Research Workshop Series

Multivariate and Multilevel Data Analysis

using SPSS, AMOS, SmartPLS and Mplus

Module 1:
Data Analysis using SPSS
TRAINER: WAQAR AKBAR
ORGANIZER: FACULTY OF MANAGEMENT
SCIENCE
SPSS – What Is It?
SPSS means “Statistical Package for the Social Sciences” and
was first launched in 1968.
Since SPSS was acquired by IBM in 2009, it's officially known
as IBM SPSS Statistics but most users still just refer to it as
“SPSS”.
SPSS - Overview Main
Features
SPSS is software for editing and analyzing all sorts of data.
These data may come from basically any source: scientific
research, a customer database, Google Analytics or even
the server log files of a website. SPSS can open all file
formats that are commonly used for structured data such as
spreadsheets from MS Excel or OpenOffice;
plain text files (.txt or .csv);
relational (SQL) databases;
Stata and SAS.
SPSS Data View
◦ This sheet -called data view- always displays our data
values
SPSS Variable View
◦ An SPSS data file always has a second sheet called
variable view. It shows the metadata associated with the
data. Metadata is information about the meaning of
variables and data values. This is generally known as the
“codebook” but in SPSS it's called the dictionary.
SPSS Output Window
◦ After clicking Ok, a new window opens up: SPSS’ Output
Viewer window. It holds a nice table with all statistics on
all variables we chose
Preparing a codebook
Preparing a codebook involves deciding (and documenting)
how you will go about:
◦ Defining and labelling each of the variable
◦ Assigning number to each of the possible

Instruction: Prepare a codebook for survey questionnaire in

EXCEL
Creating a data file and
entering data
Defining Variables
◦ Name, Type, Width, Decimals, Label, Values, Missing,
Columns, Align, Measure
Entering Data in SPSS
◦ Using Data Editor
◦ Using EXCEL
Screening and Cleaning the
Data
Screening process involves
◦ Checking for errors
◦ Need to check each of your variables for scores that are
out of range
◦ Findings and correcting
◦ Need to find where in the data file this error occurred

Instruction: Open (error.sav) file for this task

Screening and Cleaning the
Data
Step 1:Checking For Errors
◦ Checking Categorical Variable (follow procedure handout)
◦ Checking Continuous Variable (follow procedure handout)

Step 2: Finding and Correcting the errors in Data

File (follow procedure handout)
5. Descriptive Statistics
Instruction: Open (Survey.sav) file for this task
Categorical Variable
◦ Frequency
◦ Crosstabs

Continuous Variable
◦ Descriptive Statistics

Missing Data
◦ Exclude Case Listwise
◦ Exclude cases Pairwise
◦ Replace with mean
Descriptive Statistics: Missing
Values
The Exclude cases listwise option will include cases in the analysis only if
they have full data on all of the variables listed in your Variables box for that
case. A case will be totally excluded from all the analyses if it is missing even
one piece of information. This can severely, and unnecessarily, limit your
sample size.
The Exclude cases pairwise option, however, excludes the case (person) only
if they are missing the data required for the specific analysis. They will still be
included in any of the analyses for which they have the necessary
information.
The Replace with mean option, which is available in some SPSS statistical
procedures (e.g. multiple regression), calculates the mean value for the
variable and gives every missing case this value. This option should never be
used, as it can severely distort the results of your analysis, particularly if you
have a lot of missing values.
Descriptive Statistics
Assessing Normality
◦ Descriptive
◦ 5% trimmed Mean
◦ Extreme Values
◦ Skewness and Kurtosis
◦ Test of Normality
◦ K-S test & Shapiro-Wilk
◦ This assesses the normality of the distribution of scores,
◦ A non-significant value (more than 0.05) indicates normality
◦ Histogram
◦ Normal Q-Q plot (Probability plot)
◦ Detrended Normal Q-Q plots
◦ Obtained by plotting the actual deviation of the scores from the straight
line.
◦ There should be no clustering of points , with most collecting around the
zero line
◦ Outliers
◦ Histogram
◦ Boxplot
◦ Rectangle represents 50% of the cases, with the whiskers (the
lines protruding from the box) going out to the smallest and
largest values.
◦ The line inside the rectangle is the median value.
◦ Outliers appears as little circle with a number attached
◦ Extend more than 1.5 box-length from the edge of the box.
◦ Extreme points indicated with asterisk are those that extend
more than three box length from the edge of the box.
Descriptive Statistics
Exercise : data file Staffsurvey.sav

