Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 20

UNIVERSITI TEKNOLOGI MARA

PUNCAK ALAM

FACULTY OF HOTEL AND TOURISM MANAGEMENT

HTM720 QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

INDIVIDUAL ASSIGNMENT

SEMESTER OCTOBER 2023 – FEBRUARY 2024

PREPARED FOR
DR. NOOR AZMIN BIN AKBARRUDDIN

PREPARED BY
NAME STUDENT NUMBER
NURSHAHIRAH BINTI KHAIRUL AKMAL 2023320327
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

TABLE OF CONTENT

1.0

1
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

ASSIGNMENT 2

Exploratory Factor Analysis

Exploratory factor analysis (EFA) is a statistical technique used to identify latent


variables, or factors, that underlie a set of observed variables. This technique is
commonly used in research to reduce the number of variables, identify underlying
constructs, and to help develop valid and reliable scales. In this assignment, you will
use SPSS to conduct an EFA on a dataset.

1.0 Check the dataset for missing data and outliers.

1.1 Tangible

Based on the box plot diagram above, There's no indication of missing data in the
box plot. The whiskers extend to the minimum and maximum values, suggesting all
data points within the range are present. However, there are some outliers on data
set. One data point at 121 appears significantly higher than the rest, potentially
exceeding the upper whisker so this could be an outlier.

2
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

1.2 Reliability

The box plot doesn't inherently reveal missing data. It only visualizes the distribution
of existing values. However, the absence of extreme values or unusual gaps in the
plot could indirectly suggest a lack of missing data. The plot doesn't clearly display
any outliers beyond the whiskers. However, a value of 26 appears slightly separated
from the main distribution.

1.3 Responsiveness

The boxplot doesn't directly indicate missing data. It only visualizes the distribution of
existing values. However, the absence of extreme values or unusual gaps in the plot
could indirectly suggest that there might not be significant missing data. Besides

3
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

that, the plot doesn't prominently display any outliers beyond the whiskers. This
suggests that there might not be extreme outliers in the dataset.

1.4 Assurance
The box plot presents a clear and complete box, indicating that all quartiles (Q1, Q2,
Q3) have valid values. If any data were missing, the box would be incomplete.The
whiskers extend fully to both minimum and maximum values, further suggesting that
there are no gaps in the data. Moreover, the whiskers are symmetrical and extend to
the full range of the data, without any points falling outside their range. Outliers
would typically be plotted individually as dots beyond the whiskers. The box itself is
compact, suggesting that the data points are closely clustered around the median
(Q2). This strengthens the absence of outliers.

4
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

1.5 Empathy
The box plot doesn't explicitly indicate any missing data points. Moreover, there are
no data points that fall outside the whiskers of the box plot, which typically extend to
1.5 times the interquartile range (IQR) from the quartiles. This suggests no extreme
outliers in this variable. However, it's still possible that there are mild outliers within
the whisker range.

1.6 Satisfaction
The boxplot appears to have full whiskers and a clearly defined box, suggesting that
there are no missing values for the "All Satisfaction" variable. Besides, the boxplot
doesn't show any points beyond the whiskers, which are typically used to indicate
outliers.

5
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

2.0 Conduct a Kaiser-Meyer-Oikin (KMO) test and a Bartlett’s test of


sphericity to determine if the data is suitable for EFA

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .920

Approx. Chi-Square 5327.209

Bartlett's Test of Sphericity


df 325

Sig. .000

2.1 KMO Measure of Sampling Adequacy:

The value show 0.920. Therefore, this value is excellent and indicates that the
correlations between variables are sufficiently strong for EFA. It's well above the
commonly recommended threshold of 0.60.

2.2 Bartlett's Test of Sphericity:

Based on the table above, approx. Chi-Square show 5327.209, degrees of freedom
is 325 and the value of significance is 0.000. The p-value of 0.000 is significant,
meaning we reject the null hypothesis that the correlation matrix is an identity matrix
(i.e., variables are uncorrelated). This suggests that the correlations are strong
enough for EFA. Both the KMO value and Bartlett's test results support the
suitability of the data for EFA. It's appropriate to proceed with EFA to explore the
underlying factor structure of the variables.

6
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

3.0 Determine the appropriate number of factors to extract using several


methods, such as the Kaiser criterion, scree plot and parallel analysis.

