in general a researcher chooses a measurement scale
that influences the course of operationalisations
a
study :
nominal : unordered
categories
>
-
&
> ordinal : ordered categories ,
unequal intervals
-
rational :
equal intervals and zero means zero
>
-
interval : equal intervals and no zero
quantitative data ,
a measure can be described with
>
-
discrete (full) and continuous (between) numbers
choice for a plot depends on the measurement scale
categorical bar graph -
pie -
or
-
chart
anantitative en
myogram or boxplot -
histogram :
·
every bar is a bin
·
height is number of observations
-
in the bin
!
difference with bar graphs
·
no
gaps
:
describing data in a
graph
nalk Center spread
I adad I
shape
area& #
- &
-
median -
raange
-
peans
balance
prin
mean
variance symmetry
-
-
en -snewedness
& -
outliers
he
Beans use this to create the five-number
-
summary
graphical representation
&
fire-number summary :
bouplot
variations on bonplot
·
modified boxplot : don't note outliers (1. 5 JGR above or
below G, and Q) to explain
·
side-by-side boxplot snewed distributions
the mean and standard deviation
for
symmetric distributions , use
to explain y b . + bik -
:
can do linear transformations on dat
you 9
coral corb)
bo affects only of center and b , center spread
>
-
measures a
density curve is a representation of a histogram
>
-
graph stays the same when an observation changes (not for
histograms)
rules for density curves
, area gives proportion of observations that fall in
·
always area of 3 that range
meaning
n-oris-y-anis
has no
·
always on or above the
a class of density curves which are symmetrical are the
good approximations of
,
>are
Normal with Normal distribution
-
curves outcomes
completely determined by and o
·
N
·
one property every normal curve has is that 6PY falls within
30,
954 20 9917 % so also
and known 60, 95, 991 7-rule
as the
,
you can standardize an observation and transform it into the
standard Normal distribution ->
graph with N
= O and 0-b
·
area stays the same
·
all normal distributions are transformations of the standard Normal
distribution
to see how"normal" is, make Normal quantile
a graph a
plot
·pot eachprint against it's norma sceermal
summarize the
you can use a
scatterplot to
graphically
S relationship between two variables
antitative
studied on
ha samen aris the independant variable and garis the dependant
casesvariable
strength
I I
shape direction
follows trine
positiee
linear a
quadrativ i
-
exponentia neutral
quantify the direction and strength of relationship
allneu with
the correlation creffecient+
corariance is the shared variance betweenm and y
·
correlation conffecient is basically the standardized covariance
you can
plot a live in a scatterplot that can
predicty
from u
,
called the
regression line Cregression is the general technique of
fitting lines &
non-resistant
which is the
&
for prediction /
every line in regression there is error
distance of every point to the live also called residual
·
ei =
y
-
g
·
the best fitting line is the one with the least squared prediction
that influences the course of operationalisations
a
study :
nominal : unordered
categories
>
-
&
> ordinal : ordered categories ,
unequal intervals
-
rational :
equal intervals and zero means zero
>
-
interval : equal intervals and no zero
quantitative data ,
a measure can be described with
>
-
discrete (full) and continuous (between) numbers
choice for a plot depends on the measurement scale
categorical bar graph -
pie -
or
-
chart
anantitative en
myogram or boxplot -
histogram :
·
every bar is a bin
·
height is number of observations
-
in the bin
!
difference with bar graphs
·
no
gaps
:
describing data in a
graph
nalk Center spread
I adad I
shape
area& #
- &
-
median -
raange
-
peans
balance
prin
mean
variance symmetry
-
-
en -snewedness
& -
outliers
he
Beans use this to create the five-number
-
summary
graphical representation
&
fire-number summary :
bouplot
variations on bonplot
·
modified boxplot : don't note outliers (1. 5 JGR above or
below G, and Q) to explain
·
side-by-side boxplot snewed distributions
the mean and standard deviation
for
symmetric distributions , use
to explain y b . + bik -
:
can do linear transformations on dat
you 9
coral corb)
bo affects only of center and b , center spread
>
-
measures a
density curve is a representation of a histogram
>
-
graph stays the same when an observation changes (not for
histograms)
rules for density curves
, area gives proportion of observations that fall in
·
always area of 3 that range
meaning
n-oris-y-anis
has no
·
always on or above the
a class of density curves which are symmetrical are the
good approximations of
,
>are
Normal with Normal distribution
-
curves outcomes
completely determined by and o
·
N
·
one property every normal curve has is that 6PY falls within
30,
954 20 9917 % so also
and known 60, 95, 991 7-rule
as the
,
you can standardize an observation and transform it into the
standard Normal distribution ->
graph with N
= O and 0-b
·
area stays the same
·
all normal distributions are transformations of the standard Normal
distribution
to see how"normal" is, make Normal quantile
a graph a
plot
·pot eachprint against it's norma sceermal
summarize the
you can use a
scatterplot to
graphically
S relationship between two variables
antitative
studied on
ha samen aris the independant variable and garis the dependant
casesvariable
strength
I I
shape direction
follows trine
positiee
linear a
quadrativ i
-
exponentia neutral
quantify the direction and strength of relationship
allneu with
the correlation creffecient+
corariance is the shared variance betweenm and y
·
correlation conffecient is basically the standardized covariance
you can
plot a live in a scatterplot that can
predicty
from u
,
called the
regression line Cregression is the general technique of
fitting lines &
non-resistant
which is the
&
for prediction /
every line in regression there is error
distance of every point to the live also called residual
·
ei =
y
-
g
·
the best fitting line is the one with the least squared prediction