1. Follow the procedures covered in this descriptive statistics section to

answer the following questions:
a) What percentage of the staff in this organisation are permanent employees?
(Use the variable employstatus.)
b) What is the average length of service for staff in the organisation? (Use the
variable service.)
c) What percentage of respondents would recommend the organisation to
others as a good place to work? (Use the variable recommend.)
2. Assess the distribution of scores on the Total Staff Satisfaction scale
(totsatis) for employees who are permanent versus casual
(employstatus).
a) Are there any outliers on this scale that you would be concerned about
b) Are scores normally distributed for each group?
6. Manipulating the Data
• Calculating Total Scale Scores
◦ Step 1: Reverse any negatively worded items.
◦ Optimum Scale: Item 2, 4, 6 (Procedure handout)
◦ Step 2: Add together scores from all the items
that make up the subscale or scale
◦ (Procedure handout)

◦ Use (survey.sav) File for this exercise

• Collapsing a Continuous Variable into Groups
• For some analyses (e.g. Analysis of Variance), you may
wish to divide the sample into equal groups according
to respondents' scores on some variable (e.g. to give
low, medium and high scoring groups) or you have
very skewed distributions. Use visual binning
• Visual Binning will create a new categorical variable
leaving the original variable as continuous variable
• Age variable (Procedure handout)
• Use Survey.sav file for the this activity
• Collapsing the number of categories of a categorical
Variable
• There are some situations where you may want to reduce or
collapse the number of categories of a categorical variable.
• You may want to do this for research or theoretical reasons
(e.g. collapsing the marital status into just two categories
representing people 'in a relationship‘/'not in a relationship)
• Education variable (procedure handout)
• File : Survey.sav
 7. Reliability of a Scale
• When you are selecting scales to include in your study, it is
important to find scales that are reliable
• One of the main issues concerns the scale's internal consistency.
• This refers to the degree to which the items that make up the scale
'hang together'. Are they all measuring the same underlying construct?
• One of the most commonly used indicators of internal
consistency is Cronbach's alpha coefficient.
• Ideally, the Cronbach alpha coefficient of a scale should be above.7
(DeVellis 2003).
Example: Survey.sav ; Variable: lifsat1,2,3,4,5 (Procedure handout)
Warning: must check all negatively worded items
Interpretation
Inter-Item Correlation Matrix
◦ Check for negative values. All values should be positive, indicating that the items are
measuring the same underlying characteristic.

Cronbach’s Alpha
◦ Values above .7 are considered acceptable; however, values above .8 are preferable.

Corrected Item-Total Correlation

◦ values shown in the Item-Total Statistics table give you an indication of the degree to
which each item correlates with the total score.
◦ Low values (less than .3) here indicate that the item is measuring something different
from the scale as a whole.
◦ If your scale’s overall Cronbach alpha is too low (e.g. less than .7) and you have
checked for incorrectly scored items, you may need to consider removing items with
low item-total correlations.
Alpha if Item Deleted
◦ the impact of removing each item from the scale is given.
◦ Compare these values with the final alpha value obtained.
◦ If any of the values in this column are higher than the final alpha value, you
may want to consider removing this item from the scale.
Exercise
Data file: staffsurvey.sav.
1. Check the reliability of the Staff Satisfaction
Survey, which is made up of the agreement items in
the data file: Qla to QI0a. None of the items of this
scale need to be reversed.
8. Correlation
Correlation analysis is used to describe the strength and
direction of the linear relationship between two variables.
PRELIMINARY ANALYSES FOR CORRELATION
◦ Before performing a correlation analysis, it is a good idea to
generate a scatterplot.
◦ This enables you to check for violation of the assumptions of
linearity and homoscedasticity
Follow handout to generate scatterplot
File: Survey.sav
Interpretation of output
from scatterplot
Step 1: Checking for outliers
◦ Check your scatterplot for outliers
◦ If you identify an outlier and want to find out the ID number of the
case, you can use the Data Label Mode icon in the Chart Editor.
Double-click on the chart to activate the Chart Editor window. Click
on the icon that looks a bit like a bullseye (or choose Data Label
Mode from the Elements menu) and move your cursor to the point
on the graph you wish to identify. Click on it once and a number will
appear.
Step 2: Inspecting the distribution of data points
Are the data points spread all over the place? This suggests a very low
correlation.
• Are all the points neatly arranged in a narrow cigar shape? This
suggests quite a strong correlation.
• Could you draw a straight line through the main cluster of points, or
would a curved line better represent the points? If a curved line is
evident (suggesting a curvilinear relationship) Pearson correlation
should not be used, as it assumes a linear relationship.
• What is the shape of the cluster? Is it even from one end to the other?
Or does it start off narrow and then get fatter? If this is the case, your
data may be violating the assumption of homoscedasticity.
Step 3: Determining the direction of the relationship between the
variables
◦ The scatterplot can tell you whether the relationship between your
two variables is positive or negative.
Pearson Correlation
RQ: Is there a relationship between the amount of control people have over
their internal states and their levels of perceived stress? Do people with high
levels of perceived control experience lower levels of perceived stress?
◦ Follow Procedure handout; Example file survey.sav

