Professional Documents
Culture Documents
The purpose of this paper is to assess the impact of performance appraisal on employee
performance and focusing on the role of Motivation. Employee’s performance is the major
issue in an organization. Our study focuses on to find out the impact of performance appraisal
on employee’s performance and also analyses that motivation affects the relationship of
performance appraisal and employee’s performance. The study findings show the presence of
significant positive and negative outcomes when the organisation uses performance appraisal
as a motivation tool. Further, the study finds that the use of more than one appraisal
techniques helps yield greater satisfaction and consequently higher motivational levels.
We used Two hypotheses are analyzed by using sampling techniques; we were select 36
numbers of employees as sample by using simple random sampling among the different
organizations of Islamabad. Primary data were collect through standard questionnaire. For
analyzing data, we applied different tools and techniques through IBM SPSS 20 and Amos
Software. Results presented there is positive relationship between performance appraisal and
employee’s performance. Motivation as a moderator positively affected the relationship
between performance appraisal and employee’s performance. Different Organizations can use
appraisal system as a strategic approach by integrate it with business policies and HR
practices and can improve the performance standers of its employees.
1
Acknowledgement
All praises are for Allah, the Most Beneficent, ever Merciful and respect for Holy Prophet
Muhammad (Peace be upon Him). We thank Almighty Allah for giving us the strength and
courage to successfully complete this task. It is our privilege to thank our parents who always
encouraged us to do things beyond the limits. We owe a deep sense of gratitude to our
beloved teacher Dr Qlander Hayat. His help, guidance, keen interest and above all his
overwhelming attitude to help his students had been solely responsible for the completion of
our work. His guidance has always been a source of motivation throughout this project.
Dedication
This study is dedicated to all of my teachers, friends, and family, who have encouraged me
throughout this journey to achieve the highest level of education available. I would also like
to dedicate this dissertation to my parents, who has been a blessing, and whom I love dearly!
Table of Contents
Title Page…………………………………………………………………………
Preface
Acknowledgement
Dedication
Table of Contents
Introduction
Theoretical Framework (Model)
Questionnaire (With Variable Names)
Coding of Variables in SPSS
Preliminary Examination of Data
Exploring data with graphs
Univariate Profiling: Examining the Shape of the Distribution
Bivariate Profiling: Examining the Relationship Between Variables
Bivariate Profiling: Examining Group Differences
Multivariate Profiles
New Measures of Association
Missing Data
A Four-Step Process for Identifying Missing Data and Applying Remedies
Outliers
Detecting and Handling Outliers
Assumptions of Multivariate Analysis (Four Assumptions)
Assessing Individual Variables Versus and Variate
Data transformations
Exploratory Factor Analysis (EFA) Six Stage Process
4
Introduction
To evaluate the impact of performance appraisal on employee’s performance, we need to
firmly know what is performance appraisal and employee performance. Performance
appraisal is one of the most significant aspects of human resource activities designed and
performed in the companies as well as in every walk of life. In terms of definition, it is an act
of formally evaluating performance of non managerial staff at least once in a year (John
T.Addison, Clive R.Belfield, 2007). Performance appraisal can be defined as an act of
testing, evaluating, measuring and justifying the performance displayed by the employees
during a specific period of time (Meysam Fakharyan, Mohammad Reza Jalilvand, Behrooz
Dini, Ebrahim Dehafarin, 2012). Maximizing the performance of organizations is the main
issue for an organization(Bob cardy, 1997).Good organization performance refers to the
employee’s performance. Satisfactory performance of employees does not happen
automatically. Managerial standards, Knowledge and Skill, Commitment and Performance
appraisals effecting employee’s performance. But we are focusing on performance appraisal.
The history of performance appraisal is quite brief. Its roots can be traced in the early 20th
century to Taylor’s pioneering time and motion studies. The performance appraisal system
start in practiced mainly in the 1940s and with the help of this system, merit rating was used
for the first time near the Second World War as a method of justifying an employee’s wages
(Lillian & Sitati,2011). There are number of organizations and banks in Pakistan that using
the performance management system for making better their employee’s performance
because it leads to achieve organizational performance. However performance appraisal is
very important process but it deemed to be the “weak point” of managing human force
(Pulakios, 2009). Therefore performance appraisal is important to manage employee’s work
effectively. (Armstrong, 2001) tells performance as behavior the way in which organization’s
teams and individuals get work done.(Mooney, 2009) suggested that performance is not only
related to results but it also relates with activities and behaviors of employees that they
adopted to achieve their given goals. (Dessler, 2005) define performance appraisal as
“comparing the employee’s present and past performance to his/her performance standards”.
(Grubb,2007) says performance appraisals a procedure to evaluate how individual personnel
are performing and how they can improve their performance and contribute to overall
organizational performance. (Beach, 2001) Performance appraisal is the systematic
evacuation of employees according to the
job and potential development. (Pınar Güngör, 2011) says “Motivation is the ability of person
to modify his/her behavior”. Motivation is a driven force that leads and directed a person
toward some specific goals.