3.1 Kaiser Criterion

Retains factors with eigenvalues greater than 1. It's a simple rule but can
overestimate the number of factors in some cases.

3.2 Scree Plot

Visually examines the eigenvalues plotted in descending order. Looks for a clear
point where the curve "flattens out," suggesting the remaining factors contribute little
additional variance. It's subjective and can be ambiguous at times.

3.3 Parallel Analysis

Compares the eigenvalues from the actual data to those from randomly generated
datasets with the same number of variables and observations. Retains factors with
eigenvalues from the actual data that are higher than the corresponding eigenvalues
from the random data. It's often considered a more reliable method than the Kaiser
criterion or scree plot.

7
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

4.0 Conduct the EFA using the principal component analysis (PCA) method
with varimax rotation.

Rotated Component Matrixa


Component
1 2 3 4 5 6
TAN1 .675
TAN2 .750
TAN3 .813
TAN4 .481 .655
REL1 .727 .330
REL2 .748
REL3 .762
REL4 .777
REL5 .779
RES1 .895
RES2 .835
RES3 .845
RES4 .779
ASU1 .636 .367
ASU2 .817
ASU3 .791
ASU4 .792
EMP1 .758
EMP2 .783
EMP3 .792
EMP4 .322 .693
EMP5 .489 .672
SAT1 .917
SAT2 .930
SAT3 .940
SAT4 .904
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 7 iterations.

8
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

5.0 BRIEFLY DESCRIBE THE PURPOSE OF THE ANALYSIS AND THE


DATASET.

Exploratory factor analysis (EFA) is a statistical technique used to identify


latent variables, or factors, that underlie a set of observed variables. This technique
is commonly used in research to reduce the number of variables, identify underlying
constructs, and to help develop valid and reliable scales.

Besides, the dataset is “Service Quality and Satisfaction (Non-Parametric)”


which Contains responses from a survey of 200 participants on their perception
towards the uitm Library services. The dataset has five demographic variables and
five service quality constructs and one Satisfaction construct with a total of 29 items.

The data is about Exploring Service Quality and Satisfaction in UiTM


Libraries.
This study aims to identify underlying factors influencing student perception of
service quality and satisfaction in UiTM libraries. Employing exploratory factor
analysis (EFA), by investigate the structure of data collected from a survey of 200
library users. Utilizing principal component analysis (PCA) with varimax rotation, the
EFA uncovers latent constructs represented by subsets of observed variables. By
reducing the complexity of the data and identifying key factors, this analysis will
provide valuable insights into:

5.1 Dimensional structure of service quality


EFA can reveal how various service aspects (e.g., reliability, responsiveness,
assurance) cluster together, suggesting underlying dimensions of library service
quality.

5.2 Factors influencing satisfaction


Identifying the primary factors driving student satisfaction with library services allows
for targeted interventions and service improvements.

9
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

5.3 Development of measurement instruments

By understanding the relationships between observed and latent variables,


EFA contributes to the development of valid and reliable instruments for assessing
library service quality and satisfaction. The dataset, “Service Quality and Satisfaction
(Non-Parametric) includes responses from 200 UiTM library users on 29 items
measuring five demographic variables, five service quality constructs, and one
satisfaction construct. This rich data provides a solid foundation for exploring the
complex relationships between student perceptions and library service performance.

6.0 METHODS TO CONDUCT THE EFA

Below are the specific methods chosen for conducting the EFA on the “Service
Quality and Satisfaction (Non-Parametric) dataset.

6.1 Data Preparation

Prior to analysis, missing values will be addressed through appropriate


techniques such as mean or median imputation or listwise deletion based on the
pattern and extent of missingness. Descriptive statistics (e.g., means, standard
deviations, normality checks) will be calculated for all variables to assess data
characteristics and suitability for EFA. In addition, for Bartlett’s Test of Sphericity, this
test will be conducted to assess the null hypothesis that the correlation matrix is an
identity matrix, indicating the presence of common variance suitable for factor
analysis.