Interpretation
Step 1: Checking the information about the sample
Step 2: Determining the direction of the relationship
Step 3: Determining the strength of the relationship
small r=.10 to .29; medium r=.30 to .49; large r=.50 to 1.0
Step 4: Calculating the coefficient of determination
Step 5: Assessing the significance level
PRESENTING THE RESULTS
FROM CORRELATION
The relationship between perceived control of internal states (as
measured by the PCOISS) and perceived stress (as measured by the
Perceived Stress Scale) was investigated using Pearson product-moment
correlation coefficient. Preliminary analyses were performed to ensure
no violation of the assumptions of normality, linearity and
homoscedasticity. There was a strong, negative correlation between the
two variables, r = –.58, n = 426, p < .001, with high levels of perceived
control associated with lower levels of perceived stress.
Correlation is often used to explore the relationship among a group of
variables, rather than just two as described above. In this case, it would be
awkward to report all the individual correlation coefficients in a paragraph; it
would be better to present them in a table. One way this could be done is as
follows:
COMPARING THE CORRELATION
COEFFICIENTS FOR TWO GROUPS
Sometimes when doing correlational research you may want to
compare the strength of the correlation coefficients for two separate
groups.
◦ For example, you may want to look at the relationship between optimism
and negative affect for males and females separately.
◦ Follow procedure handout

Important:
Remember, when you have finished looking at males and females
separately you will need to turn the Split File option off. It stays in place
until you specifically turn it off.
◦ To do this, click on Data, Split File and select the first button: Analyze all
cases, do not create groups.
Exercise
Data file: sleep.sav.
1. Check the strength of the correlation between
scores on the Sleepiness and Associated Sensations
Scale (totSAS) and the Epworth Sleepiness Scale
(ess).
 9. Partial Correlation
• Partial correlation is similar to Pearson product-moment
correlation, except that it allows you to control for an additional
variable.
• This is usually a variable that you suspect might be influencing
your two variables of interest. By statistically removing the
influence of this confounding variable, yon can get a clearer and
more accurate indication of the relationship between your two
variables.
•This occurs when the relationship between two variables (A and
B) is influenced, at least to some extent, by a third variable (C).
This can serve to artificially inflate the size of the correlation
coefficient obtained.
 Example of research question:
After controlling for subjects' tendency to present
themselves in a positive light on self-report scales, is there
still a significant relationship between perceived control of
internal states (PCOISS) and levels of perceived stress?
File: Survey.sav
Follow Procedure Handout
Interpretation
In the top half of the table is the normal Pearson product-moment
correlation matrix between your two variables of interest not
controlling for your other variable. In this case, the correlation is –.581.
The bottom half of the table repeats the same set of correlation
analyses, but this time controlling for (removing) the effects of your
control variable (e.g. social desirability). In this case, the new partial
correlation is –.552.
PRESENTING THE RESULTS
FROM PARTIAL CORRELATION
Partial correlation was used to explore the relationship between
perceived control of internal states (as measured by the PCOISS) and
perceived stress (measured by the Perceived Stress Scale), while
controlling for scores on the Marlowe-Crowne Social Desirability Scale.
Preliminary analyses were performed to ensure no violation of the
assumptions of normality, linearity and homoscedasticity. There was a
strong, negative, partial correlation between perceived control of
internal states and perceived stress, controlling for social desirability, r =
–.55, n = 425, p < .001, with high levels of perceived control being
associated with lower levels of perceived stress. An inspection of the
zero-order correlation (r = –.58) suggested that controlling for socially
desirable responding had very little effect on the strength of the
relationship between these two variables.
Exercise
sleep.sav.
1. Check the strength of the correlation between
scores on the Sleepiness and Associated Sensations
Scale (totSAS) and the impact of sleep problems on
overall wellbeing (impact6) while controlling for
age. Compare the zero order correlation (Pearson
Correlation) and the partial correlation coefficient.
Does controlling for age make a difference?
10. Multiple Regression
• Multiple regression tells you how much of the
variance in your dependent variable can be
explained by your independent variables.
• It also gives you an indication of the relative
contribution of each independent variable. Tests
allow you to determine the statistical significance
of the results, in terms of both the model itself and
the individual independent variables.
Some of the main types of research questions that multiple regression
can be used to address are:
• how well a set of variables is able to predict a particular outcome
• which variable in a set of variables is the best predictor of an outcome
• whether a particular predictor variable is still able to predict an
outcome when the effects of another variable are controlled for (e.g.
socially desirable responding).
MAJOR TYPES OF
MULTIPLE REGRESSION
1. standard or simultaneous
2. hierarchical or sequential
3. stepwise.
Standard Multiple Regression
In standard multiple regression, all the independent (or
predictor) variables are entered into the equation
simultaneously.
Each independent variable is evaluated in terms of its
predictive power, over and above that offered by all the
other independent variables.
This is the most commonly used multiple regression
analysis. You would use this approach if you had a set of
variables (e.g. various personality scales) and wanted to
know how much variance in a dependent variable (e.g.
anxiety) they were able to explain as a group or block.
This approach would also tell you how much unique
variance in the dependent variable each of the independent
variables explained.
Hierarchical multiple
regression
In hierarchical regression (also called sequential regression), the
independent variables are entered into the model in the order specified by
the researcher based on theoretical grounds.
Variables or sets of variables are entered in steps (or blocks), with each
independent variable being assessed in terms of what it adds to the
prediction of the dependent variable after the previous variables have been
controlled for.
For example, if you wanted to know how well optimism predicts life
satisfaction, after the effect of age is controlled for, you would enter age in
Block 1 and then Optimism in Block 2.
Once all sets of variables are entered, the overall model is assessed in terms
of its ability to predict the dependent measure. The relative contribution of
each block of variables is also assessed.
Stepwise multiple regression
In stepwise regression, the researcher provides a list of independent
variables and then allows the program to select which variables it will
enter and in which order they go into the equation, based on a set of
statistical criteria.
There are three different versions of this approach: forward selection,
backward deletion and stepwise regression.
There are a number of problems with these approaches and some
controversy in the literature concerning their use (and abuse). Before
using these approaches(see Tabachnick & Fidell 2013, p. 138).
It is important that you understand what is involved, how to choose the
appropriate variables and how to interpret the output that you receive.
ASSUMPTIONS OF
MULTIPLE REGRESSION
Sample size
◦ Stevens (1996, p. 72) recommends that ‘for social science research, about 15
participants per predictor are needed for a reliable equation.
◦ Tabachnick and Fidell (2013, p. 123) give a formula for calculating sample size
requirements, taking into account the number of independent variables that
you wish to use: N > 50 + 8m (where m = number of independent variables).
If you have five indepndent variables, you will need 90 cases.