Performance management is key area of human resource practices and Performance appraisal
is a crucial part of performance management. Performance management is a system basically
used by top management like executives, managers and supervisors in order to align
employee performance with the firm’s goal. Performance management is used for different
purposes like; to define measures of performance, monitor the employee performance, and to
provide employees with feedbacks about their performance. Performance appraisal is used to
define the extent to which an employee performs his/her work effectively. According to
Ivancevich, (2007), different terms are used for performance appraisal as and when required;
like performance evaluation, performance review, personnel rating, merit rating, employee
appraisal and employee evaluation. Performance appraisal is a process of assessment
individual employee performance and how it can be improved to contribute overall
organization performance (Grubb, 2007).
The theoretical framework shows the relationships of different variables and describes how
the hypotheses are formulated. The diagram given below describes the theoretical
relationships of the variable under investigation and their relationships with each other.
Performance appraisal
Employee
performance
Motivation
There are two hypothesis that relates with Performances appraisal and Employee
performance, the variable “Motivation” mediates in such a way that the relationship between
the impact of performance appraisal and employee performance.
H1: Motivation will mediate the relationship between Performance appraisal and Employee
performance.
H2: The relationship between performance appraisal and employee performance will be
mediated by Motivation.
7
Conceptual Framework:
The conceptual framework describes the operational details of the constructs used in the
study. Impact of performance appraisal on employee performance is inevitable. The model
presented by the researchers show the impact of performance appraisal on the employee’s
performance. According to this model first of all the management or the line managers set the
benchmark performance so that it helps in evaluating the performance of the employees
individually or collectively as a group. The benchmark performance may be preset standards,
best performance among the employees or any aspired performance/ results. Secondly the
actual or given performance of the employees should be matched with the benchmark
performance and two situations will appear, actual performance would be equal or above
benchmark performance or actual performance of the employees would be below benchmark
performance. If the actual performance of the employees is equal or above the required
performance then the employees would be satisfied and their motivation would be amplified
to be consistent and meet the next benchmark resulting in improved employee’s performance
But if the actual performance of the employees is below the benchmark performance, in the
result to an act of performance appraisal by the management, the gap between actual and
benchmark performance appears. This gap between actual and benchmark performance may
be because of deficiency in the employees knowledge, skills or attitude toward the assigned
job at workplace. The management’s role is now to identify the deficiency in the employee’s
knowledge, required skills or attitude and recommend and arrange the proper training
required to fulfill/ overcome the deficiency. Proper training to the employees would be
beneficial to the employees, equipping them with the latest knowledge and skills to perform
the assigned job more efficiently and effectively by creating the process interesting and
elevating employee’s motivation levels, and also to the organization to have workforce
equipped with latest knowledge, skills and other characteristics assisting and directing
organization to achieve ultimate organizational goal.
8
The diagram shows the details of the construct’s operational dimensions
Questionnaire:
Kindly tick (√) where applicable and do not indicate your names or personnel number.
3. Department:
4. Work experience
5. Educational Level
Primary school [ ] Secondary school [ ]
College level [ ] University level [ ]
6. Are you satisfied with the current performance appraisal process in enhancing your
motivation?
Highly satisfied [ ] Satisfied [ ] Neutral [ ] Dissatisfied []
Highly Dissatisfied [ ]
7. Are you satisfied with the current appraisers in enhancing your motivation?
Highly satisfied [ ] Satisfied [ ] Neutral [ ] Dissatisfied []
Highly Dissatisfied [ ]
10
8. Are you satisfied with the solutions put in place to deal with performance appraisal
problems that would enhance your motivation?
Highly satisfied [ ] Satisfied [ ] Neutral [ ] Dissatisfied []
Highly Dissatisfied [ ]
13
Coding of Variables in SPSS:
In the process of cording the variables in SPSS, we have 40 questions in which we construct
the name of the variable, which is used to refer to that variable in syntax. According to the
questionnaire, we construct 3 variables from the theoretical framework which is Independent
variable (Performance appraisal), Moderator( Motivation) and Dependent variable
( Employee performance).
The Explore procedure produces detailed univariate statistics and graphs for numeric scale
variables for an entire sample, or for subsets of a sample. It can also be used to assess the
normality of a numeric scale variable with special inferential statistics and detailed diagnostic
plots.
Interpretation of output:
The first table, the Case Processing Summary, shows how many valid values
there were. Since we selected pairwise missing data handling, the analysis is
using all complete information for each variable. We can see that there are
more missing values for variable is 0 than there are for variable CPEM.
Descriptives
Statisti Std.
c Error
Mean 2.75 .128
Lower
2.49
95% Confidence Bound
Interval for Upper
3.01
Mean Bound
My Median 3.00
Maximum 4
Range 3
Interquartile Range 1
Interpretation of output
The Descriptives box appears next. It has detailed univariate descriptive statistics for each of
the continuous variables, including skewness and kurtosis. The standard normal distribution
has skewness = 0 and kurtosis = 0, so we can interpret the sample skewness and kurtosis of
our variables in relation to that. For variable, the skewness is .72 (slightly right skewed) and
the kurtosis is .60 (slightly heavier tails than a normal distribution, but not by much). These
numbers alone aren't very good indicators of departures from normality, but they can
supplement the graphs and the normality tests.
