily disagree with folks that think personas and archtype creation
is just interviews & usability tests. It's also quantitative based in
usage statistics, surveys, as well as psychographic, technographic and
demographic. The analysis is more science, the narrative to fillout
the archtype is art. If no other value exist, and many do- it is that
when you make controversial design decisions that contradict the whims
and pet assumptions of stakeholders, you have something to back up
your argument. When it comes to prioritizing features and initiatives
- you have something to inform cost benefit of every keep/drop
decision. In the end - designing is just as much about power and
politics and with user research you are negotiating your design from a
position of weakness and it's better to have more than emty rhetoric,
lack of data and ignorance of users when you enter that battle with
stakeholders- and it is a battle, make no mistake. You may be given
some degree of a honeymoon sometimes where people buys your designs
without challenge but eventually those halcyon days will end and when
they do, better have done the researg so you don't look like a horses
ass. :-)

Cheers!

will evans
emotive architect &
hedonic designer
will at semanticfoundry.com
617.281.1281
twitter: semanticwill
aim: semanticwill
gtalk: wkevans4
skype: semanticwill
_________________________
Sent via iPhone

On Nov 16, 2008, at 8:50 AM, "Michael Stiso" <mikestiso at gmail.com>
wrote:

> Hi, all. I was inspired to post this question by the very interesting
> ACD/UCD discussion, during which the personas concept has frequently
> been
> mentioned. I've always been a little uncertain of personas, but many
> people
> seem to love them, and so I'm wondering if I'm missing or
> misunderstanding
> something.
>
> Below are my three main problems with the concept. I'm hoping some
> of you
> might be able to tell me whether (and how) I'm on- or off-base with
> them.
>
> 1) *Frankenstein.* As I understand it, the better persona
> practitioners will
> base their constructions on real-world data. Essentially, they use
> various
> methods to gather a bunch of data on behaviors, attitudes, and
> demographics
> from some population, and then reorganize and combine the various data
> points into some mock person. If that is correct, then it would seem
> that
> the resulting persona doesn't represent any actual user -- it's just
> made up
> of parts of real users, like a Frankenstein's monster. As James
> Page<http://gamma.ixda.org/discuss.php?post=35466#35613>said in a
> comment on the ACD/UCD thread, the result is a fiction.
>
> 2) *Efficiency.* If personas are made up of pieces of real users, I'm
> uncertain about how much benefit is gained from from recombining those
> pieces into a narrative that makes sense, as opposed to simply
> looking at
> the dataset and any potential relationships within it. For example,
> in his
> original post in the last ACD/UCD thread, Jared Spool describes the
> following:
>
>> "Recently, I had a client show me their persona descriptions that
>> talked
>> about the car the family had and the family dog. My first
>> inclination was to
>> suggest they take this information out. However, their project was a
>> home-improvement information site and providing filters for pet-
>> friendly
>> improvement projects and easy-to-bring-home materials was an obvious
>> no-brainer out of this simple info."
>>
> Unless the persona creators got lucky, the car and pet details in
> their
> creations came from some actual data -- a demographics survey, focus
> groups,
> user interviews, or whatever. If so, I wonder whether the effort of
> creating
> those fictional representations of home-improvement customers could
> have
> been saved by just looking at the simple data on which they were
> based. Put
> another way, was it the persona that helped, or was it the simple
> finding
> that some of their customers drive small cars and own pets?
>
> 3) *Variability.* Persona creation, and the conclusions that come
> from them,
> seem to be more of an art than a science. (Ultimately, design is an
> art, but
> the recommendations on which those designs are based should perhaps
> fall
> more in the science realm.) My impression (and that's all it is) is
> that the
> process of imagining a persona turns datasets into inkblots, with
> different
> practitioners looking at the same thing and coming up with different
> interpretations. If so, that makes me question the utility of the
> concept.
> Going back to the ACD/UCD discussion, I wonder if there would be less
> variability in design recommendations if they were based on analyses
> of the
> central activities/tasks that users must perform, along with real data
> showing associations between user characteristics and those
> activities.
> That's just speculation, but if there's any truth in it, then the
> costs and
> benefits of personas as a design tool may be an interesting
> investigation.
>
> Thanks for your time if you've read all of this. I'm hoping it will
> generate
> a good discussion on the topic and give me some insight into the
> issue.
>
> Mike
>
> -------
>
> Michael Stiso, Ph.D.
> HCI Researcher
> SINTEF ICT
> Oslo, Norway
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help

16 Nov 2008 - 10:46am
Josh Seiden
2003

I've probably posted this reference before, but Roger Martin wrote a
good article comparing two ways of thinking about problems and
processes. He calls this dynamic reliability vs. validity.

"Reliability seeks to produce consistent,
predictable outcomes by utilizing a system
that is restricted to the use of objective data....

"Validity, on the other hand, seeks to
produce outcomes that meet the desired
objective, even if the system employed
can%u2019t produce a consistent,predictable out-
come."

Personas are a classic example of a method that is "valid" in
Martin's terms.

The whole article can be found here:
https://www.rotman.utoronto.ca/integrativethinking/ValidityVSReliability.pdf

JS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

16 Nov 2008 - 12:11pm
Michael Stiso
2006

> Personas are a classic example of a method that is "valid" in
Martin's terms.

That's an interesting article, but I must admit it has me confused
as to whether personas would be an attempt at validity or
reliability. I see personas as an attempt to represent a product's
users, the goal being to reliably categorize users into types so that
the product can be designed to fit those types. I see that as putting
the method on the reliability side, although I'm uncertain as to
whether it actually is reliable.

Or looked at from another angle, can personas be considered valid if
they don't represent actual users?

Thank you for the link.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

16 Nov 2008 - 12:40pm
Mark Schraad
2006

Great article to mention Joshua.

It is a great read for those of us in an MBA heavy corporate
environment. I find myself coming back to read and reference this
article often. Martin touches this in another somewhat charged
article from the same publication called "designing in hostile
territory"... also worth reading.

As an aside... the relationship between reliability and validity...
is very similar to that between efficiency and effectiveness. The two
are not mutually exclusive and in fact interact to the point that
it's not possible to optimize both.

Mark

On Nov 16, 2008, at 8:46 AM, Joshua Seiden wrote:

> I've probably posted this reference before, but Roger Martin wrote a
> good article comparing two ways of thinking about problems and
> processes. He calls this dynamic reliability vs. validity.
>
> "Reliability seeks to produce consistent,
> predictable outcomes by utilizing a system
> that is restricted to the use of objective data....
>
> "Validity, on the other hand, seeks to
> produce outcomes that meet the desired
> objective, even if the system employed
> can%u2019t produce a consistent,predictable out-
> come."
>
> Personas are a classic example of a method that is "valid" in
> Martin's terms.
>
> The whole article can be found here:
> https://www.rotman.utoronto.ca/integrativethinking/
> ValidityVSReliability.pdf
>
> JS
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help

16 Nov 2008 - 12:43pm
Marijke Rijsberman
2004

Michael,

I'm glad you posted your concerns. I think personas are a great tool for
designers trying to resist the natural slide into accommodating every
possible scenario in the world. In principle. There are definitely
situations when I think they are very helpful.

However, in practice I've seen personas much abused, because they are done
> without a good understanding of market segmentation (and thus present
characteristics and goals as representative when they are not)
> without a good understanding of the domain (and thus present
characteristics that are pretty much irrelevant to the problem at hand)
> without a good understanding of how they will be used (and thus present
patterns in a vacuum, without any idea of what they might apply to or when
they might be valid)

They advantage of personas is that they are easy to make vivid. The danger
is that vividness is made to stand in for a true understanding of the field.
As a well-respected practitioner in the field said to me on Friday, in many
cases the best thing would be to create movie out of your field research,
but the editing is just too much work. Personas are cheap.