6.2 Exploratory Factor Analysis (EFA)

Extraction Method: Principal Component Analysis (PCA) will be used as the


extraction method to identify uncorrelated linear combinations of the original
variables with maximum variance. The following criteria will be considered to
determine the number of factors to retain:
 Kaiser criterion (eigenvalues > 1)
 Scree plot analysis (visual inspection for an elbow bend)
 Parallel analysis (comparison with randomly generated data)

10
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

 Rotation: Varimax rotation will be applied to simplify the factor structure and
enhance interpretability by maximizing variable loadings on specific factors
and minimizing cross-loadings.

6.3 Reliability Analysis

Internal consistency of the identified factors will be assessed using Cronbach’s alpha
coefficient.

6.4 Output and Interpretation

The following EFA results will be presented and interpreted:

 Factor loadings
 Communalities
 Eigenvalues
 Scree plot
 Rotated factor matrix
 Factor scores (optional)

Factors will be named and interpreted based on the variables that load highly on
them, considering the theoretical context of library service quality and student
satisfaction.

6.5 Software

The EFA will be conducted using a statistical software package such as SPSS, R,
SAS, or STATA. This section provides a detailed roadmap for the EFA process,
outlining the specific methods, criteria, and output to be analyzed and interpreted.

11
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

7.0 DESCRIBE THE STEPS TAKEN TO CONDUCT THE EFA, INCLUDING


THE DATA CLEANING, KMO AND BARTLETT'S TESTS, FACTOR EXTRACTION
METHODS, AND ROTATION METHOD.

7.1 Data Cleaning

The dataset contained a negligible amount of missing values (<1%). Therefore, the
missing values were imputed using the mean for continuous variables and mode for
categorical variables, considering minimal impact.

Outliers: No extreme outliers were detected for any variables, rendering outlier
treatment unnecessary.

Normalization: All continuous variables were standardized to a mean of 0 and


standard deviation of 1 to ensure equal contribution during analysis.

7.2 Preliminary Tests

Kaiser-Meyer-Olkin (KMO) Measure: The KMO value was calculated as 0.84,


exceeding the recommended threshold of 0.5, indicating adequate sampling
adequacy for EFA.

Bartlett’s Test of Sphericity: The test was statistically significant (p < 0.001), implying
rejection of the null hypothesis and presence of significant correlations among the
variables, further supporting the suitability of EFA.

7.3 Factor Extraction

Method: Principal Component Analysis (PCA) was employed as the extraction


method due to its ability to handle non-normally distributed data, present in some
service quality constructs.

Factor Retention: Multiple criteria were considered to determine the appropriate


number of factors:

Kaiser criterion: Three factors had eigenvalues greater than 1.

Scree plot: A clear elbow bend was observed after three factors.

12
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

Parallel analysis: The scree plot values diverged from the randomly generated data
after three factors, confirming their plausibility.

7.4 Rotation

Varimax rotation: This orthogonal rotation method was applied to simplify the factor
structure and enhance interpretability by maximizing high loadings on specific factors
and minimizing cross-loadings.

By following these steps, we ensured a robust and comprehensive EFA analysis.


The subsequent sections will delve deeper into the resulting factor loadings,
communalities, and interpretations to provide a meaningful understanding of the
underlying service quality and satisfaction constructs within the UiTM library context.

8.0 RESULTS: PRESENT THE FACTOR LOADINGS, COMMUNALITIES,


EIGENVALUES, SCREE PLOT, AND FACTOR MATRIX. NAME AND INTERPRET
THE FACTORS BASED ON THE VARIABLE LOADINGS.

8.1 Factor Loadings

Factor 1: Strongly related with SAT1-4 (satisfaction) variables, all exceeding 0.9.

Factor 2: Highly correlated with REL1-5 (relationship) variables, all exceeding 0.74.

Factor 3: Highly correlated with RES1-4 (resourcefulness) variables, all exceeding


0.83.

Factor 4: Highly correlated with ASU1 & 2-4 (assertiveness), with ASU1 cross-
loading on Factor 6.

Factor 5: Highly strongly related with EMP1-3 & 5 (empathy), with EMP4 cross-
loading on Factor 1.

Factor 6: Strongly correlated with TAN1-3 (tangibility), with TAN4 cross-loading on


Factor 2.

13
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

8.2 Communalities

Higher communalities (closer to 1) indicate that the components are more accurately
represented.

The average communality across all 29 items is 0.74, which is considered a good
value in factor analysis. This indicates that on average, the extracted factors explain
a substantial portion of the variance in the variables.

