3.
About
Roomie
Roomie
is
an
online
pla,orm
that
aims
at
helping
university
students
find
suitable
roommates
in
Hong
Kong
according
to
their
personality,
life
style
and
other
daily
issues.
3
1.Introduction
Target
popula4on
of
Roomie
While
the
primary
users
of
Roomie
are
students
from
Mainland
China,
as
most
of
them
cannot
apply
for
the
university
dormitories,
we
also
prefer
to
target
graduates
who
determine
to
work
in
Hong
Kong
but
lack
roommates
to
share
an
apartment,
as
well
as
local
students
who
prefer
to
live
separately
from
parents.
4. Goal
of
usability
tes4ng
Our
goal
for
the
usability
tesCng
was
to
collect
both
quanCtaCve
and
qualitaCve
data
of
new
users
learning
how
to
use
Roomie
to
search
for
potenCal
roommates
for
the
first
Cme.
We
hope
that
the
results
of
the
usability
tesCng
can
be
used
by
us
to
understand
the
strengths
and
weaknesses
of
this
project
and
serve
as
a
benchmark
for
future
usability
tesCng.
4
1.Introduction
5. Overview
Usability
tesCng
occurred
during
the
one-‐week
span
of
March
9-‐16.
We
divided
our
tesCng
into
two
rounds,
the
first
round
on
March
9
and
the
second
round
on
March
16.
ALer
the
first
round
tesCng,
we
opCmized
our
prototype
according
to
parCcipants’
suggesCons,
then
held
the
second
round
tesCng
one
week
aLer.
Ten
parCcipants
were
recruited
who
had
never
heard
of
our
project
before
but
yet
had
a
need
to
look
for
roommates,
or
had
related
roommate-‐finding
experiences.
During
the
half-‐an-‐hour
tesCng
sessions,
each
parCcipant
was
asked
to
do
the
task
of
searching
for
suitable
roommates
on
Roomie.
Time
spent
on
the
task
and
qualitaCve
comments
were
collected
for
each
parCcipant.
Each
session
was
recorded
using
QuickTime
to
allow
for
later
analysis.
5
2.
Methodology
6. Prototype Building
We used Axure to make the prototype of Roomie online
platform. Axure is a business drawing software for creating
mockups and wireframes of application user interfaces.
This easy to use software generates HTML prototypes and
functional specifications.
6
2.
Methodology
7. Par4cipant
Selec4on
&
Recruitment
Since
our
goal
was
to
find
the
strengths
and
weaknesses
of
this
project,
we
recruited
users
who
had
never
heard
Roomie
before
but
had
a
need
to
look
for
roommates,
or
had
related
roommate-‐finding
experiences.
They
should
be
future
students,
present
students
or
graduates
of
a
certain
university
in
Hong
Kong.
We
recruited
our
parCcipants
through
our
social
networks,
and
offered
them
a
small
giL
as
an
incenCve
for
parCcipaCon.
7
2.
Methodology
8. Test
Protocol
Each
test
session
last
for
one
hour,
including
10
minutes
for
a
pre-‐test
background
interview
and
post-‐test
debriefing
quesCons,
and
20
minutes
for
the
tasks.
We
began
each
test
session
by
asking
them
some
general
quesCons
about
how
they
got
along
with
their
former
roommates,
whether
they
had
some
good
or
bad
experiences,
and
what
they
cared
the
most
when
looking
for
a
roommate.
Then
we
gave
them
two
tasks,
and
asked
them
to
complete
each
one
while
talking
aloud
about
their
thought
process
and
any
confusion
they
encountered.
8
2.
Methodology
9. Test
Protocol
The
two
tasks
we
gave
them
were
organized
by
the
sequenCal
order.
The
summary
of
tasks
is
listed
below.
Task Number
Task Summary
1
Register for Roomie
2
Search for roommates on Roomie
9
2.
Methodology
ALer
the
parCcipant
completed
the
tasks,
we
conducted
a
post-‐test
brief,
asking
them
about
their
interesCng
or
confused
points
about
Roomie.
We
also
asked
them
to
score
each
page
according
to
their
saCsfacCon
(4-‐point
Likert
scale,
from
1
=dissaCsfied
to
4=saCsfied)
and
asked
if
they
thought
they
would
be
likely
to
use
Roomie,
based
on
their
iniCal
impression.
10. Test
Environment
and
Equipment
The
tests
were
conducted
in
Learning
Commons
of
Wu
Ho
Man
Yuen
Building.
We
used
QuickTime
to
record
our
test
screens.
We
each
took
turns
moderaCng
one
session
and
observing
one
session.
We
set
up
QuickTime
on
a
Mac,
and
give
this
Mac
to
parCcipant
to
use
during
the
test.
10
2.
Methodology
Data
Collec4on
We
collected
both
qualitaCve
and
quanCtaCve
data
about
the
parCcipants’
behavior.