It doesn't do user research any favors to cut corners in this way.

marijke

Marijke Rijsberman
http://www.interfacility.com
http://landfill.wordpress.com

16 Nov 2008 - 4:08pm
SemanticWill
2007

Personas are not meant to represent one person or user. They are meant
to represent classes or segments of users. This is a valid use of
them. Creating stereotypes and aggregates is useful in many endeavors.
You by default will always design for someone- question is whether
that person is the developer, founder, CEo or his daughter that is
positive some feature will be kewl. Those methods have worked- it's
like like design by developer doesn't create smash hits like basecamp
by 37s - the problem is that no one besides them finds their INS
models particularly usable or intuitive and barely learnable- their
breakout success maybe more attributable to their business model or
the eloquence and celebrity stature of the founders but not by any
metric a Ux professional would call repeatable and predictable.

will evans
emotive architect &
hedonic designer
will at semanticfoundry.com
617.281.1281
twitter: semanticwill
aim: semanticwill
gtalk: wkevans4
skype: semanticwill
_________________________
Sent via iPhone

On Nov 16, 2008, at 10:11 AM, Mike Stiso <mikestiso at gmail.com> wrote:

>> Personas are a classic example of a method that is "valid" in
> Martin's terms.
>
> That's an interesting article, but I must admit it has me confused
> as to whether personas would be an attempt at validity or
> reliability. I see personas as an attempt to represent a product's
> users, the goal being to reliably categorize users into types so that
> the product can be designed to fit those types. I see that as putting
> the method on the reliability side, although I'm uncertain as to
> whether it actually is reliable.
>
> Or looked at from another angle, can personas be considered valid if
> they don't represent actual users?
>
> Thank you for the link.
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help

16 Nov 2008 - 4:42pm
Michael Stiso
2006

> Creating stereotypes and aggregates is useful in
> many endeavors.

Some endeavors. But if decisions are being based on a stereotype of
something that doesn't actually exist -- an aggregate in which bits
and pieces represent characteristics of different users, but in which
the whole represents no one -- then I question the usefulness of the
method. Wouldn't it be more efficient just to focus on the bits and
pieces rather than combine them into a fictional character?

On the other hand, I agree with what Marijke had to say on the
matter. Personas do make good communication tools, I think,
particularly with clients -- at least as long as the creators don't
go overboard on the details.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

16 Nov 2008 - 4:42pm
Ali Naqvi
2008

Hi Michael,
http://www.snitker.com/public/dokumenter/A4personasplakat2.pdf

This has been made by a Danish PHd.

Ali

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

17 Nov 2008 - 3:52am
James Page
2008

>
> Personas are not meant to represent one person or user. They are meant to
> represent classes or segments of users. This is a valid use of them.

It is hard for a Peronsa to represent classes or segments of users. Chapman
and Milham put it far better than I could.

Unfortunately, one cannot use scattered data points to assemble a composite
> description with any known relationship to a population. Each
> additional datum – if it has any informational content – has a probability
> less than 1.0.
>

See:
http://cnchapman.files.wordpress.com/2007/03/chapman-milham-personas-hfes2006-0139-0330.pdf

We are told so many times not to use us for our designs, ourselves, or our
mothers as the target for a design. But surely this is better than something
that is purely fictional.

Are they useful as a good communication tools?

Probably not as you are communicating something that is fictional, and does
not represent anything. Mike suggest just using the real data subjects. Why
not just put a fake name on one of data subjects for external use?

James

On Sun, Nov 16, 2008 at 10:42 PM, Ali Naqvi <ali at amroha.dk> wrote:

> Hi Michael,
> http://www.snitker.com/public/dokumenter/A4personasplakat2.pdf
>
> This has been made by a Danish PHd.
>
> Ali
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

17 Nov 2008 - 4:50am
martinpolley
2007

The way I understand it, that is a bad idea because you could end up
designing to the (non-representative) quirks of the individual, rather than
needs that are shared by many such individuals.

Cheers,
--
Martin Polley
Technical writer, interaction designer
+972 52 3864280
Twitter: martinpolley
<http://capcloud.com/>

On Mon, Nov 17, 2008 at 11:52 AM, James Page <jamespage at gmail.com> wrote:

> >
>
> [...] you are communicating something that is fictional, and does
> not represent anything. Mike suggest just using the real data subjects. Why
> not just put a fake name on one of data subjects for external use?
>
>

17 Nov 2008 - 1:46pm
Elizabeth Bacon
2003

Hi Michael, et al,

I helped formalize persona creation methods while I was at Cooper,
and I recently spoke on the subject of some common persona
misconceptions with a current Cooperista, Steve Calde. Please see
http://www.devise.com/further_reading and look to the bottom of the
page for an embedded version of our "Death to Personas! Long Live
Personas!" presentation.

But I sure wish you'd posted your concerns about personas before
that presentation was authored, as the three you raise make for
interesting topics of discussion. :)

It seems to me that what's missing most here is a grounding in their
formal practice. (For that, we can all look forward to Kim Goodwin's
upcoming book!) But here goes a quick pass at addressing your
concerns, especially the first one which nobody's really targeted
yet in this thread:

1) Frankenstein. The persona creation methodology does not involve
one "combin[ing] the various data points into some mock person".
Instead, each individual's characteristics are mapped against
specific dimensions of interest discovered during field research
(ideally!!). Then you'll do it again for another dimension, and
another and another. For example... if a key dimension of interest is
tech-savviness, you'll map the place of individuals A, B and C on
that spectrum, and then you'll map their place against another
dimension of interest such as concern-about-security. And then
you'll see what kind of pattern has developed by (for example)
seeing that B & C are both tech-savvy and unconcerned about security
and are also grouped on a variety of other dimensions. And A is an
outlier who's not tech-savvy and is concerned about security, and
also isolated on a variety of other dimensions. Voila, you have
identified two separate user archetypes, one fed by B&C data and the
other by A (this being a simplified example, of course); they are
proto-personas who represent patterns in your research data. They are
not an artificial combination of disparate body parts, to refer back
to your proposed monster. ;)

2) Efficiency. How efficient is it to share a bunch of numerical data
points, versus how much more efficient is it to map those data points
onto a consolidated graph? The unification of research patterns into
a single, narrative user archetype is fundamentally an efficient
method to model one's ethnographically-inspired research. It's
similar to making graphical models of complex workflows, or diagrams
of contexts of use. Models are inherently more efficient than
discursive texts or indexes of findings. The narrative form of a
persona description can be taken too far, however, and this is a
problem discussed in our presentation. Overly biographical and
life-goal-oriented personas can be distracting and are indeed
inefficient for teams to consume. It takes some practice to find the
right level at which to document personas, and often for me still
involves fine-tuning for the audience at hand.

3) Variability. Hopefully my quick elucidation about the original
persona creation methodology helps you to see that the mapping of
individuals to dimensions of interest is a relatively scientific
method. There are occasional disagreements in the research team about
where to place an individual on the scale of a dimension, but such
disagreements tend to be resolved quickly and usually indicate there
are multiple dimensions of interest in that one area. And having
taught the Cooper U Interaction Design Practicum course, I can attest
that different classrooms filled with fresh attendees have no problem
with repeatedly mapping and identifying the archetypes originally
discovered in the research data.

I look forward to learning more about the relationship between
validity vs. reliability from that document Josh sent around. This
morning I did quickly peruse the Chapman-Milham Personas piece. It's
probably statistically true that "as features are added, the overall
probability of a composite decreases". That's why it's utterly
crucial that personas be created and then utilized within a specific
domain -- one should not re-purpose personas from a research
conducted around virus software, for example, to a design problem on
cloud computing; to do so would expand and re-consider their
characteristics using fiction, not fact. As for their repeated points
about it being "unclear about what data underlies these""
personas...well, yeah, that's what you get when you write an
academically-inclined paper far removed from the research & design
process. ;) The data was there for the team to use but it generally
isn't passed along to the client. Still, I think their proposals for
future research are excellent; and to their first suggestion, about
creating a set of customer data and giving it to independent teams --
that's what I've seen done at Cooper U with repeated validity.