17
Extreme Values
Case Number Value
1 1 3.67
2 2 3.67
Highes 3 29 3.42
t
4 21 3.33
5 5 3.08a
CPEM
1 4 2.08
2 19 2.17
Lowest 3 16 2.25
4 9 2.33
5 20 2.42
a. Only a partial list of cases with the value 3.08 are shown
in the table of upper extremes.
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
Statistic df Sig. Statistic df Sig.
CPEM .139 36 .075 .966 36 .316
Interpretation of output
The Tests of Normality table contains the Kolmogorov-Smirnov and Shapiro-Wilk tests for
both of our variables. For variable, the K-S and Shapiro-Wilk test p-values are both very
small (p < 0.001), so the decision to reject is very clear. However, for variable, the results are
not as clear-cut: the K-S p-value is p = 0.000 (which is just barely below the significance
level 0.075), and the Shapiro-Wilk p-value is p = 0.316. These tests are suggesting
contradictory conclusions:
Univariate Profiling: Examining the Shape of the Distribution:
A histogram is a graphical representation of a single variable that represents the frequency
of occurrences (data values) within data categories. The frequencies are plotted
to examine the shape of the distribution of values. If the integer values ranged from 1 to 10,
the researcher could construct a histogram by counting the number of responses for each
integer value. For continuous variables, categories are formed within which the frequency
of data values is tabulated. If examination of the distribution is to assess its normality (see
section on testing assumptions for details on this issue), the normal curve can be
superimposed on the distribution to assess the correspondence of the actual distribution to
the desired (normal) distribution. The histogram can be used to examine any type of metric
variable.
19
Univariate Profile: Stem & Leaf Diagram:
Interpretation:
The normal curve is also superimposed on the distribution. As will be shown in a later
section, empirical measures indicate that the distribution of CPEM deviates significantly from
the normal distribution. But how does it differ? The empirical measure that differs most is the
kurtosis, representing the peakedness or flatness of the distribution. The values indicate that
the distribution is flatter than excepted. What does the histogram show? The middle of the
distribution falls below the superimposed normal curve, while both tails are higher than
expected. Thus, the distribution shows no appreciable skewness to one side or the other, just a
shortage of observations in the center of the distribution. This comparison also provides
guidance on the type of transformation that would be effective if applied as a remedy for non-
normality. All of this information about the distribution is shown through a single histogram.
20
DESCRIPTIVE STATISTICS AND GRAPHICS CPEM
Interpretation of Output:
A Normal Q-Q (or Quantile-Quantile) Plot compares the observed quantiles of the data
(depicted as dots/circles) with the quantiles that we would expect to see if the data were
normally distributed (depicted as a solid line). If the data is approximately normally
distributed, the points will be on or close to the line. When looking at a Q-Q plot, you should
look for points that stray far from the line of expected values, as well as trends in the
observed values.
Interpretation:
The Detrended Normal Q-Q Plot shows the same information as the Normal Q- Q Plot, but in
a different manner. In the Detrended Plot, the horizontal line at the origin represents the
quantiles that we would expect to see if the data were normal; the dots represent the
magnitude and direction of deviation in the observed quantiles. Each dot is calculated by
subtracting the expected quantile from the observed quantile. (This implies that if a dot is
below the trend line on the Normal Q-Q plot, it will appear above the trend line on the
Detrended Normal Q-Q plot, because observed - expected > 0.)
22
Univariate Profile: Frequency Distribution: Variable CPEM
Statistics
CPEM
Valid 36
N
Missing 0
CPEM
Interpretation:
As we view the boxplot of CPEM , we do see substantial differences across the groups that
confirm the statistical results. We can also see that the primary differences are between
groups 2, 2.5, 3, 3.5 and versus group 4. Essentially, groups 2.5 and 3.0 seem about equal. If
we performed more statistical tests looking at each pair of groups separately, the tests would
confirm that the only statistically significant differences are group 2 versus 3 and group 3
versus 4. Also, we can see that group 2.50 to 3.00 has substantially more dispersion (a larger
box section in the boxplot), which prevents its difference from group 2. The boxplots thus
provide more information about the extent of the group differences of CPEM than just the
statistical test.
Multivariate Profile:
The researcher may desire to compare observations characterized on a multivariate profile,
whether it be for descriptive purposes or as a complement to analytical procedures. To
address this need, a number of multivariate graphical displays center around one of three
types of graphs. The first graph type is a direct portrayal of the data values, either by (a)
glyphs, or metro glyphs, which are some form of circle with radii that correspond to a data
value; or (b) multivariate profiles, which portray a bar like profile for each observation. A
second type of multivariate display involves a mathematical transformation of the original
data into a mathematical relationship, which can then be portrayed graphically.
25
Scatterplot Matrix
Interpretation:
In the interpretation of Multiple variate profiles, we made multiple variate graphical displays
of scatterplot Matrix and simple 3D scatter. We see the relationship between Independent and
Dependent variables on x-axis and y-axis on scatterplot matrix and we also see the same
relationship on simple 3D matrix, in which have two independent variables on x-axis and one
Dependent variable on y-axis, and the also using the Nominal variable of Gender on the top
of the diagram.