Our
quanCtaCve
data
consisted
of
the
performance
data
of
Cme
spent
on
tasks
for
each
parCcipant,
and
their
saCsfacCon
(4-‐point
Likert
scale)
towards
Roomie.
The
qualitaCve
data
we
collected
were
quotes,
comments
and
observaCons
noted
down
by
the
observer
during
the
test.
11. Score
analysis:Test
1*
3.3
3.6 3.6
2.5
3 3.1
2.8
0
0.5
1
1.5
2
2.5
3
3.5
4
登
陆
注
册
页
邮
箱
注
册
填
写
信
息
个
人
主
页
首
页他
人
的
个
人
主
页
加
好
友
Tes t 1均分
Test1均分
11
3.Results
*Test 1 stands for the first-round test. After the first round, we optimized our
prototype and conducted Test 2, which is the second–round test.
14. Time
spent
on
task:
Test
1
&
Test
2
T ime spent on task
0
5
10
15
20
25
1 2 3 4 5
Minutes
Test1
Test2
14
3.Results
From
this
chart
we
can
see
clearly
that
aLer
opCmizing
the
prototype,
parCcipants
spent
less
Cme
on
compleCng
tasks
than
the
first
Cme.
It
means
that
now
our
prototype
is
easier
to
understand,
and
the
searching
process
is
clearer
than
before.
15. Findings
In
the
first-‐round
test,
we
found
5
broken
experiences
and
figured
out
soluCons
for
each
broken
experience.
In
the
second-‐
round
test,
we
found
8
problems
that
needed
to
be
solved.
15
3.Results
Prototype
V1.0
Prototype
V2.0
Prototype
V3.0
Fixed 5 problems
from Test 1
Fixed 8 problems
from Test 2
16.
3
person
points
out
that
the
picture
on
the
first
page
is
not
appropriate.
“The
picture
in
registraCon
page
cannot
clearly
represent
the
funcCon
of
Roomie.”
“The
picture
make
the
website
look
like
a
social
media.”
16
3.Results Test 1
Broken
experience
1
17. SoluCon
1
Changed
to
a
more
clear
and
high
quality
picture,
cannot
find
a
suitable
picture
with
Asian
faces.
17
3.Results Test 1
18. Broken
experience
2
5
users
menConed
that
the
design
of
user’s
own
informaCon
page
is
not
clean
and
clear.
“I
find
the
table
very
stressful
and
the
classificaCon
of
informaCon
is
not
clear.
”
18
3.Results Test 1
19. SoluCon
2
Change
table
to
listed
informaCon,
add
more
sucCons
in
user’s
own
informaCon
page.
19
3.Results Test 1
20. 3
person
points
out
that
“The
advanced
search
is
confusing”,
“
I
do
not
understand
the
relaConship
between
apartment
and
roommate.”
20
3.Results Test 1
Broken
experience
3
21. SoluCon
3
Change
boxes
to
listed
opCons
to
make
it
easier
to
understand,
and
also
add
missed
bufons.
21
3.Results Test 1
22. 4
users
menConed
that
more
personal
informaCon
such
as
lifestyle
is
necessary.
“I
hope
to
see
more
personalized
informaCon
to
know
what
kind
of
person
the
potenCal
roommate
is.”
22
3.Results Test 1
Broken
experience
4
23. SoluCon
4
Provide
choices
related
to
lifestyle
and
apartment
requirements
.
23
3.Results Test 1
24. 3
person
points
out
that
the
whole
process
of
registraCon
and
search
is
not
very
smooth,
some
bufons
are
missing.
“I
think
there
should
be
a
next
or
finish
bufon.”
“
I
can’t
go
back
to
the
homepage.”
24
3.Results Test 1
Broken
experience
5
25. SoluCon
5
Adding
navigaCon
at
the
top
of
each
page,
and
change
“submit”
to
“finish”.
25
3.Results Test 1
26. After we fixed problems from Test 1, we updated our
prototype from V1.0 to V2.0 and conducted Test 2.
26
Prototype
V1.0
Prototype
V2.0
Fixed 5 problems
from Test 1
27. In
the
“fill-‐in
personal
informaCon”
page,
we
want
to
collect
more
informaCon
about
the
user.
But
3
users
complained
that
they
don’t
want
to
fill
in
all
the
informaCon.
“Some
informaCon
I
don’t
want
to
share
to
others.
It’s
all
required?
”
27
3.Results Test 2
Broken
experience
1
28. We
add
“*”
in
front
of
those
mandatory
fields
to
disCnguish
from
opConal
ones,
so
as
users
can
clearly
figure
out
which
part
they
must
fill
in.
28
3.Results Test 2
SoluCon
1
29. 2
users
pointed
out
that
they
don’t
want
to
type
in
the
descripCons
of
their
characters,
or
it
took
some
Cme
to
think
about
how
to
describe
themselves.