Personas are not any kind of be-all end-all method, so don't
misconstrue my points. They can, however, be a powerful tool in a
designer's toolkit. And it's essential to remember that no design
process stops with personas...they are a most helpful input to a
scenario-based approach to design. Hope this helps!

Cheers,
Liz

Vice-President, IxDA
CDO, Devise

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

17 Nov 2008 - 2:48pm
jaketrimble
2008

Liz finishes this precisely...

"And it's essential to remember that no design process stops with
personas"

And Will's stance in my opinion is spot on in suggesting that
personas are essential when you are...sorry for the quote.."going to
the mattresses" :)

Personas are to me the foundation of a building. A building in which
the lobby and the second floor are almost never the same.

In the same way that Mike scrutinizes personas so can one question
only focusing "on the bits and pieces" as he puts it. Bits and
pieces are exactly that. They are meant to form a whole. It is our
job to put those pieces together. Personas become even more important
when the data sets themselves are not representative to the whole
puzzle.

It is important to understand that it is not a "fictional
character". You are building a real base, a solid foundation that
will propel your workflow from start to finish.

- 2cents from Jake

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

17 Nov 2008 - 7:04pm
Adam Korman
2004

I want to kind of expand on Liz's point a little, and maybe take this
in a slightly different direction. Personas are to digital products as
a tape measure is to a bookcase. Critics of personas often seem to
have an intense focus on validity of the tool, without a lot of
discussion about what it is people are trying to do with personas. As
I understand and use personas, their value has less to do with their
accuracy and more to do with their human qualities and specificity.

A persona provides me as a designer (and the team at large) a specific
person to identify with. A lot of products don't treat people very
well, and having a persona around as a constant reminder that the
product is interacting with a specific person helps us avoid treating
people like anonymous "users." Personas help keep me empathetic.

A persona also provides a shorthand for whatever research has been
done and helps keeps things focused in context. So, when I say "Does
Joan [the persona] need that feature?" it's often just a quicker (and
nicer) way of saying "Remember that we did a bunch of research and
came to agreement on who the target is for our design? This random
request from a single user doesn't seem to match our data, so let's
stay on target until there's some more evidence that this is a real
concern."

Now, I'm kind of downplaying the importance of personas being a good
archetype of the user base, which is not to say that I don't think
that's important too, but being "right" is only aspect of why they are
a handy tool.

-Adam

On Nov 17, 2008, at 3:46 AM, Elizabeth Bacon wrote:

> Personas are not any kind of be-all end-all method, so don't
> misconstrue my points. They can, however, be a powerful tool in a
> designer's toolkit.

17 Nov 2008 - 7:18pm
Steve Baty
2009

Michael,

2008/11/17 Michael Stiso <mikestiso at gmail.com>

> 1) *Frankenstein.* As I understand it, the better persona practitioners
> will
> base their constructions on real-world data. Essentially, they use various
> methods to gather a bunch of data on behaviors, attitudes, and demographics
> from some population, and then reorganize and combine the various data
> points into some mock person. If that is correct, then it would seem that
> the resulting persona doesn't represent any actual user -- it's just made
> up
> of parts of real users, like a Frankenstein's monster. As James
> Page<http://gamma.ixda.org/discuss.php?post=35466#35613>said in a
> comment on the ACD/UCD thread, the result is a fiction.
>

I think this point is the major misconception about the creation of a good
persona. As Liz outlines, and Will alludes to, a persona is an archetype
representing a segment of the audience. To derive those archetypes one does
not copy and paste characteristics from real users willy-nilly to craft a
pleasant-looking, but fictional user.

Just like in market research, persona creation is a segmentation exercise
that should be a) driven by real user research; b) analysed using
appropriate techniques. In this case, one of several forms of multi-variate
statistical analysis ranging from the simple (cross-tabulation, or radial
maps) to the complex (clustering analysis, multi-dimensional scaling, or
factor analysis). In each case the aim is to identify groupings or clusters
of users who share a largely overlapping set of characteristics in those
dimensions relevant to your design problem.

Personas - when done in this manner - *are not* fictional; they're
representative, and that's a whole world of difference.

Regards
Steve

--
Steve 'Doc' Baty | Principal Consultant | Meld Consulting | P: +61 417 061
292 | E: stevebaty at meld.com.au | Twitter: docbaty

Blog: http://docholdsfourth.blogspot.com
Contributor - UXMatters - www.uxmatters.com

17 Nov 2008 - 1:27pm
Josh Seiden
2003

"Probably not as you are communicating something that is fictional,
and does not represent anything."

There are so many problems with this statement James! Is it your
contention that fiction doesn't represent anything? Or that
practicing designers are representing NOTHING with their personas?
You've got to be kidding me!

Designers are representing their research, their ideas, their
thinking. If you want, they're representing their opinions and
biases as well.

Design is a creative act. We all have to represent our creative
thinking. We use many tools to do this--and we don't need to claim
that our tools are scientific for our work to be valid.

As for fiction not representing anything... I'll leave that for
another day!

JS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

18 Nov 2008 - 12:05am
Jarod Tang
2007

Hi James,

> We are told so many times not to use us for our designs, ourselves, or our
> mothers as the target for a design. But surely this is better than something
> that is purely fictional.
Persona is/should be based on user research data underneath (at least
for design). This is defined from early practitioners like Alan Cooper
(and he proved why the instantiation of persona should based on
concrete user research in his books). To say it's fictional, one may
miss the point of persona usage for design .

Regards,
Jarod

--
http://designforuse.blogspot.com/

18 Nov 2008 - 3:00am
Michael Stiso
2006

Ok, this information has been helpful. What Elizabeth and Steve say
about the statistical methods (clustering, etc.) behind persona
creation makes sense, and that kind of analysis is something that I
already do.

The problem is that I've never witnessed nor heard of personas being
created in such a manner (which is perhaps part of a larger problem).
Rather, I see examples of personas described in an amount of detail
that would necessitate a large and comprehensive dataset, built up
iteratively based on ongoing analyses. That would be ideal, I think,
but it seems to be something of a rarity in industry, and so my
concern is that such detail is often added based just on a hunch.

Also, classification and grouping methods have been around for a
while. If personas make use of the same methods, does persona
creation mainly involve adding another step to the process:
instantiating the different classifications into individual
characters? If so, the primary benefit of doing would seem to be the
achievement of a possible means of communication with others
(clients, developers, etc.).

I can see using the instantiations as a focus for design, too, but
there's a risk of treating the creation as more than a statistical
representation, which could lead to assumptions about the
character's behavior, motivation, and characteristics that are based
on the designer's stereotypes rather than on actual data. However, I
can imagine that being either a good or a bad thing, depending on the
designer.

Thanks for all the info so far.

Mike

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

18 Nov 2008 - 4:47am
Sam Menter
2008

Hi
The biggest problem I've found with personas comes when they haven't been
explained fully to a client - for example a stakeholder has been forwarded
the persona document, and they see the personas as representing a
demographic rather than a segmentation based on likely user tasks.

I guess it's probably inevitable that the designer makes some assumptions
based on the persona, but hopefully these assumptions lean more towards
informed decisions if the persona has been well researched!