Simple 3D Scatter
New Measure of Association:
Recent research using non-normal data compared 12 different measures, including
Pearson correlation, Spearman’s rank-order correlation, various transformations (e.g.,
nonlinear transformations or the Rank-Based Inverse Normal Transformation), and
resampling approaches (e.g., the permutation test or boot strapping measures) .While the
Pearson correlation worked fairly well, the Rank-Based Inverse Normal Transformation
worked well in more situations across samples sizes and degrees of non-normality. The
significance of these results is not to propose using a new method of association per se, but
instead to expose researchers to the multiplicity of existing measures that may be more suited
to the task of measuring association across a variety of situations.
The rapid development of data mining, particularly in the arena of Big Data, brought to
attention the need for more sophisticated measures of association than the traditional
correlation coefficient. The capability for close examination of literally thousands of
relationships to see if the correlation captured the relationship was impossible. What was
needed were more “robust” measures of association/dependence which could assess the more
complicated patterns that might be encountered. To this end, several new measures have been
developed, including Hoeffding’s D, dCor (the distance correlation) and MIC (mutual
information correlation). Hoeffding’s D is a nonparametric measure of association based on
departures from independence and can work with many types of data. dCor is a distance-
based measure of association which also is more sensitive to nonlinear patterns in the data.
Perhaps the most interesting measure is MIC (mutual information correlation) which has been
shown as capable of not only identifying nonlinear relationships, but a wide range of distinct
patterns which are not of the traditional nonlinear type (e.g., two lines intersecting at an
angle, a line and parabola, a pattern like the letter X and an ellipse, along with many others).
Based on pattern matching, it provides a method for quickly scanning large amounts of data
for these more atypical relationships that would otherwise go undetected.
28
Pearson Correlation
Bootstrap
Bootstrap Specifications
Correlations
Impact of 36 36 36
Performance N
0 .022 .037
Bias
Appraisal Process on 0 .238 .293
Bootstrap Std. Error
Employee Motivation. . -.101 -.425
c
BCa 95% Confidence Lower
. .869 .733
Interval Upper .429 ** 1 .701**
Pearson Correlation .009 .000
Sig. (2-tailed) 36 36 36
.022 0 .003
Effect of Performance N
.238 0 .087
Appraisers on Bias
-.101 . .469
Employee Motivation Bootstrap Std. Error
.869 . .851
c
BCa 95% Confidence Lower .179 .701** 1
Interval Upper .296 .000
Pearson Correlation 36 36 36
Sig. (2-tailed) .037 .003 0
Challenges of .293 .087 0
N
Performance -.425 .469 .
Bias
Appraisal on .733 .851 .
Bootstrap Std. Error
Employee Motivation
c
BCa 95% Confidence Lower
Interval Upper
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
c. Unless otherwise noted, bootstrap results are based on 1000 bootstrap samples
Interpretation:
Pearsons Correlation SPSS Output provides a matrix of correlation coefficients for the three
variables. It also displays a matrix of significance values for these coefficients. Each variable
is perfectly correlated with itself (obviously) and so r = 1. Effect of job performance on
motivation is negatively related to challenges of performance appraisal on employee
motivation with a correlation coefficient of r = -0.101 which is significant at p < 0.001 (as
indicated by the ** after the coefficient). This significance value tells us that the probability
of this correlation being a ‘fluke’ is very low (close to zero in fact). Hence, we can have
confidence that this relationship is genuine and not a chance result. Usually, social scientists
accept any probability value below 0.05 as being statistically meaningful and so any
probability value below 0.05 is regarded as indicative of genuine effect. The output also
shows that impact of performance appraisal process on employee motivation is positively
related to the amount of time spent revising, with a coefficient of r =0.0429 , which is also
significant at p < 0.001. Finally, challenges of performance appraisal on employee motivation
appears to be negatively related to the amount of time spent revising (r = -0.425, p < 0.001).
Nonparametric Correlations:
Spearman’s correlation coefficient is a nonparametric statistic and so can be used when the
data have violated parametric assumptions and/or the distributional assumptions. Spearman’s
tests works by first ranking the data, and then applying Pearson’s equation to those ranks. As
an example of nonparametric data, a drugs company was interested in the effects of steroids
on cyclists. To test the effect they measured each cyclist’s position in a race (whether they
came first, second or third etc.) and how many steroid tablets each athlete had taken before
the race. Both variables are nonparametric, because neither of them was measured at an
interval level. The position in the race is ordinal data because the exact difference between
the ability of the runners is unclear. It could be that the first athlete won by several metres
while the remainder crossed the line simultaneously some time later, or it could be that first
and second place was very tightly contested but the remainder were very far behind.