So
there
is
a
high
possibility
that
they
will
leave
in
this
stage.
“I
don’t
want
to
type,
I
want
to
choose.”
29
3.Results Test 2
Broken
experience
2
30. Besides
words
to
describe
the
characterisCc,we
first
provide
menu
for
users
to
choose
their
character
type.
30
3.Results Test 2
SoluCon
2
31. 31
2
person
points
out
that
they
can
not
understand
the
word
‘洋楼’and‘唐楼’,
and
actually
the
type
of
the
building
is
not
that
important.
Moreover,
the
type
of
the
apartment
is
something
that
they
concern
more.
“I
don’t
know
how
to
choose, I
have
no
idea
what
it
is.”
3.Results Test 2
Broken
experience
3
32. To
be
more
clear,
we
change
“唐楼”
and
“洋楼”
to
“无电梯
楼宇”
and
“有电梯楼宇”.
We
also
add
Type
of
Apartment
to
provide
more
details.
32
3.Results Test 2
SoluCon
3
33. ALer
registraCon
process,
user
will
landing
to
personal
homepage,
but
there
is
no
instrucCons
for
what
to
do
next.
“OK,
I’ve
registered.
So
what
I
goanna
do
next?
”
33
3.Results Test 2
Broken
experience
4
34. SoluCon
4
Besides
words
to
describe
the
characterisCc,we
first
provide
menu
for
users
to
choose
their
character
type.
34
3.Results Test 2
35. RecommendaCon
funcCon
on
the
homepage
is
overlooked
by
users.
“See
I
can
search
the
website.
If
there
are
recommended
roommates
for
me,
that
would
be
great!”
35
3.Results Test 2
Broken
experience
5
36. SoluCon
5
We
put
recommendaCon
part
on
the
top
of
homepage
to
reinforce
its
funcCon,
and
put
search
bar
on
the
right
side.
If
users
want
to
search
more
informaCon,
they
can
click
Advanced
Search
and
go
to
search
page.
36
3.Results Test 2
37. 37
2 person
points
out
that
when
searching,
they
put
gender
in
priority.
“Where
is
the
gender
in
the
search
box?
I
would
definitely
want
that
opCon
before
all
the
other
choices.
”
NO GENDER!
3.Results Test 2
Broken
experience
6
38. We
rearrange
the
choices
and
add
“同一性别”
so
as
it
can
be
seen
more
easily.
38
3.Results Test 2
SoluCon
6
39. In
the
process
of
finding
search
funcCon,
user
click
“people
I
interested
in”
below
head
portrait
instead
of
“search”
bufon
on
the
header.
However, “people
I
interested
in”
is
not
the
right
search
bufon.
There
is
some
misunderstanding
of
“people
I
interested
in”.
“I’ll
click
this(‘people
I
interested
in’)
to
see
who
is
my
suitable
roommate.
39
3.Results Test 2
Broken
experience
7
40. We
build
list
page
of
users
who
I
am
interested
in
aLer
click
“people
I
interested
in
”
bufon.
Users
can
check
and
manage
all
the
list
here.
40
3.Results Test 2
SoluCon
7
41. 41
2 person
points
out
that
they
can’t
understand
what
grouping
means
when
I
want
to
leave
a
note
to
other
users.
“I
don’t
know
what
the
grouping
here
means,
will
it
appear
every
Cme
I
leave
message
for
someone?
Will
other
people
see
my
grouping
informaCon?
Would
it
be
more
appropriate
if
it
shows
when
I
liked
someone?”
3.Results Test 2
Broken
experience
8
42. We
revise
Grouping
facCon,
add
a
drop
list
for
users
to
select
and
add
group
category.
42
3.Results Test 2
SoluCon
8
43. Further
demand
of
renCng
a
house
can
not
be
fulfilled
in
our
website
“Besides
finding
a
roommate,
I
want
to
find
a
flat.
If
there
is
only
informaCon
of
roommates,
I
think
that’s
not
enough.”
SoluCon
Unfortunately,we
cannot
solve
this
problem
so
far.
43
3.Results Test 2
Broken experience 9
44. The
tasks
are
about
the
basic
funcCons
of
Roomie.
Most
users
successfully
finished
all
the
tasks.
All
of
them
experienced
some
difficulCes
but
all
successfully
figured
out
a
way
to
overcome
them.
From
the
first-‐round
test
we
found
many
weaknesses
of
Roomie
and
opCmized
it
according
to
parCcipants’
suggesCons.
And
in
the
second-‐round
test,
though
Roomie
had
been
opCmized,
parCcipants
sCll
figured
out
some
broken
experiences
that
need
to
be
solved.
It
means
that
to
saCsfied
users’
real
needs,
we
sCll
have
a
lot
to
do.
44
4. Conclusion
45. • Roomie usability test video
• Prototype (in the folder “V3.0/ index.html”)
45
5. Appendix