All the best
Sam

www.pixelthread.co.uk

2008/11/18 Mike Stiso <mikestiso at gmail.com>

> Ok, this information has been helpful. What Elizabeth and Steve say
> about the statistical methods (clustering, etc.) behind persona
> creation makes sense, and that kind of analysis is something that I
> already do.
>
> The problem is that I've never witnessed nor heard of personas being
> created in such a manner (which is perhaps part of a larger problem).
> Rather, I see examples of personas described in an amount of detail
> that would necessitate a large and comprehensive dataset, built up
> iteratively based on ongoing analyses. That would be ideal, I think,
> but it seems to be something of a rarity in industry, and so my
> concern is that such detail is often added based just on a hunch.
>
> Also, classification and grouping methods have been around for a
> while. If personas make use of the same methods, does persona
> creation mainly involve adding another step to the process:
> instantiating the different classifications into individual
> characters? If so, the primary benefit of doing would seem to be the
> achievement of a possible means of communication with others
> (clients, developers, etc.).
>
> I can see using the instantiations as a focus for design, too, but
> there's a risk of treating the creation as more than a statistical
> representation, which could lead to assumptions about the
> character's behavior, motivation, and characteristics that are based
> on the designer's stereotypes rather than on actual data. However, I
> can imagine that being either a good or a bad thing, depending on the
> designer.
>
> Thanks for all the info so far.
>
> Mike
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

18 Nov 2008 - 5:16am
SemanticWill
2007

Another fantastic book on qualitative and quantitive research for
personas is The User is Always Right by Steve Mulder.

will evans
emotive architect &
hedonic designer
will at semanticfoundry.com
617.281.1281
twitter: semanticwill
aim: semanticwill
gtalk: wkevans4
skype: semanticwill
_________________________
Sent via iPhone

On Nov 17, 2008, at 11:46 AM, Elizabeth Bacon
<lists at elizabethbacon.com> wrote:

> Hi Michael, et al,
>
> I helped formalize persona creation methods while I was at Cooper,
> and I recently spoke on the subject of some common persona
> misconceptions with a current Cooperista, Steve Calde. Please see
> http://www.devise.com/further_reading and look to the bottom of the
> page for an embedded version of our "Death to Personas! Long Live
> Personas!" presentation.
>
> But I sure wish you'd posted your concerns about personas before
> that presentation was authored, as the three you raise make for
> interesting topics of discussion. :)
>
> It seems to me that what's missing most here is a grounding in their
> formal practice. (For that, we can all look forward to Kim Goodwin's
> upcoming book!) But here goes a quick pass at addressing your
> concerns, especially the first one which nobody's really targeted
> yet in this thread:
>
> 1) Frankenstein. The persona creation methodology does not involve
> one "combin[ing] the various data points into some mock person".
> Instead, each individual's characteristics are mapped against
> specific dimensions of interest discovered during field research
> (ideally!!). Then you'll do it again for another dimension, and
> another and another. For example... if a key dimension of interest is
> tech-savviness, you'll map the place of individuals A, B and C on
> that spectrum, and then you'll map their place against another
> dimension of interest such as concern-about-security. And then
> you'll see what kind of pattern has developed by (for example)
> seeing that B & C are both tech-savvy and unconcerned about security
> and are also grouped on a variety of other dimensions. And A is an
> outlier who's not tech-savvy and is concerned about security, and
> also isolated on a variety of other dimensions. Voila, you have
> identified two separate user archetypes, one fed by B&C data and the
> other by A (this being a simplified example, of course); they are
> proto-personas who represent patterns in your research data. They are
> not an artificial combination of disparate body parts, to refer back
> to your proposed monster. ;)
>
> 2) Efficiency. How efficient is it to share a bunch of numerical data
> points, versus how much more efficient is it to map those data points
> onto a consolidated graph? The unification of research patterns into
> a single, narrative user archetype is fundamentally an efficient
> method to model one's ethnographically-inspired research. It's
> similar to making graphical models of complex workflows, or diagrams
> of contexts of use. Models are inherently more efficient than
> discursive texts or indexes of findings. The narrative form of a
> persona description can be taken too far, however, and this is a
> problem discussed in our presentation. Overly biographical and
> life-goal-oriented personas can be distracting and are indeed
> inefficient for teams to consume. It takes some practice to find the
> right level at which to document personas, and often for me still
> involves fine-tuning for the audience at hand.
>
> 3) Variability. Hopefully my quick elucidation about the original
> persona creation methodology helps you to see that the mapping of
> individuals to dimensions of interest is a relatively scientific
> method. There are occasional disagreements in the research team about
> where to place an individual on the scale of a dimension, but such
> disagreements tend to be resolved quickly and usually indicate there
> are multiple dimensions of interest in that one area. And having
> taught the Cooper U Interaction Design Practicum course, I can attest
> that different classrooms filled with fresh attendees have no problem
> with repeatedly mapping and identifying the archetypes originally
> discovered in the research data.
>
> I look forward to learning more about the relationship between
> validity vs. reliability from that document Josh sent around. This
> morning I did quickly peruse the Chapman-Milham Personas piece. It's
> probably statistically true that "as features are added, the overall
> probability of a composite decreases". That's why it's utterly
> crucial that personas be created and then utilized within a specific
> domain -- one should not re-purpose personas from a research
> conducted around virus software, for example, to a design problem on
> cloud computing; to do so would expand and re-consider their
> characteristics using fiction, not fact. As for their repeated points
> about it being "unclear about what data underlies these""
> personas...well, yeah, that's what you get when you write an
> academically-inclined paper far removed from the research & design
> process. ;) The data was there for the team to use but it generally
> isn't passed along to the client. Still, I think their proposals for
> future research are excellent; and to their first suggestion, about
> creating a set of customer data and giving it to independent teams --
> that's what I've seen done at Cooper U with repeated validity.
>
> Personas are not any kind of be-all end-all method, so don't
> misconstrue my points. They can, however, be a powerful tool in a
> designer's toolkit. And it's essential to remember that no design
> process stops with personas...they are a most helpful input to a
> scenario-based approach to design. Hope this helps!
>
> Cheers,
> Liz
>
> Vice-President, IxDA
> CDO, Devise
>
>
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help

18 Nov 2008 - 5:36am
SemanticWill
2007

That's a good point- I think Dan Brown may have made the point in
Communicating Design but in many coversatiins with Todd Z W - I would
say we all agree - if at all possible- never just email your
deliverable - especially conceptually abstract ones like personas,
concept models/maps, wireflows, sitemaps, ix models. Present in person
or via conference webex or video call- they need backstory and context
and they need you there to shape the audiences perception of them.
Don't sent them ahead of time either so opinions are and arguments are
not formulated before you discuss. Presentation of the deliverables is
just as cruciall as the documents themselves and you are on the hook
for designing their introduction to them :-)

will evans
emotive architect &
hedonic designer
will at semanticfoundry.com
617.281.1281
twitter: semanticwill
aim: semanticwill
gtalk: wkevans4
skype: semanticwill
_________________________
Sent via iPhone

On Nov 18, 2008, at 5:47 AM, "Sam Menter" <sammenter at gmail.com> wrote:

> Hi
> The biggest problem I've found with personas comes when they haven't
> been
> explained fully to a client - for example a stakeholder has been
> forwarded
> the persona document, and they see the personas as representing a
> demographic rather than a segmentation based on likely user tasks.
>
> I guess it's probably inevitable that the designer makes some
> assumptions
> based on the persona, but hopefully these assumptions lean more
> towards
> informed decisions if the persona has been well researched!
>
> All the best
> Sam
>
> www.pixelthread.co.uk
>
>
> 2008/11/18 Mike Stiso <mikestiso at gmail.com>
>
>> Ok, this information has been helpful. What Elizabeth and Steve say
>> about the statistical methods (clustering, etc.) behind persona
>> creation makes sense, and that kind of analysis is something that I
>> already do.
>>
>> The problem is that I've never witnessed nor heard of personas being
>> created in such a manner (which is perhaps part of a larger problem).
>> Rather, I see examples of personas described in an amount of detail
>> that would necessitate a large and comprehensive dataset, built up
>> iteratively based on ongoing analyses. That would be ideal, I think,
>> but it seems to be something of a rarity in industry, and so my
>> concern is that such detail is often added based just on a hunch.
>>
>> Also, classification and grouping methods have been around for a
>> while. If personas make use of the same methods, does persona
>> creation mainly involve adding another step to the process:
>> instantiating the different classifications into individual
>> characters? If so, the primary benefit of doing would seem to be the
>> achievement of a possible means of communication with others
>> (clients, developers, etc.).
>>
>> I can see using the instantiations as a focus for design, too, but
>> there's a risk of treating the creation as more than a statistical
>> representation, which could lead to assumptions about the
>> character's behavior, motivation, and characteristics that are based
>> on the designer's stereotypes rather than on actual data. However, I
>> can imagine that being either a good or a bad thing, depending on the
>> designer.
>>
>> Thanks for all the info so far.
>>
>> Mike
>>
>>
>> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
>> Posted from the new ixda.org
>> http://www.ixda.org/discuss?post=35624
>>
>>
>> ________________________________________________________________
>> Welcome to the Interaction Design Association (IxDA)!
>> To post to this list ....... discuss at ixda.org
>> Unsubscribe ................ http://www.ixda.org/unsubscribe
>> List Guidelines ............ http://www.ixda.org/guidelines
>> List Help .................. http://www.ixda.org/help
>>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help

18 Nov 2008 - 9:24am
Eva Kaniasty
2007

Mike,

I think you make a mistake in assuming that creating the persona is an extra
step in the research process that could be eliminated - i.e. the persona
itself has no value, it is the research behind it that's really useful.