30
Correlations
Impact of Challenges Employee
Performanc of Satisfaaction
e Appraisal Performance
Process on Appraisal on
Employee Employee
Motivation. Motivation
tailed) 36 36 36
Upper 36 36 36
Kendall's
Correlation -.007 .000 -.001
tau_b
Coefficient Sig. (2- .153 .000 .116
N .536 . -.135
Challenges of Performance
pc . -.071 -.597
BCa 95%
Spearman's . .645 .277
Correlation 36 36 36
pc BCa 95%
Lower Confidence Interval
Upper
Correlation
tailed)
Challenges of Performance N
N 36 36 36
c. Unless otherwise noted, bootstrap results are based on 1000 bootstrap samples
32
Partial Correction:
Correlations
Df 0 33
Impact of Performance
Appraisal Process on Bias .000 .031
Employee Motivation. Std. Error .000 .232
Bootstrapa
BCa 95% Confidence Lower . -.544
Effect of Performance Interval Upper . .323
Appraisers on
Correlation -.189 1.000
Employee Motivation
Significance (2-tailed) .278 .
Challenges of Df 33 0
Performance Appraisal
Bias .031 .000
on Employee
Std. Error .232 .000
Motivation
Bootstrapa
BCa 95% Confidence Lower -.544 .
Interpretation of Output:
SPSS output shows the Partial correlation of Effect of Performance Appraisal on employee
motivation and Impact of performance Appraisal on employee motivation, controlling for
revision time. Note that the top and bottom of the table contain identical values, so we ignore
one half of the table. The partial correlation between EPAM and IPEM is -.189, which is the
same value as we computed in equation. The value is considerably less then when we didn’t
adjust for the effect of revision time. Although the correlation is till statically significant ( its
p value still below then 0.05) and ( -0.544-0.323) still doesn’t contain zero, the relationship is
diminished.
33
In term of variance, the value of R square for the partial correlation 0.06, which means that
EPEM shares only 6.1% of the variance in the IPEM that is left over by revision time(
compared to 19.4% when revision time was not factored in ). Running this analysis has
shown us that EPEM alone does explain some of the variation in epem scores, but there is a
complex relationship between epem, revision and ipem that might otherwise have been
ignored
Regression Line
Descriptive Statistics
Mean Std. Deviation N
Challenges of Performance
Appraisal on Employee 2.8403 .36098 36
Motivation
Impact of Performance
Appraisal Process on 2.8409 .39916 36
Employee Motivation.
Effect of Performance
Appraisers on Employee 2.8397 .43060 36
Motivation
34
Correlations
Challenges of
Performance Appraisal on 1.000 .179 .701
Employee Motivation
Impact of Performance
Pearson Correlation Appraisal Process on .179 1.000 .429
Employee
Motivation.
Effect of Performance
Appraisers on .701 .429 1.000
Employee
Motivation
Challenges of
Performance Appraisal on . .148 .000
Employee
Motivation
Impact of
Sig. (1-tailed) Performance .148 . .005
Appraisal Process on
Employee Motivation.
Effect of Performance
Appraisers on .000 .005 .
Employee
Motivation
Challenges of
Performance Appraisal on 36 36 36
Employee
Motivation
Impact of
N Performance 36 36 36
Appraisal Process on
Employee Motivation.
Effect of Performance
Appraisers on 36 36 36
Employee Motivation
Model Summaryb
Interpretation Of Output
The first table provided by SPSS is a summary of the model that gives the value of R and R2
for the model. For these data, R is 0.714 and because there is only one predictor, this value
represents the simple correlation between advertising and album sales (you can confirm this
by running a correlation). The value of R2 is 0.510, which tells us that advertising
expenditure can account for 51.0% of the variation in CPEM. There might be many factors
that can explain this variation, but our model, which includes only advertising expenditure
explains 26%: 71% of the variation in CPEM is unexplained. Therefore, there must be other
variables that have an influence also.
Coefficientsa
36
Collinearity Diagnosticsa
Model Dimension Eigenvalue Condition Index Variance Proportions
Interpretation of Output:
For this model, the CPEM, t(36-3-1=32) = 12.26, p < .001, the amount of EPAM to release,
t(32) = 12.12, p < .001 and attractiveness of the band, t(32) = 4.55, p < .001, are all
significant predictors of CPEM.3 Remember that these significance tests are accurate only if
the assumptions discussed in our lecture are met.
If these assumptions aren’t met, or you want to ignore them, you could look at the table of
bootstrap confidence intervals for each predictor and their significance value3. These tell us
that Constant, b = 1.401 [0.360, 3.891], p = .001, IPEM, b = -0.135 [-1.104, 0.278], p = .001,
and EPEM, b = 0.641 [0.765, 5.673], p = .001, all significantly predict CPEM. The
confidence intervals are constructed such that in 95% of samples the boundaries contain the
population value of b. Therefore, it’s likely that the confidence interval we have constructed
for this sample will contain the true value of b in the population. Therefore, we can use the
confidence intervals to tell us the likely size of the parameter in the population (i.e., the true
value). If the confidence interval contains zero then this means that the true value might be
zero (i.e., no effect at all) or opposite to what we observed in the sample (e.g., a negative b
instead of.
37
Residuals Statisticsa
Minimum Maximum Mean Std. Deviation N
Charts:
38
40
Missing Values:
Missing data, where valid values on one or more variables are not available for analysis, are a
fact of life in multivariate analysis. In fact, rarely does the researcher avoid some form of
missing data problem. The researcher’s challenge is to address the issues raised by missing
data that affect the generalizability of the results. To do so, the researcher’s primary concern
is to identify the patterns and relationships underlying the missing data in order to maintain as
close as possible the original distribution of values when any remedy is applied. The extent of
missing data is a secondary issue in most instances, affecting the type of remedy applied.