Multicollinearity and singularity

◦ Multicollinearity exists when the independent variables are highly correlated
(r=.9 and above).
◦ Singularity occurs when one independent variable is actually a combination
of other independent variables.
Outliers
◦ Outliers on your dependent variable can be identified from the standardised
residual plot that can be requested (described in the example presented
later in this chapter).
◦ Tabachnick and Fidell (2013, p. 128) define outliers as those with
standardised residual values above about 3.3 (or less than –3.3).

Normality, linearity, homoscedasticity, independence of residuals

◦ normality: the residuals should be normally distributed about the predicted
DV scores
◦ linearity: the residuals should have a straight-line relationship with predicted
DV scores
◦ homoscedasticity: the variance of the residuals about predicted DV scores
should be the same for all predicted scores.
 Example
File Name: Survey.sav
Variables
 Total Perceived Stress (tpstress): total score on the Perceived Stress scale. High scores
indicate high levels of stress.
 Total Perceived Control of Internal States (tpcoiss): total score on the Perceived Control
of Internal States scale. High scores indicate greater control over internal states.
 Total Mastery (tmast): total score on the Mastery scale. High scores indicate higher
levels of perceived control over events and circumstances.
Total Social Desirability (tmarlow): total scores on the Marlowe-Crowne Social
Desirability scale, which measures the degree to which people try to present
themselves in a positive light.
 Age: age in years.
Example of research
questions:
1. How well do the two measures of control (mastery, PCOISS)
predict perceived stress? How much variance in perceived
stress scores can be explained by scores on these two scales?
2. Which is the best predictor of perceived stress: control of
external events (Mastery scale) or control of internal states
(PCOISS)?