I recently went through a persona project which included going through
volumes of qualitative and quantitative research. It was extremely
valuable, but also very time consuming. There is no way each person on my
team would have time to go through and digest all that research. Even if
they did, none of us would be able to retain all that information for very
long, which means that every time we wanted to use that research to help our
decision-making, we would have to go back and dig through the research to
find the relevant information. Needless to say, this would not be very
efficient or very likely.

A persona is a model that represents all that research in a form that is
both succinct and easy to remember. The purpose of the artifact is to help
facilitate communication and decision-making. That's what makes it hugely
valuable as far as I'm concerned.

Eva Kaniasty
http://www.linkedin.com/in/kaniasty

On Tue, Nov 18, 2008 at 4:00 AM, Mike Stiso <mikestiso at gmail.com> wrote:

>
> Also, classification and grouping methods have been around for a
> while. If personas make use of the same methods, does persona
> creation mainly involve adding another step to the process:
> instantiating the different classifications into individual
> characters? If so, the primary benefit of doing would seem to be the
> achievement of a possible means of communication with others
> (clients, developers, etc.).
>
>

18 Nov 2008 - 9:39am
Todd Warfel
2003

On Nov 18, 2008, at 6:36 AM, William Evans wrote:

> Presentation of the deliverables is just as cruciall as the
> documents themselves and you are on the hook for designing their
> introduction to them :-)

I wouldn't even consider presenting something like personas any way
other than in person unless it's a client I know we've presented
personas to before.

Personas are a convention that still need some handholding and
explaining. People have heard of them. Some people have participated
in creating them. However, even in watching this conversation, we can
see the struggle around personas.

If you want them to be successful:
* Base them on data
* Validate them
* Set expectations
* Present them in person

Cheers!

Todd Zaki Warfel
President, Design Researcher
Messagefirst | Designing Information. Beautifully.
----------------------------------
Contact Info
Voice: (215) 825-7423
Email: todd at messagefirst.com
AIM: twarfel at mac.com
Blog: http://toddwarfel.com
Twitter: zakiwarfel
----------------------------------
In theory, theory and practice are the same.
In practice, they are not.

18 Nov 2008 - 10:03am
Michael Stiso
2006

> I think you make a mistake in assuming that creating
> the persona is an extra step in the research process
> that could be eliminated - i.e. the persona itself has no
> value, it is the research behind it that's really useful.

I wasn't assuming that, though, Eva. Rather, it seems to me that the
main use of the additional step is to obtain a communication tool,
whether that be for communicating with clients, developers, or other
designers. Your use of personas would seem to support that idea.

As for whether the step can be eliminated or not, I suppose that
depends on the project and the team.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

18 Nov 2008 - 10:07am
Chris Noessel
2005

A brief caution about one ugly trap of personas: Yes, they should be
based on user research. But their benefit does not come solely
through how a persona embodies the research. (It's a pretty lossy
compression of data if you think about it.)

A major benefit comes from the effect they have in design on the
designer, i.e. they get you into an intentional stance when
designing. This is to avoid designing for yourself, or for some
stretchy variables in a market segment, or for the technology.
Getting into an intentional stance clears away all those extraneous
trappings and focuses you on helping personas achieve their goals.

It's much, much better to have these personas based on sound user
research, so that you're not designing for the wrong expertise,
expectations, contexts, or goals, but a persona still has benefit to
design thinking even if you have to make that persona up on the spot.

The trap to avoid then, is stressing out over the scientific-ness of
either the research or the personas' relation to that research. Do a
good qualitative job, and that will be enough.

Longer article on Cooper's blog: http://tinyurl.com/5btgxe

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

19 Nov 2008 - 6:33am
Sam Menter
2008

Chris - I think that's a key point:

Personas help us avoid the pitfalls of:

1. Designing for ourselves (aka art)
2. Designing for some stretchy variables in a market segment / demographic
3. Designing for the technology

And I'd add,

4. Trying to design for everyone, thereby creating design that meets
nobody's needs

s

2008/11/18 Chris <chrisnoessel at hotmail.com>:
> A brief caution about one ugly trap of personas: Yes, they should be
> based on user research. But their benefit does not come solely
> through how a persona embodies the research. (It's a pretty lossy
> compression of data if you think about it.)
>
> A major benefit comes from the effect they have in design on the
> designer, i.e. they get you into an intentional stance when
> designing. This is to avoid designing for yourself, or for some
> stretchy variables in a market segment, or for the technology.
> Getting into an intentional stance clears away all those extraneous
> trappings and focuses you on helping personas achieve their goals.
>
> It's much, much better to have these personas based on sound user
> research, so that you're not designing for the wrong expertise,
> expectations, contexts, or goals, but a persona still has benefit to
> design thinking even if you have to make that persona up on the spot.
>
> The trap to avoid then, is stressing out over the scientific-ness of
> either the research or the personas' relation to that research. Do a
> good qualitative job, and that will be enough.
>
> Longer article on Cooper's blog: http://tinyurl.com/5btgxe
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

19 Nov 2008 - 8:49am
James Page
2008

Hi All,
I think it may help people here if I inject some theory into
this discussion.

The first point is that people keep making claims that the method has some
scientific validity. For example Liz says that "Hopefully my quick
elucidation about the original persona creation methodology helps you to see
that the mapping of individuals to dimensions of interest is a relatively
scientific method"

Either a method is scientific or pseduscientific. There is no middle ground.
Is the distinction only important as academic argument? The answer is no,
and it helps to understand a small bit of history to see why.

The distinction between science or psedu science came about because there
where two political movements that claimed that they where scientific, and
by following them would lead to improvement to everybody in society. The two
political movements where communism, and national socialism.

Karl Poppers, who you could say his early life was upturned by both
movements, thought that it is important to qualify what is scientific and
what is pseduscientific. [DISCLAIMER] Some members of my family where
slightly put out by these movements as well] He came up with the idea that
unless a theory has a negative hypothesis and is replicable, then it is not
scientific. In one sweep he had disqualified both Marxism,
and communism from claim of being scientific.

One of sciences that wiped out by this definition was Eugenics. Eugenics was
one of the academic justifications of Nazism. Another science to disappear was
biotype. This is the
idea that you could predict if somebody was a criminal by their body
measurement.