These patterns and relationships are a result of a missing data process, which is any
systematic event external to the respondent (such as data entry errors or data collection
problems) or any action on the part of the respondent (such as refusal to answer) that leads to
missing values. The need to focus on the reasons for missing data comes from the fact that
the researcher must understand the processes leading to the missing data in order to select the
appropriate course of action.
Statistics
Valid 36
N
Missing 0
MVA (Missing Value Analysis) describes the missing value patterns in a data file (data
matrix). It can estimate the means, the covariance matrix, and the correlation matrix by using
listwise, pairwise, regression, and EM estimation methods. Missing values themselves can be
estimated (imputed).
N Mean Std. Deviation Missing No. of Extremesa
Count Percent Low High
42
a. Number of cases outside the range (Q1 - 1.5*IQR, Q3 + 1.5*IQR).
b. . indicates that the inter-quartile range (IQR) is zero.
39
Multiple Imputation
Warnings
Missing Values:
Interpretation of Output:
Missing value analysis helps address several concerns caused by incomplete data. If cases
with missing values are systematically different from cases without missing values, the
results can be misleading. Also, missing data may reduce the precision of calculated statistics
because there is less information than originally planned. Another concern is that the
assumptions behind many statistical procedures are based on complete cases, and missing
values can complicate the theory required. In our missing values of overall summary and
missing pattern is zero (0), which means there is no missing values in our data. In variable
pattern of missing data type is Non-missing.
44
Exploratory Factor Analysis
We introduce exploratory factor analysis as our first multivariate technique because it can
play a unique role in the application of other multivariate techniques. Broadly speaking,
factor analysis provides the tools for analyzing the structure of the interrelationships
(correlations) among a large number of variables (e.g., test scores or items, questionnaire
responses, social media usage patterns, digital tracking) by defining sets of variables that are
highly interrelated, known as factors or components. These groups of variables (factors or
components), which are by definition highly intercorrelated, are assumed to represent
dimensions within the data. If we are only concerned with reducing the number of variables,
then the dimensions can guide in creating new composite measures. However, if we have a
conceptual basis for understanding the relationships between variables, then the dimensions
may actually have meaning for what they collectively represent. In the latter case, these
dimensions may correspond to concepts that cannot be adequately described by a single
measure (e.g., store atmosphere is defined by many sensory elements that must be measured
separately but are all interrelated). Exploratory factor analysis presents several ways of
representing these groups of variables for use in other multivariate.
Factor analysis is frequently used to develop questionnaires: after all if we want to measure
an ability or trait, you need to ensure that the questions asked relate to the construct that you
intend to measure. Therefore I wanted to design a questionnaire to measure a trait that I
termed ‘SPSS Impact of Performance Appraisal on Employee Performance and Motivation’.
I decided to devise a questionnaire to measure various aspects of respondents’ Performance
appraisal of employees towards SPSS. I generated questions based on Impact of Performance
Appraisal on Employee Performance and Motivation ’and came up with 40 possible
questions to include. Each question was a statement followed by a five-point Likert scale
ranging from ‘strongly disagree’ through ‘neither agree or disagree’ to ‘strongly agree’
Descriptive Statistics
Mean
Initial Std. Deviation
Extraction Analysis N
EPM1
EPM1 1.94
1.000 .853 .826 36
EPM2
EPM2 2.08
1.000 .880 .770 36
EPM3
EPM3 2.42
1.000 .665 .692 36
EPM4
EPM4 1.75
1.000 .775 .770 36
IPEM5
IPEM5 3.06
1.000 .870 .893 36
IPEM6
IPEM6 2.78
1.000 .777 .591 36
IPEM7
IPEM7 2.83
1.000 .878 .697 36
IPEM8
IPEM8 2.89
1.000 .917 .919 36
IPEM9
IPEM9 2.89
1.000 .901 .887 36
IPEM10
IPEM10 2.58
1.000 .828 .841 36
IPEM11
IPEM11 2.64
1.000 .763 .867 36
IPEM12
IPEM12 2.97
1.000 .839 .696 36
IPEM13
IPEM13 2.78
1.000 .771 .637 36
IPEM14
IPEM14 3.11
1.000 .842 .523 36
IPEM15
IPEM15 2.72
1.000 .762 .882 36
EPEM16
EPEM16 2.50
1.000 .865 .811 36
EPEM17
EPEM17 2.75
1.000 .804 .841 36
EPEM18
EPEM18 2.64
1.000 .867 .798 36
EPEM19
EPEM19 3.00
1.000 .865 .862 36
EPEM20
EPEM20 2.61
1.000 .8591.022 36
EPEM21
EPEM21 3.14
1.000 .866 .683 36
EPEM22
EPEM22 2.92
1.000 .871 .806 36
EPEM23
EPEM23 3.08
1.000 .759 .874 36
EPEM24
EPEM24 2.83
1.000 .825 .737 36
EPEM25
EPEM25 2.97
1.000 .844 .560 36
EPEM26
EPEM26 2.78
1.000 .851 .866 36
EPEM27
EPEM27 2.86
1.000 .844 .867 36
EPEM28
EPEM28 2.83
1.000 .814 .878 36
CPEM29
CPEM29 2.72
1.000 .848 .914 36