3. If we control for the possible effect of age and socially

desirable responding, is this set of variables still able to predict
a significant amount of the variance in perceived stress?
Standard Multiple Regression
Two Questions will be addressed:
1. How well do the two measures of control (mastery, PCOISS) predict
perceived stress? How much variance in perceived stress scores can be
explained by scores on these two scales?
2. Which is the best predictor of perceived stress: control of external
events (Mastery scale) or control of internal states (PCOISS)?
Diagnosis
Step 1: Checking the Assumptions
◦ Multicollinearity
◦ Correlation Matrix
◦ Collinearity Diagnosis: Tolerance and VIF
◦ commonly used cut-off points for determining the presence of
multicollinearity (tolerance value of less than .10, or a VIP value of
above 10).
◦ Outliers, normality, homoscedasticity, independence
of residuals
◦ See: Normal Plot (P-P) and Scatter plot
◦ See: Mahlanobis Distance
Critical values for evaluating
Mahalanobis distance values
◦ To find which case has this value,
◦ go back to your Data Editor window, select Data and then Sort Cases from the menu.
◦ Sort by the new variable at the bottom of your data file (Mahalanobis Distance, MAH-
1) in Descending order.
◦ Casewie Diagnostics
◦ This presents information about cases that have standardised residual values above
3.0 or below –3.0. In a normally distributed sample, we would expect only 1 per cent
of cases to fall outside this range.
◦ the Casewise Diagnostics table that this person recorded a Total perceived stress
score of 14, but our model predicted a value of 28.85
◦ To check whether this strange case is having any undue influence on the results for
ourmodel as a whole, we can check the value for Cook’s Distance.
◦ According to Tabachnick and Fidell (2013, p. 75), cases withvalues larger than 1 are a
potential problem.
Step 2: Evaluating the model
R Square
◦ how much of the variance in the dependent variable (perceived stress) is
explained by the model

Adjusted R Square
◦ When a small sample is involved, the R square value in the sample tends to
be a rather optimistic overestimation of the true value in the population (see
Tabachnick & Fidell 2013).
◦ The Adjusted R square statistic ‘corrects’ this value to provide a better
estimate of the true population value.

ANOVA
◦ This tests the null hypothesis that multiple R in the population equals 0.
Step 3: Evaluating each of the
independent variables
Coefficients
◦ To compare the different variables it is important that you look at the
standardized coefficients, not the unstandardised ones. ‘
◦ Standardised’ means that these values for each of the different variables
have been converted to the same scale so that you can compare them.
◦ If you were interested in constructing a regression equation, you would use
the unstandardised coefficient values listed as B.

Sig.
◦ tells you whether this variable is making a statistically significant unique
contribution to the equation.

.
Part correlation coefficients
◦ If you square this value, you get an indication of the contribution of that
variable to the total R square.
◦ In other words, it tells you how much of the total variance in the dependent
variable is uniquely explained by that variable and how much R square would
drop if it wasn’t included in your model.
more practical purposes
The beta values obtained in this analysis can also be used for other
more practical purposes than the theoretical model testing shown here.
Standardised beta values indicate the number of standard deviations
that scores in the dependent variable would change if there was a one
standard deviation unit change in the predictor.
In the current example, if we could increase Mastery scores by one
standard deviation (which is 3.97, from the Descriptive Statistics table)
the perceived stress scores would be likely to drop by .42 standard
deviation units.
Hierarchical Multiple
Regression
In hierarchical regression (also called sequential regression), the
independent variables are entered into the equation in the order
specified by the researcher based on theoretical grounds.
Variables or sets of variables are entered in steps (or blocks), with each
independent variable being assessed in terms of what it adds to the
prediction of the dependent variable, after the previous variables have
been controlled for.
For example, if you wanted to know how well optimism predicts life
satisfaction, after the effect of age is controlled for, you would enter age
in Block 1 and then Optimism in Block 2.
Once all sets of variables are entered, the overall model is assessed in
terms of its ability to predict the dependent measure. The relative
contribution of each block of variables is also assessed.
Example Question
If we control for the possible effect of age and
socially desirable responding, is this set of variables
still able to predict a significant amount of the
variance in perceived stress?
Diagnosis
Step 1: Evaluating the Model
R Square
◦ After the variables in Block 1 (social desirability) have been entered, the
overall model explains 5.7 per cent of the variance (.057 × 100). After Block 2
variables (Total Mastery, Total PCOISS) have also been included, the model as
a whole explains 47.4 per cent (.474 × 100).

R Square change
square change value is .42. This means that Mastery and PCOISS explain
an additional 42 per cent (.42 × 100) of the variance in perceived stress,
even when the effects of age and socially desirable responding are
statistically controlled for.
ANOVA
indicates that the model as a whole is significant.
Step 2: Evaluating each of the
independent variables
Coefficient and Sig
◦ only two variables that make a unique statistically significant contribution
(less than .05).
◦ Remember, these beta values represent the unique contribution of each
variable, when the overlapping effects of all other variables are statistically
removed.
PRESENTING THE RESULTS FROM MULTIPLE REGRESSION

How to write the results?