Back to Persona's and Liz's presentation. She gives an example about Tom. By
my count there are at least 21 bits of data points about Tom. Using the
example given by Chapman and Milham (which again uses at least 21 data
points.
http://cnchapman.files.wordpress.com/2007/03/chapman-milham-personas-hfes2006-0139-0330.pdf
That would show that "Under those assumptions, the composite data for
"Patrick" would represent (0.5)21 * 100% = 0.000048% of the population, or
approximately 134 people in the
United States." [If your target population (lets say Car Mechanics) is
smaller than that then the number of people your Persona
could represent approaches zero]

They go on further to say that :-
The key point is "there is essentially no way to generalize from
a well-specified persona to a population of interest, and thus no way to say
anything about the users of interest. There is no way to distinguish
which characteristics of a given persona are indicative of users and which
are irrelevant

The point is that unless you can show that you are designing for Users and
not something fictional then it is hard to call it User Centred Design

Is there a way out of the theory Trap. I think yes there is. Idea one is to
treat a design as a Hypothesis and test it. Idea Two is to go back to the
Sciences that have contributed methods to UCD, like Anthropology, and see
how they overcome some of the Theory Challenges. For example many people on
the list complain about the time that it takes to go through the research,
and to distil the ideas. Ethnography was developed as a descriptive language.
Or go to Activity Theory which is another descriptive process. If you use
either the language of Ethnography or the methods of AT it will save you
time. Forget about trying to get data to jump out at you. This is called
Grounded Theory and it is time consuming and very hard to follow correctly.
Also come up with some ideas before the research and then test them (in a
negative wayi.e.... my theory is not true if.....), again this will both
save you time and can be quite reliable.

All the best

James

On Tue, Nov 18, 2008 at 6:05 AM, Jarod Tang <jarod.tang at gmail.com> wrote:

> Hi James,
>
> > We are told so many times not to use us for our designs, ourselves, or
> our
> > mothers as the target for a design. But surely this is better than
> something
> > that is purely fictional.
> Persona is/should be based on user research data underneath (at least
> for design). This is defined from early practitioners like Alan Cooper
> (and he proved why the instantiation of persona should based on
> concrete user research in his books). To say it's fictional, one may
> miss the point of persona usage for design .
>
> Regards,
> Jarod
>
> --
> http://designforuse.blogspot.com/
>

19 Nov 2008 - 9:07am
Josh Seiden
2003

> Either a method is scientific or pseduscientific. There is no middle ground.

I just completely disagree with the way you're framing the conversation. Although some here may be making claims about personas being "scientific," I certainly do not make this claim.

People are confusing science with rigor. Rigorous thinking is not the sole province of the sciences. Rigorous thought is found in many domains: design, fiction, art, and yes, sometimes science as well.

So to say a method is either scientific or pseudoscientific may be true in the limited sense in which you're speaking. But it implies that every method aspires to science-and that which does not achieve science is by definition "fake science." That's hogwash. (Or perhaps more precisely, balderdash.) Many many creative methods exist that do not aspire to science. To call those methods "fake science" is as misleading as it is to call science "pseudo-creative." It's a characterization that completely misses the point.

I wish that more designers would acknowledge this about our methods.

JS

19 Nov 2008 - 9:49am
Jarod Tang
2007

Hi James,

> Hi All,
> I think it may help people here if I inject some theory into
> this discussion.
> The first point is that people keep making claims that the method has
> some scientific validity. For example Liz says that "Hopefully my quick
> elucidation about the original persona creation methodology helps you to see
> that the mapping of individuals to dimensions of interest is a relatively
> scientific method"
> Either a method is scientific or pseduscientific. There is no middle ground.

Josh already give his explanation on this, which I fully agree.

> The point is that unless you can show that you are designing for Users and
> not something fictional then it is hard to call it User Centred Design
> Is there a way out of the theory Trap. I think yes there is. Idea one is to
> treat a design as a Hypothesis and test it. Idea Two is to go back to the
> Sciences that have contributed methods to UCD, like Anthropology, and see
> how they overcome some of the Theory Challenges. For example many people on
> the list complain about the time that it takes to go through the research,
> and to distil the ideas. Ethnography was developed as a
> descriptive language. Or go to Activity Theory which is another descriptive
> process. If you use either the language of Ethnography or the methods of AT
> it will save you time. Forget about trying to get data to jump out at you.
> This is called Grounded Theory and it is time consuming and very hard to
> follow correctly. Also come up with some ideas before the research and then
> test them (in a negative wayi.e.... my theory is not true if.....), again
> this will both save you time and can be quite reliable.

For any domain, there's two layers, one at theory layer (laws/theory)
while the other at the application layer(methods, etc.) , for
interaction design, it's about the same. Persona is belongs to the
application domain.

As guys above ( include me ) said, the persona is based on the user
research (more than x% is y or blahblahblah ), and the key result is
the people's context, motivation, needs, experience and related
stuffs. And we adopt the persona to put the analyze result into some
design & communication
format, that's the major purpose of this application method. And it
avoid this too general elastic users ( which is exactly harm the real
UCD) in the whole design process.

>From Action Theory's perspective ( theory perspective), human and
outside world's relationship is realized by activity, And activity's
structure's key components are the human(subject), and his goal
(object), and possibly the artifact as the mediator ( physical or none
physical). persona is exactly
the application method to describe the human in the activity ( his
needs, motivation, constrains, experience ) , which turns the theory
explanation into practicable method.

We don't use the theory directly, instead there's many methods applied
everyday, persona is one of the good example.

Regards,
Jarod

--
http://designforuse.blogspot.com/

19 Nov 2008 - 4:02pm
Elizabeth Bacon
2003

I totally agree with Josh's recent point. Mea culpa; it was lazy
writing for me to say that persona creation is "relatively
scientific". What I meant to say was that the persona creation
process at its methodological origins involves a rigorous,
repeatable, analytical approach to research data. That doesn't make
it a true science; but not being a true science doesn't make it
invalid to design practice, which is a remarkable union of art and science, creativity and analysis.

Cheers,
Liz

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

19 Nov 2008 - 4:37pm
Christine Boese
2006

James,

I love where you're going here, and recourse to rigor (and Popper) seems
like a neat antidote to a lot of research methods (I'm thinking more about
the field of marketing, than anything) that get more than a bit sloppy in
their applied versions.

So I'm just nailing down some stuff at the end, cuz I wasn't sure exactly
where you were going, so forgive me if I've misunderstood what you wrote.
Might be good to tease it out a bit more, cuz that's where it gets
interesting.

Too often, reference to the "sciences" and properly rigorous research
methods equates (in some people's minds) to overly foundationalist
assumptions that require generalizable, quantitative data only. I don't
believe you are going there, from what you reference at the very end, but
that makes me want to push on where you are going.

Descriptive, rich, qualitative methods are by definition NOT generalizable.
That would be the whole point. One can inductively triangulate data, amass
evidence that reinforces emerging categories of data, develop heuristics,
and even conduct parallel studies and discover points of intersection
between similar qualitative or ethnographic-type studies.

So replicate to some extent, but generalize, never. True, people doing
multi-modal studies are trying to work with qualitative and quantitative
methods in tandem, so you may get some cross pollination there. Content
analysis, linguistics, these are rich areas for combining methods, again, to
triangulate, or to use emerging qualitative data to develop quantitative
hypotheses.

Quantitative heads tend toward more restrictive, or limited definitions of
what is "real" research, which methods are most rigorous, yield the best
data, and so on. They like to tout generalizability as some kind of Holy
Grail that only they can claim, like it gives them some kind of
foundationalist claim to capital T Truth. Blah.

Not all of them think like this, however. You tend to get that kind of POV
more often in journalists who write stories about "science," and bias their
coverage toward methods they more readily grasp or can easily convert into
sound bites. (you'd be amazed at how widespread this POV is among
journalists at CNN, for instance--I can speak from experience)

To this end, then, journalists unconsciously tend to reinforce
misconceptions about real research, real science, which mushes up the whole
pseudo-science problem more, as they often tout as authoritative
quantitative studies that are so severely limited and short-sighted in their
hypothesis development or baseline assumptions as to render their so-called
valid and generalizable data utterly worthless.

Which is worse: to have a method that is rigorously generalizable, but
doesn't actually fit what we find in the world, or a rich data set which
hews closely to the actual behaviors of actual people with deep insight and
understanding, but should not ever be generalized beyond that level of
detail?

Which is closer to real small-t truths?

