Research 2 Factor analysis

?
  • Created by: CaliFish
  • Created on: 20-04-17 12:40
Define Factor Analysis
Factor analysis is a data reduction tool, that finds common underlying dimensions in the data, by reducing a large number of variables to a smaller set, that are representative and meaningful, whilst keeping as much info as possible
1 of 70
give 3 over names for these variables
items, constructs, questions on the questionnaire
2 of 70
what is a factor
the meaningful cluster of these variables
3 of 70
what is another name for a factor
a latent variable
4 of 70
are these factors independent of each other
yes, but the extent of this depends on the analysis you run
5 of 70
how does it reduce the data
by removing redundancy and duplication
6 of 70
define redundancy
a variable unrelated to any of the concepts
7 of 70
define duplication
a variable measuring the same ting as another variable
8 of 70
what is the minimum number of participants for a factor analysis
100
9 of 70
what number of participants is wanted
300
10 of 70
if under 300, what is the ratio for the number of participants and the number of variables
10:1
11 of 70
if under 300, what is the ratio for the number of variables and the number of factors
4:1
12 of 70
if under 300, what is the ratio for the number of participants and the number of factors
6:1
13 of 70
in small samples correlation coefficients do what
fluctuate more than normal
14 of 70
how many steps are involved in exploratory factor analysis
5
15 of 70
name them
1. preliminary analysis: data screening 2. suitability checks 3. factor extraction 4. factor rotation 5. reliability analysis
16 of 70
define data screening
The process of inspecting raw data for errors and correcting them prior to doing data analysis.
17 of 70
what are the three substages within preliminary analysis: data screening
SD checks, Normality checks, Correlation checks
18 of 70
the SD's should not be above what
1.5
19 of 70
why
because otherwise all the data would be loaded at the two ends
20 of 70
the SD's should not be below what
0.5
21 of 70
why
because otherwise all the data is clustered around the centre
22 of 70
what level should data be measured at
interval level
23 of 70
how can you check for normality
K-S, Skew and kurtosis scores and histograms
24 of 70
draw positive skew
***
25 of 70
draw negative skew
***
26 of 70
draw positive kurtosis (what is this called)
***
27 of 70
draw negative kurtosis (what is this called)
***
28 of 70
is the normality assumption important
yes, if you want to generalise the results of your analysis beyond the sample you're using
29 of 70
where do you look to check your correlations
The R matrix
30 of 70
should your variables be correlated
yes, to an extent
31 of 70
why should your variables be correlated
if items are measuring the same construct, they will be correlated
32 of 70
item's below what value, should be removed
0.3
33 of 70
define multicollinearity
variables are highly correlated
34 of 70
item's above what value, should be removed
0.8
35 of 70
define singularity
variables are perfectly correlated
36 of 70
what correlation coefficient value is this represented by
1.0
37 of 70
how else, other than the correlation coefficient can you test for multicollinearity and singularity
by checking the DETERMINANT value
38 of 70
what value is acceptable
anything above 0.00001
39 of 70
the determinant value ranges between what
0-1
40 of 70
how do you check if the final data set (it is finalised after dealing with SD's, normality and correlations) is suitable for factor analysis
measure sampling adequacy and test for the assumption of sphericity
41 of 70
what SPSS statistic measures sampling adequacy
KMO
42 of 70
explain exactly what it is measuring
measuring the proportion of variance among variables that might be common variance / the relationships between variables
43 of 70
what is the KMO value range
0 - 1
44 of 70
the higher the value the better but Kaiser recommends accepting values greater than what?
0.5
45 of 70
how do you report KMO?
a sentence verbally explaining how strong/weak the realtionship between the variables are (KMO = x), indicating it is acceptable/unacceptable to proceed with analysis
46 of 70
what statistic is used to test for sphericity
Bartletts test
47 of 70
what is it testing for
Tests the null hypothesis that the original correlation matrix is an identity matrix
48 of 70
what is an identity matrix
a matrix where all coefficients are 0
49 of 70
we want our variables to have a what
relationship
50 of 70
this is shown through what
a significant value
51 of 70
how do you report for bartletts test of sphericity
Bartlett's test of sphericity was/was not significant, X squared (df) = value , p < .value, indicating it is acceptable/unacceptable to proceed with analysis
52 of 70
what are the two factor extraction methods
extraction based on Eiegenvalues greater than 1 using kaisers criteria and the sree plot
53 of 70
Kaisers criterian is only accurate if what
There are less than 30 variables and communalities after extraction are greater than 0.7 or When the sample size exceeds 250 and the average communality is greater than 0.6
54 of 70
IF still no, what can you do
check the scree plot
55 of 70
what do eigenavlues represent
the amount of variance explained by a factor
56 of 70
what does a communality represent
how much percent of variance in a variable is explained by all factor
57 of 70
do you want eigenvalues and communalities to be high or low
the higher the better, this means the more variance being explained
58 of 70
what does the scree plot, plot
plots the eigenvalues of the factors extracted
59 of 70
you can measure how many factors you have by doing what
how many points are to the left of the point of inflection
60 of 70
why do we use factor rotation
to aid interpretation
61 of 70
what does factor loading tell us
the relative contribution of a variable to a factor
62 of 70
what does it do to help us
it maximises the loading of a variable on one factor while minimsing its loadings on all other factors, this distributes the variance explained (Relative importance) more evenly between factors
63 of 70
explain this in short
it optimises the data
64 of 70
what are the two types of factor rotation
orthogonal and oblique
65 of 70
when should you use orthogonal
when factors are uncorrelated
66 of 70
give an example of one
varimax
67 of 70
when should you use oblique
when factors are inter-correlated
68 of 70
give an example one
direct oblimin
69 of 70
what do you have to do for orthogonal that you DONT have to do for oblique
keep 90 degree
70 of 70

Other cards in this set

Card 2

Front

give 3 over names for these variables

Back

items, constructs, questions on the questionnaire

Card 3

Front

what is a factor

Back

Preview of the front of card 3

Card 4

Front

what is another name for a factor

Back

Preview of the front of card 4

Card 5

Front

are these factors independent of each other

Back

Preview of the front of card 5
View more cards

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Research methods resources »