PRESENTING THE RESULTS
FROM MULTIPLE REGRESSION
Hierarchical multiple regression was used to assess the ability of two control
measures (Mastery Scale, Perceived Control of Internal States Scale: PCOISS)
to predict levels of stress (Perceived Stress Scale), after controlling for the
influence of social desirability and age. Preliminary analyses were conducted
to ensure no violation of the assumptions of normality, linearity,
multicollinearity and homoscedasticity. Age and social desirability were
entered at Step 1, explaining 6% of the variance in perceived stress. After
entry of the Mastery Scale and PCOISS Scale at Step 2 the total variance
explained by the model as a whole was 47.4%, F (4, 421) = 94.78, p < .001.
The two control measures explained an additional 42% of the variance in
stress, after controlling for age and socially desirable responding, R squared
change = .42, F change (2, 421) = 166.87, p < .001. In the final model, only
the two control measures were statistically significant, with the Mastery
Scale recording a higher beta value (beta = –.44, p < .001) than the PCOISS
Scale (beta = –.33, p < .001).
ADDITIONAL EXERCISES
Health: Data file: sleep.sav.
1. Conduct a standard multiple regression to explore factors that impact on
people’s level of daytime sleepiness. For your dependent variable, use the
Sleepiness and Associated Sensations Scale total score (totSAS). For
independent variables, use sex, age, physical fitness rating (fitrate) and scores
on the HADS Depression Scale (depress). Assess how much of the variance in
total sleepiness scores is explained by the set of variables (check your R square
value). Which of the variables make a unique significant contribution (check
your beta values)?
2. Repeat the above analysis, but this time use a hierarchical multiple
regression procedure entering sex and age in the first block of variables and
physical fitness and depression scores in the second block. After controlling for
the demographic variables of sex and age, do the other two predictor variables
make a significant contribution to explaining variance in sleepiness scores?
How much additional variance in sleepiness is explained by physical fitness and
depression, after controlling for sex and age?
11. Factor analysis
It takes a large set of variables and looks for a way the data may be
‘reduced’ or summarised using a smaller set of factors or components.
It does this by looking for ‘clumps’ or groups among the inter
correlations of a set of variables. This is an almost impossible task to do
‘by eye’ with anything more than a small number of variables.

STEPS INVOLVED IN FACTOR ANALYSIS

Step 1: Assessment of the suitability of the
data for factor analysis
Step 1: Assessment of the suitability of the data for factor analysis
◦ There are two main issues to consider in determining whether a particular data
set is suitable
◦ for factor analysis: sample size, and the strength of the relationship among the
variables

Two statistical measures are also generated by IBM SPSS to help assess the
factorability of the data: Bartlett’s test of sphericity (Bartlett 1954), and the
Kaiser-
Meyer-Olkin (KMO) measure of sampling adequacy (Kaiser 1970, 1974).
Bartlett’s test of sphericity should be significant (p < .05) for the factor
analysis to be considered appropriate.
The KMO index ranges from 0 to 1, with .6 suggested as the minimum value
for a good factor analysis (Tabachnick & Fidell 2013).
Step 2: Factor extraction
Factor extraction involves determining the smallest number of factors
that can be used to best represent the interrelationships among the set
of variables.
There are a variety of approaches that can be used to identify (extract)
the number of underlying factors or dimensions.
The most commonly used approach is principal components analysis
There are a number of techniques that can be used to assist in the
decision concerning the number of factors to retain:
◦ Kaiser’s criterion;
◦ scree test; and
◦ parallel analysis.
Kaiser’s criterion
◦ only factors with an eigenvalue of 1.0 or more are retained for further
investigation

Scree test
◦ This involves plotting each of the eigenvalues of the factors (IBM SPSS does
this for you) and inspecting the plot to find a point at which the shape of the
curve changes direction and becomes horizontal