Chris

On Wed, Nov 19, 2008 at 9:49 AM, James Page <jamespage at gmail.com> wrote:

> Hi All,
> I think it may help people here if I inject some theory into
> this discussion.
>
> The first point is that people keep making claims that the method has some
> scientific validity. For example Liz says that "Hopefully my quick
> elucidation about the original persona creation methodology helps you to
> see
> that the mapping of individuals to dimensions of interest is a relatively
> scientific method"
>
> Either a method is scientific or pseduscientific. There is no middle
> ground.
> Is the distinction only important as academic argument? The answer is no,
> and it helps to understand a small bit of history to see why.
>
> The distinction between science or psedu science came about because there
> where two political movements that claimed that they where scientific, and
> by following them would lead to improvement to everybody in society. The
> two
> political movements where communism, and national socialism.
>
> Karl Poppers, who you could say his early life was upturned by both
> movements, thought that it is important to qualify what is scientific and
> what is pseduscientific. [DISCLAIMER] Some members of my family where
> slightly put out by these movements as well] He came up with the idea that
> unless a theory has a negative hypothesis and is replicable, then it is not
> scientific. In one sweep he had disqualified both Marxism,
> and communism from claim of being scientific.
>
> One of sciences that wiped out by this definition was Eugenics. Eugenics
> was
> one of the academic justifications of Nazism. Another science to disappear
> was
> biotype. This is the
> idea that you could predict if somebody was a criminal by their body
> measurement.
>
> Back to Persona's and Liz's presentation. She gives an example about Tom.
> By
> my count there are at least 21 bits of data points about Tom. Using the
> example given by Chapman and Milham (which again uses at least 21 data
> points.
>
> http://cnchapman.files.wordpress.com/2007/03/chapman-milham-personas-hfes2006-0139-0330.pdf
> That would show that "Under those assumptions, the composite data for
> "Patrick" would represent (0.5)21 * 100% = 0.000048% of the population, or
> approximately 134 people in the
> United States." [If your target population (lets say Car Mechanics) is
> smaller than that then the number of people your Persona
> could represent approaches zero]
>
> They go on further to say that :-
> The key point is "there is essentially no way to generalize from
> a well-specified persona to a population of interest, and thus no way to
> say
> anything about the users of interest. There is no way to distinguish
> which characteristics of a given persona are indicative of users and which
> are irrelevant
>
> The point is that unless you can show that you are designing for Users and
> not something fictional then it is hard to call it User Centred Design
>
> Is there a way out of the theory Trap. I think yes there is. Idea one is to
> treat a design as a Hypothesis and test it. Idea Two is to go back to the
> Sciences that have contributed methods to UCD, like Anthropology, and see
> how they overcome some of the Theory Challenges. For example many people on
> the list complain about the time that it takes to go through the research,
> and to distil the ideas. Ethnography was developed as a descriptive
> language.
> Or go to Activity Theory which is another descriptive process. If you use
> either the language of Ethnography or the methods of AT it will save you
> time. Forget about trying to get data to jump out at you. This is called
> Grounded Theory and it is time consuming and very hard to follow correctly.
> Also come up with some ideas before the research and then test them (in a
> negative wayi.e.... my theory is not true if.....), again this will both
> save you time and can be quite reliable.
>
> All the best
>
> James
>
> On Tue, Nov 18, 2008 at 6:05 AM, Jarod Tang <jarod.tang at gmail.com> wrote:
>
> > Hi James,
> >
> > > We are told so many times not to use us for our designs, ourselves, or
> > our
> > > mothers as the target for a design. But surely this is better than
> > something
> > > that is purely fictional.
> > Persona is/should be based on user research data underneath (at least
> > for design). This is defined from early practitioners like Alan Cooper
> > (and he proved why the instantiation of persona should based on
> > concrete user research in his books). To say it's fictional, one may
> > miss the point of persona usage for design .
> >
> > Regards,
> > Jarod
> >
> > --
> > http://designforuse.blogspot.com/
> >
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

20 Nov 2008 - 6:11am
Terry Fitzgerald
2008

I am a Product Designer (UX Designer) for a major international software
> development company. One of my roles is the creation/updating of personas.
> We have about 185 of them now. The primary value of them, at least in our
> organization, is to provide a relatable reference to our intended users. Our
> personas are used not only by the Developers but also sales and marketing
> and the folks who write the documentation.
>

There has been a lot of chatter in this thread about "scientific" methods
etc but one of the things I did not see was the use of real person
interviews to vette the personas. We do this quite regularly with our
customers, our sales and support teams etc to ask them if the prersonas are
representative of real people they know and work with. We use this feedback
to round out the persona design.

It is an amazing tool for dialoging with end users about their needs because
in the end that is the purpose of the persona. It gives the customer/user
the impression that you are asking about them not about processes. Personas
can also be an effective tool for analtysing/designing workflows because the
conversation shifts from boxes and arrows to the integration of people in
the organziation with when and how they are introduced into a process. The
whole design exercise takes on a new flavour.

--
Regards

Terry
http://www.linkedin.com/in/terryfitzgerald
BLOG: http://hitsandmesses.blogspot.com

24 Nov 2008 - 9:25am
James Page
2008

Chris,
You hit what I was trying to say on the nail. There are
the qualitative science and the quantitative sciences. With USD we are
mixing the two, and I have no problems with the two forms been mixed.

The issue is as you put is people being "sloppy" in application.

Your argument was far clearer than the one I was making.

Yes, I was spurred on by the debate from the Marketing world, about the poor
performance of certain research methods. If people have not been following
the argument, it is that certain research methods restrict creative thought
and a good idea could be thrown out because research shows it will not work.
Examples given for this argument include the "Dove campaign for real beauty"
(which shows real women as beautiful, instead of models), the walkman,
etc.... The proponents of the argument include some largest names in
Marketing including Uni Leaver, the vice chairman of Ogivy... etc....

They are not making an argument that research is bad, just that some forms
are harmful. For a good posting on the subject see:
http://lbtoronto.typepad.com/lbto/2007/05/pretesting.html

I was not trying to have a foundationalist argument, but just to defend
Popper a tiny bit, he was not against the social sciences. What he was doing
was showing that t(truth) could not be proven. That something could
only shown not to be true. There is only negative truth, and no way to prove
something positively.

What all science is trying to do build the evidence up, it is never
complete. Both approaches can help build that evidence up. But they must be
used in a safe manner, otherwise we will throw away the next Dyson
vacuum cleaner,
or the next Walkman.

Your example of Journalism is that what gets reported is the findings
without the caveats. There are those two nasty errors in statistics, Type 1
and Type II. What you read in the newspaper is that the Turkey gets fed
every day. You don't read that there is a 1 in 365 chance of that finding
been wrong.

>From the foundationalist view Persona are unsafe, and I think you are making
an argument that from the non-foundationalist view that they are unsafe.

I don't believe you are going there, from what you reference at the very
> end, but that makes me want to push on where you are going.

I am all for good research and it can help inform the development of a
product. The challenge is that all the methods have limitations that need to
be known. If the theory side of USD can developed it would strengthen the
industry, and would reduce the risk of bad research creating bad products.

With Ethnography (and my only qualification here is that my father was an
Anthropologist) I believe we need to look more closely in how we present the
rich data collected from the participants (not fictionalised), and also use
more participant observers (immersion with the participants), use informers.
Keep studying the participants while the product is been developed. Do not
think of it as waterfall process but an ongoing one. Ethnography is about
how people relate to each other and objects. If we are developing a new
object then the peoples interaction with the new object will change. How do
you change the interaction between the personas, how do you know that you
have got it correct?

Is there a way of designer immersion into the participants while the
designers are building the product?

