Lecture 3 (September 18 , 2012) th Morad Moazami
The
test
on
October
11
is
multiple
choice.
It
will
be
60
multiple
choice
questions,
and
there
will
be
ne
hour
and
thirty
five
minutes
to
complete
the
exam.
Descriptive
vs.
Correlational
Studies:
Descriptive
Studies:
These
studies
involve
observing
and
classifying
behavior.
For
example,
Shriley
Brice
Heath
in
the
1970s,
went
and
stayed
with
300
families
over
a
period
of
time
and
found
that
middle
class
tended
to
include
their
children
in
their
daily
relationships
more
than
the
lower-‐class
families.
And
this
was
reflected
in
the
middle-‐classes
grades
in
school
later
on.
Festinger
and
a
number
of
different
psychologists
in
1956
wanted
to
study
cognitive
dissonance.
That
means
that
if
your
attitude
doesn't
mash
your
behavior,
you
will
think
two
things
that
are
hard
to
believe.
So
they
went
and
stayed
with
this
cult,
called
“The
Seekers”
and
their
leader
was
Marian
Keech.
They
believed
that
on
December
aliens
would
come
down,
and
Keech
would
get
messages
from
this
alien
race,
and
they
believed
that
the
world
would
end.
So
the
psychologists
embedded
themselves
within
this
group,
because
they
were
really
interested
in
what
would
happen
when
this
spaceship
wouldn't
come.
The
world
didn't
end.
Flying
saucers
didn't
come,
and
they
didn't
stop
believing,
instead,
Mrs.
Keech
said
that
she
had
received
a
message
from
the
aliens,
and
that
because
of
their
faith,
they
had
saved
the
universe.
So
they
believed
even
more
strongly
in
their
religion.
In
psychology,
these
kinds
of
studies
are
usually
done
in
the
first
steps
of
research
or
they
are
part
of
a
larger
research
project,
so
these
dissonance
project
people
had
this
chance
to
go
out
in
the
real
world
and
find
cognitive
dissonance
in
real
life.
So
these
steps
add
to
more
rigorous
testing.
Correlational
Studies:
In
correlational
studies,
we
are
specifically
examining
how
variables
are
related
to
one
another.
This
is
a
kind
of
thing
we
do
in
an
experiment.
The
only
difference
is
that
in
correlational
studies,
the
researcher
isn't
manipulating
any
of
the
variables.
For
example,
there
is
research
that
has
gotten
a
lot
of
media
attention
that
links
depression
and
cellphones
to
adolescence,
saying
that
cellphones
and
young
adults
are
related
to
teens
being
more
depressed.
These
are
correlational
studies,
because
they
obviously
aren't
manipulating
these
kinds
of
things.
What
is
unclear
here
is
if
the
kids
who
have
problems
sleeping
to
begin
with
go
to
their
phones
to
try
to
connect
to
other
people,
etc.
It
is
unclear
what
variable
is
cuasing
changes
to
other
variables.
We
cant
say
that
cellphone
use
causes
depression
in
teens.
We
can
just
say
that
teenagers
that
are
more
depressed
use
their
phones
more
often.
Another
example
is
that
self-‐esteem
is
linked
to
academic
success
and
vice-‐versa.
Correlational
studies
allow
researchers
to
make
claims
about
associates
between
variables,
but
not
causal
claims.
You
cannot
make
claims
that
one
variable
is
causing
changes
in
another
variable.
The
Third
Variable
Problem
and
Confounds:
The
third-‐variable
problem:
Is
specific
to
correlational
research,
because
it
arises
when
researchers
cannot
manipulate
the
variable
they
believe
is
causing
changes
in
another
variable.
For
example,
they
find
out
that
children
who
attend
pre-‐school
have
better
reading
skills,
but
there
could
also
be
a
third
variable
that
is
explaining
changes
in
what
we’re
looking
at.
Confounds
are
thought
in
the
context
of
an
experiment.
You
can
think
of
it
as
a
third
variable
you
are
not
particularly
studying.
It
is
something
else
aside
from
your
independent
variable
that
is
linked
to
your
experiment
–
they
provide
alternative
explanations
to
the
changes
in
your
work.
It
is
a
bad
thing,
because
it
is
between
your
experimental
changes.
For
example,
you’re
interested
in
alcohol
in
regards
to
driving.
You
have
a
group
of
drunk
people
and
non-‐drunk
people,
and
because
you’re
worried
about
drunk
people
driving,
you
give
them
smart
cars,
and
you
give
the
non-‐drunks
SUVS.
This
is
a
weak
study,
because
you
have
confounds
(another
type
of
variable:
the
type
of
car
they
are
driving)
that
changes
the
experiment
entirely.
Good
Research
Requires
Data
That
Is…
Good
research
requires
data
that
is:
Accurate,
valid,
and
reliable.
Accuracy:
Whatever
it
is
you
are
measuring
must
be
accurate.
You
have
two
kinds
of
error:
Random
error
and
systematic
error.
Random
error,
for
example,
is
when
your
timer
is
working,
but
you
make
minor
errors
in
regards
to
timing
an
experiment,
for
example,
but
in
the
end,
it
all
cancels
out
,
because
some
will
be
too
long,
some
too
short,
and
therefore,
it
all
equals
out
in
the
end,
and
won’t
really
affect
the
experiment.
Random
error
is
typically
not
so
bad.
You
want
to
be
as
accurate
as
possible,
but
random
error
is
not
as
bad
as
systematic
error,
because
that
way,
there
is
a
problem
with
one
of
your
variables.
So
for
every
single
trial,
of
for
example
timing
an
experiment,
if
the
clock
doesn't
work,
it’s
not
going
to
correct
itself
out
with
multiple
trials
like
random
error
does.
Valid:
validity
refers
to
the
extent
to
which
the
collected
data
address
the
research
hypothesis
in
the
way
intended.
It
is
the
question
of
whether
you’re
measuring
what
you
mean
to
measure.
Our
question
is
whether
our
processing
speed
increases
over
age.
So
our
hypothesis
is
that
university
students
have
faster
reaction
times
in
the
Stroop
test
compared
to
elementary
students.
So
we
take
data
from
university
students
and
we
take
date
from
elementary
school
students.
But
if
our
question
was
whether
university
students
enjoy
completing
the
Stroop
test
more
as
they
age,
then
our
question
would
be
whether
anyone
would
enjoy
doing
the
Stroop
test.
The
point
is
that
More
Less