Parallel analysis
◦ Parallel analysis involves comparing the size of the eigenvalues with those
obtained from a randomly generated data set of the same size. Only those
eigenvalues that exceed the corresponding values from the random data set
are retained.
Step 3: Factor rotation and
interpretation
There are two main approaches to rotation, resulting in either
orthogonal (uncorrelated) or oblique (correlated) factor solutions.
Within the two broad categories of rotational approaches there are a
number of different techniques
◦ orthogonal: Varimax, Quartimax, Equamax;
◦ oblique: Direct Oblimin, Promax).
◦ The most commonly used orthogonal approach is the Varimax method,
which attempts to minimise the number of variables that have high loadings
on each factor.
◦ The most commonly used oblique technique is Direct Oblimin.
EXAMPLE (data file:
survey.sav)
Research question:
What is the underlying factor structure of the Positive and Negative
Affect Scale? Past research suggests a two-factor structure (positive
affect/negative affect). Is the structure of the scale in this study, using a
community sample, consistent with this previous research?
What you need: A set of correlated continuous variables.
What it does: Factor analysis attempts to identify a small set of factors
that represents the underlying relationships among a group of related
variables.
Interpretation of output—Part 1

Step 1
Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) value is .6 or
above and that the Bartlett’s Test of Sphericity value is significant
Correlation Matrix
◦ look forcorrelation coefficients of .3 and above

Step 2 :Total Variance Explained

◦ Initial Eigenvalues

Step 3: Screeplot
Step 4:parallel analysis MonteCarloPA
◦ the number of variables you are analysing (in this case, 20);
◦ the number of participants in your sample (in this case, 435); and
◦ the number of replications(specify 100).

systematically compare the first eigenvalue you obtained in IBM SPSS

with the corresponding first value from the random results generated
by parallel analysis. If your value is larger than the criterion value from
parallel analysis, you retain this factor; if it is less, you reject it.
Step 5
ComponentMatrix
◦ shows the unrotated loadings of each of the items on the four components
Step 6
Pattern Matrix (Rotated Solution)
Using the default options in IBM SPSS, we obtained a four-factor
solution. It is now necessary to go back and ‘force’ a two-factor
solution
Part 2: Oblimin rotation of
two-factor solution
Component Correlation Matrix
◦ This shows you the strength of the relationship between the two factor

Pattern Matrix
◦ shows the factor loadings of each of the variables.

Structure Matrix
◦ unique to the Oblimin output, provides information about the correlation
between variables and factors.

Communalities
◦ This gives information about how much of the variance in each item is
explained. Low values (e.g. less than .3) could indicate that the item does not
fit well with the other items in its component
PRESENTING THE RESULTS FROM FACTOR ANALYSIS
The 20 items of the Positive and Negative Affect Scale (PANAS) were subjected to
principal components analysis (PCA) using SPSS version 18. Prior to performing
PCA, the suitability of data for factor analysis was assessed . Inspection of the
correlation matrix revealed the presence of many coefficients of .3 and above.
The Kaiser-Meyer- Olkin value was .87, exceeding the recommended value of .6
(Kaiser 1970, 1974) and Bartlett’s Test of Sphericity (Bartlett 1954) reached
statistical significance, supporting the factorability of the correlation matrix.
Principal components analysis revealed the presence of four components with
eigenvalues exceeding 1, explaining 31.2%, 17%, 6.1% and 5.8% of the variance
respectively. An inspection of the screeplot revealed a clear break after the
second component. Using Catell’s (1966) scree test, it was decided to retain two
components for further investigation. This was further supported by the results
of Parallel Analysis, which showed only two components with eigenvalues
exceeding the corresponding criterion values for a randomly generated data
matrix of the same size (20 variables × 435 respondents).
The two-component solution explained a total of 48.2% of the variance,
with Component 1 contributing 31.25% and Component 2 contributing
17.0%. To aid in the interpretation of these two components, oblimin
rotation was performed. The rotated solution revealed the presence of
simple structure (Thurstone 1947), with both components showing a
number of strong loadings and all variables loading substantially on only
one component.
The interpretation of the two components was consistent with previous
research on the PANAS Scale, with positive affect items loading strongly
on Component 1 and negative affect items loading strongly on Component
2. There was a weak negative correlation between the two factors (r =
–.28). The results of this analysis support the use of the positive affect
items and the negative affect items as separate scales, as suggested by the
scale authomrs (Watson, Clark & Tellegen 1988).
ADDITIONAL EXERCISES
Data file: staffsurvey.sav
1. conduct a principal components analysis with Oblimin rotation on the
ten agreement items that make up the Staff Satisfaction Survey (Q1a to
Q10a). You will see that, although two factors record eigenvalues over
1, the screeplot indicates that only one component should be retained.
Run Parallel Analysis using 523 as the number of cases and 10 as the
number of items. The results indicate only one component has an
eigenvalue that exceeds the equivalent value obtained from a random
data set. This suggests that the items of the Staff Satisfaction Scale are
assessing only one underlying dimension (factor).
12. T-tests
• independent-samples t-test, used when you want to compare the
mean scores of two different groups of people or conditions
• paired-samples t-test, used when you want to compare the mean
scores for the same group of people on two different occasions, or
when you have matched pairs.