James

On Wed, Nov 19, 2008 at 10:37 PM, Christine Boese <christine.boese at gmail.com
> wrote:

> James,
>
> I love where you're going here, and recourse to rigor (and Popper) seems
> like a neat antidote to a lot of research methods (I'm thinking more about
> the field of marketing, than anything) that get more than a bit sloppy in
> their applied versions.
>
> So I'm just nailing down some stuff at the end, cuz I wasn't sure exactly
> where you were going, so forgive me if I've misunderstood what you wrote.
> Might be good to tease it out a bit more, cuz that's where it gets
> interesting.
>
> Too often, reference to the "sciences" and properly rigorous research
> methods equates (in some people's minds) to overly foundationalist
> assumptions that require generalizable, quantitative data only. I don't
> believe you are going there, from what you reference at the very end, but
> that makes me want to push on where you are going.
>
> Descriptive, rich, qualitative methods are by definition NOT generalizable.
> That would be the whole point. One can inductively triangulate data, amass
> evidence that reinforces emerging categories of data, develop heuristics,
> and even conduct parallel studies and discover points of intersection
> between similar qualitative or ethnographic-type studies.
>
> So replicate to some extent, but generalize, never. True, people doing
> multi-modal studies are trying to work with qualitative and quantitative
> methods in tandem, so you may get some cross pollination there. Content
> analysis, linguistics, these are rich areas for combining methods, again, to
> triangulate, or to use emerging qualitative data to develop quantitative
> hypotheses.
>
> Quantitative heads tend toward more restrictive, or limited definitions of
> what is "real" research, which methods are most rigorous, yield the best
> data, and so on. They like to tout generalizability as some kind of Holy
> Grail that only they can claim, like it gives them some kind of
> foundationalist claim to capital T Truth. Blah.
>
> Not all of them think like this, however. You tend to get that kind of POV
> more often in journalists who write stories about "science," and bias their
> coverage toward methods they more readily grasp or can easily convert into
> sound bites. (you'd be amazed at how widespread this POV is among
> journalists at CNN, for instance--I can speak from experience)
>
> To this end, then, journalists unconsciously tend to reinforce
> misconceptions about real research, real science, which mushes up the whole
> pseudo-science problem more, as they often tout as authoritative
> quantitative studies that are so severely limited and short-sighted in their
> hypothesis development or baseline assumptions as to render their so-called
> valid and generalizable data utterly worthless.
>
> Which is worse: to have a method that is rigorously generalizable, but
> doesn't actually fit what we find in the world, or a rich data set which
> hews closely to the actual behaviors of actual people with deep insight and
> understanding, but should not ever be generalized beyond that level of
> detail?
>
> Which is closer to real small-t truths?
>
> Chris
>
> On Wed, Nov 19, 2008 at 9:49 AM, James Page <jamespage at gmail.com> wrote:
>
>> Hi All,
>> I think it may help people here if I inject some theory into
>> this discussion.
>>
>> The first point is that people keep making claims that the method has some
>> scientific validity. For example Liz says that "Hopefully my quick
>> elucidation about the original persona creation methodology helps you to
>> see
>> that the mapping of individuals to dimensions of interest is a relatively
>> scientific method"
>>
>> Either a method is scientific or pseduscientific. There is no middle
>> ground.
>> Is the distinction only important as academic argument? The answer is no,
>> and it helps to understand a small bit of history to see why.
>>
>> The distinction between science or psedu science came about because there
>> where two political movements that claimed that they where scientific, and
>> by following them would lead to improvement to everybody in society. The
>> two
>> political movements where communism, and national socialism.
>>
>> Karl Poppers, who you could say his early life was upturned by both
>> movements, thought that it is important to qualify what is scientific and
>> what is pseduscientific. [DISCLAIMER] Some members of my family where
>> slightly put out by these movements as well] He came up with the idea that
>> unless a theory has a negative hypothesis and is replicable, then it is
>> not
>> scientific. In one sweep he had disqualified both Marxism,
>> and communism from claim of being scientific.
>>
>> One of sciences that wiped out by this definition was Eugenics. Eugenics
>> was
>> one of the academic justifications of Nazism. Another science to disappear
>> was
>> biotype. This is the
>> idea that you could predict if somebody was a criminal by their body
>> measurement.
>>
>> Back to Persona's and Liz's presentation. She gives an example about Tom.
>> By
>> my count there are at least 21 bits of data points about Tom. Using the
>> example given by Chapman and Milham (which again uses at least 21 data
>> points.
>>
>> http://cnchapman.files.wordpress.com/2007/03/chapman-milham-personas-hfes2006-0139-0330.pdf
>> That would show that "Under those assumptions, the composite data for
>> "Patrick" would represent (0.5)21 * 100% = 0.000048% of the population, or
>> approximately 134 people in the
>> United States." [If your target population (lets say Car Mechanics) is
>> smaller than that then the number of people your Persona
>> could represent approaches zero]
>>
>> They go on further to say that :-
>> The key point is "there is essentially no way to generalize from
>> a well-specified persona to a population of interest, and thus no way to
>> say
>> anything about the users of interest. There is no way to distinguish
>> which characteristics of a given persona are indicative of users and which
>> are irrelevant
>>
>> The point is that unless you can show that you are designing for Users and
>> not something fictional then it is hard to call it User Centred Design
>>
>> Is there a way out of the theory Trap. I think yes there is. Idea one is
>> to
>> treat a design as a Hypothesis and test it. Idea Two is to go back to the
>> Sciences that have contributed methods to UCD, like Anthropology, and see
>> how they overcome some of the Theory Challenges. For example many people
>> on
>> the list complain about the time that it takes to go through the research,
>> and to distil the ideas. Ethnography was developed as a descriptive
>> language.
>> Or go to Activity Theory which is another descriptive process. If you
>> use
>> either the language of Ethnography or the methods of AT it will save you
>> time. Forget about trying to get data to jump out at you. This is called
>> Grounded Theory and it is time consuming and very hard to follow
>> correctly.
>> Also come up with some ideas before the research and then test them (in a
>> negative wayi.e.... my theory is not true if.....), again this will both
>> save you time and can be quite reliable.
>>
>> All the best
>>
>> James
>>
>> On Tue, Nov 18, 2008 at 6:05 AM, Jarod Tang <jarod.tang at gmail.com> wrote:
>>
>> > Hi James,
>> >
>> > > We are told so many times not to use us for our designs, ourselves, or
>> > our
>> > > mothers as the target for a design. But surely this is better than
>> > something
>> > > that is purely fictional.
>> > Persona is/should be based on user research data underneath (at least
>> > for design). This is defined from early practitioners like Alan Cooper
>> > (and he proved why the instantiation of persona should based on
>> > concrete user research in his books). To say it's fictional, one may
>> > miss the point of persona usage for design .
>> >
>> > Regards,
>> > Jarod
>> >
>> > --
>> > http://designforuse.blogspot.com/
>> >
>> ________________________________________________________________
>> Welcome to the Interaction Design Association (IxDA)!
>> To post to this list ....... discuss at ixda.org
>> Unsubscribe ................ http://www.ixda.org/unsubscribe
>> List Guidelines ............ http://www.ixda.org/guidelines
>> List Help .................. http://www.ixda.org/help
>>
>
>

25 Nov 2008 - 10:37am
Jon-Eric Steinbomer
2003

Not to beat a dead (or dying?) horse, but you might enjoy Steve
Portigal's article in interactions, "Persona Non Grata" [the piece
mirrors his presentation content of "Personas are User Centered
Bullshit"] - http://interactions.acm.org/content/?p=262 for more
details.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=35624

25 Nov 2008 - 8:25pm
Jarod Tang
2007

There's a presentation "Death To Personas! Long Live Personas!" to
(partially) answer the "Personas are User Centered "
http://www.slideshare.net/ebacon/death-to-personas-long-live-personas-presentation

It's a well done.

Regards,
Jarod

On Wed, Nov 26, 2008 at 12:37 AM, Jon [ GMAIL ] <jkolko at gmail.com> wrote:
> Not to beat a dead (or dying?) horse, but you might enjoy Steve
> Portigal's article in interactions, "Persona Non Grata" [the piece
> mirrors his presentation content of "Personas are User Centered
> Bullshit"] - http://interactions.acm.org/content/?p=262 for more
> details.
>
>
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=35624
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

--
http://designforuse.blogspot.com/

Syndicate content Get the feed