CPEM30
CPEM30 2.86
1.000 .906 .762 36
CPEM31
CPEM31 2.94
1.000 .780 .826 36
CPEM32
CPEM32 2.58
1.000 .827 .770 36
CPEM33
CPEM33 2.86
1.000 .871 .683 36
CPEM34
CPEM34 3.00
1.000 .828 .756 36
CPEM35
CPEM35 3.08
1.000 .860 .554 36
CPEM36
CPEM36 2.75
1.000 .739 .770 36
CPEM37
CPEM37 2.50
1.000 .848 .845 36
CPEM38
CPEM38 2.92
1.000 .767 .692 36
CPEM39
CPEM39 2.81
1.000 .875 .856 36
CPEM40
CPEM40 3.06
1.000 .863 .630 36
Communalities
Interpretation Of Output:
Component Initial Eigenvalues Extraction Sums of Squared Loadings Rotation Sums of Squared Loadings
SPSS Output lists the eigenvalues associated with each linear component (factor) before
extraction, after extraction and after rotation. Before extraction, SPSS has identified 40 linear
components within the data set (we know that there should be as many eigenvectors as there
are variables and so there will be as many factors as variables). The eigenvalues associated
with each factor represent the variance explained by that particular linear component and
SPSS also displays the eigenvalue in terms of the percentage of variance explained (so, factor
1explains 18.089% of total variance). It should be clear that the first few factors explain
relatively large amounts of variance (especially factor 1) whereas subsequent factors explain
only small amounts of variance. SPSS then extracts all factors with eigenvalues greater than
1, which leaves us with four factors. The eigenvalues associated with these factors are again
displayed (and the percentage of variance explained) in the columns labelled Extraction Sums
of Squared Loadings. The values in this part of the table are the same as the values before
extraction, except that the values for the discarded factors are ignored (hence, the table is
blank after the fourth factor).
In the final part of the table (labelled Rotation Sums of Squared Loadings), the eigenvalues of
the factors after rotation are displayed. Rotation has the effect of optimizing the factor
structure and one consequence for these data is that the relative importance of the four factors
is equalized. Before rotation, factor 1 accounted for considerably more variance than the
remaining three (18.089% compared to 9.257, 8.125, and 7.672%), however after extraction
it
accounts for only 10.939% of variance (compared to 10.884, 10.369 and 8.947% respectively).
49
Interpretation of Output:
We can also use the scree plot, which we asked SPSS to produce. The scree plot is shown
below with a thunderbolt indicating the point of inflexion on the curve. This curve is difficult
to interpret because the curve begins to tail off after three factors, but there is another drop
after four factors before a stable plateau is reached. Therefore, we could probably justify
retaining either two or four factors. Given the large sample, it is probably safe to assume
Kaiser's criterion; however, you could rerun the analysis specifying that SPSS extract only
two factors and compare the results.
Component Matrixa
Component
1 2 3 4 5
This output also shows the component matrix before rotation. This matrix contains the
loadings of each variable onto each factor. By default SPSS displays all loadings; however, we
requested that all loadings less than 0.4 be suppressed in the output and so there are blank
spaces for many of the loadings. This matrix is not particularly important for interpretation.
At this stage SPSS has extracted four factors. Factor analysis is an exploratory tool and so it
should be used to guide the researcher to make various decisions: you shouldn't leave the
computer to make them. One important decision is the number of factors to extract. By
Kaiser's criterion we should extract four factors and this is what SPSS has done. However,
this criterion is accurate when there are less than 40 variables and communalities after
extraction are greater than 0.7 or when the sample size exceeds 150 and the average
communality is greater than 0.6. The communalities are shown in SPSS Output 4, and none
exceed 0.7. The average of the communalities can be found by adding them up and dividing
by the number of communalities (11.573/23 = 0.503). So, on both grounds Kaiser's rule may
not be accurate. However, you should consider the huge sample that we have, because the
research into Kaiser's criterion gives recommendations for much smaller samples.
Rotated Component Matrixa
Component
1 2 3 4 5
Interpretation Of Output:
SPSS Output shows the rotated component matrix (also called the rotated factor matrix in
factor analysis) which is a matrix of the factor loadings for each variable onto each factor.
This matrix contains the same information as the component matrix in SPSS Output of
commonalities and component factor except that it is calculated after rotation. There are
several things to consider about the format of this matrix. First, factor loadings less than 0.4
have not been displayed because we asked for these loadings to be suppressed. If you didn't
select this option, or didn't adjust the criterion value to 0.4, then your output will differ.