In both cases, you are comparing the values on some continuous

variable for two groups or on two occasions. If you have more than two
groups or conditions, you will need to use analysis of variance instead
INDEPENDENT-SAMPLES T-
TEST (Data File: survey.sav)
Research question: Is there a significant difference in the mean self-
esteem scores for males and females?
What you need: Two variables:
• one categorical, independent variable (e.g. males/females)
• one continuous, dependent variable (e.g. self-esteem scores).
What it does: An independent-samples t-test will tell you whether there
is a statistically significant difference in the mean scores for the two
groups (i.e. whether males and females differ significantly in terms of
their self-esteem levels). In statistical terms, you are testing the
probability that the two sets of scores (for males and females) came
from the same population.
Interpretation
Step 1: Checking the information about the groups
◦ Group Statistics

Step 2: Checking assumptions

◦ Independent Samples Test: Levene’s test for equality of variances
◦ If your Sig. value for Levene’s test is larger than .05 (e.g. .07, .10) you should
use the first line in the table, which refers to Equal variances assumed.
◦ If the significance level of Levene’s test is p=.05 or less (e.g. .01, .001), this
means that the variances for the two groups (males/females) are not the
same. Therefore your data violate the assumption of equal variance, use the
information in the second line of the t-test table, which refers to Equal
variances not assumed.
Step 3: Assessing differences between the groups
◦ t-test for Equality of Means
◦ If the value in the Sig. (2-tailed) column is equal to or less than .05 (e.g. .03, .
01, .001), there is a significant difference in the mean scores on your
dependent variable for each of the two groups.
◦ If the value is above .05 (e.g. .06, .10), there is no significant difference
between the two groups

Step 4: Calculating the effect size for independent-samples t-test

Eta squared
The guidelines (proposed by Cohen 1988, pp. 284–7) for interpreting
this value are:
.01=small effect
.06=moderate effect
.14=large effect
Presenting the results for
independent-samples t-test
An independent-samples t-test was conducted to compare the self-
esteem scores for males and females. There was no significant
difference in scores for males (M = 34.02, SD = 4.91) and females (M =
33.17, SD = 5.71; t (434) = 1.62, p = .11, two-tailed). The magnitude of
the differences in the means (mean difference = .85, 95% CI: –1.80 to
1.87) was very small (eta squared = .006).
PAIRED-SAMPLES T-TEST
A paired-samples t-test (also referred to as repeated measures) is used
when you have only one group of people (or companies, or machines
etc.) and you collect data from them on two different occasions or
under two different conditions.
Pre-test/post-test experimental designs are an example of the type of
situation where this technique is appropriate.
Example : experim.sav
Research question: Is there a significant change in participants’ Fear of
Statistics Test scores following participation in an intervention designed to
increase students’ confidence in their ability to successfully complete a
statistics course? Does the intervention have an impact on participants’
Fear of Statistics Test scores?
What you need: One set of participants (or matched pairs). Each person
(or pair) must provide both sets of scores. Two variables:
◦ one categorical independent variable (in this case it is Time, with two different
levels Time 1, Time 2)
◦ one continuous, dependent variable (e.g. Fear of Statistics Test scores)
measured on two different occasions or under different conditions.

What it does: A paired-samples t-test will tell you whether there is a

statistically significant difference in the mean scores for Time 1 and Time 2.
Interpretation of output from
paired-samples t-tes
Step 1: Determining overall significance
◦ Paired Samples Test: Sig.(2-tailed)

Step 2: Comparing mean values Paired Samples Statistics: Mean scores

for each of the two sets of scores.
Calculating the effect size for paired-samples t-test
Presenting the results for
paired-samples t-test
A paired-samples t-test was conducted to evaluate the impact of the
intervention on students’ scores on the Fear of Statistics Test (FOST).
There was a statistically significant decrease in FOST scores from Time 1
(M = 40.17, SD = 5.16) to Time 2 (M = 37.5, SD = 5.15), t (29) = 5.39, p
< .001 (two-tailed). The mean decrease in FOST scores was 2.67 with a
95% confidence interval ranging from 1.66 to 3.68. The eta squared
statistic (.50) indicated a large effect size.
ADDITIONAL EXERCISES
Business Data file: staffsurvey.sav. See Appendix for details of the data
file.
1. Follow the procedures in the section on independent-samples t-tests
to compare the mean staff satisfaction scores (totsatis) for permanent
and casual staff (employstatus). Is there a significant difference in mean
satisfaction scores?