Second, the variables are listed in the order of size of their factor loadings because we asked
for the output to be Sorted by size. If this option was not selected your output will look
different. Finally, for all other parts of the output I suppressed the variable labels (for reasons
of space) but for this matrix I have allowed the variable labels to be printed to aid
interpretation. Compare this matrix with the unrotated solution. Before rotation, most
variables loaded highly onto the first factor and the remaining factors didn't really get a look
in. However, the rotation of the factor structure has clarified things considerably: there are
four factors and variables load very highly onto only one factor (with the exception of one
question). The suppression of loadings less than 0.4 and ordering variables by loading size
also makes interpretation considerably easier (because you don't have to scan the matrix to
identify substantive loadings).
54
Component Transformation Matrix
Component 1 2 3 4 5
Interpretation Of Output:
IPEM12 .792
EPEM24 .713
EPEM25 .687
IPEM7 .656
EPEM16 .633
IPEM11 .614
CPEM36 .571
CPEM32 .491
CPEM33 .436 .435
EPEM20
CPEM30
CPEM38
CPEM35 .817
EPEM19 .796
EPEM22 .742
CPEM34 .667
EPEM23 .603
CPEM29 .456
EPEM21 .433
EPEM26 .409
EPEM28
IPEM13
CPEM31 .727
CPEM39 .707
EPEM27 .596
EPM1 -.569
EPM3 -.518
EPEM17 .491
EPM2 -.468
EPEM18 .449
IPEM14 .434
IPEM8
CPEM37
IPEM6 .805
IPEM5 .683
CPEM40 -.633
IPEM10 .449 .611
IPEM15 .438
IPEM9 .402
EPM4
Extraction Method: Principal Component Analysis.
Rotation Method: Promax with Kaiser Normalization.
a. Rotation converged in 8 iterations.
Interpretation of Output:
The pattern matrix holds the loadings. Each row of the pattern matrix is essentially a
regression equation where the standardized observed variable is expressed as a function of
the factors The structure matrix holds the correlations between the variables and the factors.
Interpretation of a set of oblique factors involves both the pattern and structure matrices, as
well as the factor correlation matrix. The latter matrix contains the correlations among all
pairs of factors in the solution. It is automatically printed for an oblique solution when the
rotated factor matrix is printed.
When the rotation is orthogonal (i.e. the factors are uncorrelated; orthogonal and uncorrelated
are synonymous with centered variables), then the rotated factor matrix represents both the
loadings and the correlations between the variables and factors. For oblique rotations, where
the factors are allowed to correlate (oblimin or promax in SPSS), then the loadings and
correlations are distinct. The pattern matrix holds the loadings. Each row of the pattern matrix
is essentially a regression equation where the standardized observed variable is expressed as a
function of the factors. The loadings are the regression coefficients. The structure matrix
holds the correlations between the variables and the factors.
In this output table of pattern matrix, we have 4 variable of components that are single
loaded, but we have only two variables which are double loaded and those are CPEM 33 and
IPEM 10.
57
Structure Matrix
Component
1 2 3 4
IPEM12 .667
EPEM16 .645
EPEM24 .645
CPEM33 .635 .461 .631
EPEM25 .632
IPEM11 .626
CPEM36 .560
IPEM7 .552
EPEM20 .548 .521
CPEM32 .517
IPEM15 .436 .424
CPEM30 .404
CPEM38
CPEM35 .743
EPEM22 .719
EPEM19 .686
CPEM34 .677
EPEM23 .631
EPEM21 .540 .586 .432
EPEM26 .489 .527
CPEM29 .497
EPEM28 .414 .450 .425
IPEM13
CPEM31 .642
EPM1 -.606
EPEM27 .567
EPM3 -.553
CPEM39 .548
EPM2 -.444 -.534
EPEM17 .503
EPEM18 .432
CPEM37 .427
IPEM14
IPEM8
IPEM6 .798
IPEM5 .696
CPEM40 .441 .456 -.655
IPEM10 .593
IPEM9 .410
EPM4
Extraction Method: Principal Component Analysis.
Rotation Method: Promax with Kaiser Normalization.
In the output table of structure matrix holds the correlations between the variables and the
factors. Interpretation of a set of oblique factors involves both the pattern and structure
matrices, as well as the factor correlation matrix. The latter matrix contains the correlations
among all pairs of factors in the solution. Interpretation of a set of oblique factors involves
both the pattern and structure matrices, as well as the factor correlation matrix. The latter
matrix contains the correlations among all pairs of factors in the solution. It is automatically
printed for an oblique solution when the rotated factor matrix is printed.
In this output table of structure matrix, we have 4 variable of components that are single
loaded, but we have different variables which are double loaded and those are CPEM 33,
EPEM 20, IPEM 15, EPEM 21,26,28, EPM2, CPEM40.
59
References:
Impact of performance appraisal on employee’s performance involving the
Moderating Role of Motivation(August 2013). Nadeem Iqbal( Faculty of Management
Sciences, Baha Uddin Zakariya University Multan, Pakistan), Naveed Ahmad
(Faculty of Management sciences, Indus international institute, D. G. Khan, Pakistan),
Zeeshan Haider (MBA student Baha-uddin Zakariya university, Pakistan), Yumna
Batool (MBA student Baha-uddin Zakariya university, Pakistan). Qurat-ul-ain (MBA
student Baha- uddin Zakariya university, Pakistan).
60