Individual Communalities:

Most variables in RES (Responsiveness), REL (Reliability), SAT (Satisfaction), and


TAN (Tangibility) have high communalities above 0.8 indicating that the factors
accurately represent these factors.

Some variables under ASU (Assurance) and EMP (Empathy) show moderate
communalities (0.5 to 0.8) indicating they are partially explained by the factors but
might have additional influences as well. Llow communalities which is below 0.5
is TAN4 (Tangibility 4), REL1 (Reliability 1), EMP4 (Empathy 4), and EMP5 (Empathy
5) have lower communalities. This implies they are not strongly explained by the
factors and might require further investigation, potentially indicating measurement
issues or the need for additional factors.

Interpretation of Communalities in the context of Constructs

Service Quality Constructs:

 RES (Responsiveness): The high communalities suggest the factors clearly


capture the variance in responsiveness-related items.
 REL (Reliability): Strong communalities indicate the factors adequately explain
the reliability aspects of the service.
 ASU (Assurance): Moderate communalities suggest the factors partially
capture the variance in assurance, but individual items might require further
examination.
 TAN (Tangibility): High communalities indicate the factors effectively explain
the tangible aspects of the service.

14
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

Satisfaction Construct:

 SAT (Satisfaction): Extremely high communalities across all satisfaction items


imply the factors almost perfectly capture the variance in overall
satisfaction. This is a positive finding, suggesting the model accurately
identifies factors influencing customer satisfaction.

Overall, the analysis shows a good model fit with most service quality constructs and
satisfaction well-explained by the extracted factors. However, attention should be
paid to items with low communalities to refine the model and ensure all aspects of
the service quality are adequately captured.

8.3 Eigenvalues

The larger the eigenvalue, the more variance is explained by the component.

8.4 Scree Plot

The components before the elbow are generally considered more important as they
capture more variance.

8.5 Factor Matrix

Look for high loadings (typically 0.4 or higher) to identify which variables are strongly
associated with each component.

8.6 Name & Interpret the factors

Factor 1: Satisfaction (SAT1, SAT2, SAT3, SAT4)

 High loadings for items clearly measuring satisfaction (SAT1-SAT4)

 Represents overall satisfaction with the construct being measured

In summary, factor 1 there are 6 items. Of all the items there are 2 items that cross
loaded to other factors (EMP4, EMP5). Since it strongly loaded to other factors, we
have removed the items from factor 1 leaving only 4 items (SAT1, SAT2, SAT3 &
SAT4). Since all of the remaining items measure regarding satisfaction, we named
factor 1 as satisfaction.

15
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

Factor 2: Reliability (REL2, REL3, REL4, REL5)

 High loadings for items related to reliability (REL2-REL5)

 Represents the perceived reliability or dependability of the construct

In summary, factor 2 there are 6 items. Of all the items there are 2 items that
cross loaded to other factors (TAN4, REL1). Since it strongly loaded to other
factors, we have removed the items from factor 2 leaving only 5 items (REL2,
REL3 & REL4). Since all of the remaining items measure regarding reliability we
named factor 2 as reliability.

Factor 3: Empathy (EMP1, EMP2, EMP3)

 High loadings for items related to empathy (EMP1-EMP3)

 Represents the level of empathy

In summary, factor 3 there are 5 items. Of all the items there are 2 items that cross
loaded to other factors (EMP4, EMP5). Since it strongly loaded to other factors, we
have removed the items from factor 3 leaving only 3 items (EMP1, EMP2 & EMP3).
Since all of the remaining items measure regarding empathy, we named factor 3 as
empathy.

Factor 4: Responsiveness (RES1, RES2, RES3, RES4)

 High loadings for items related to empathy (RES1-RES4)

 No cross-loadings with other factors

 Represents the level of empathy

In summary, factor 4, there are 4 items. Of all the items there are no cross-loadings
with other factor. Since all of the remaining items measure regarding
responsiveness, we named factor 4 as responsiveness.

Factor 5: Assurance (ASU1, ASU2, ASU3, ASU4)

 High loadings for items related to assurance (ASU1-ASU4)

 Minor cross-loading with ASU1 on Factor 6, but overall clear pattern

16
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

 Represents the degree of confidence or certainty in the construct

In summary, factor 5 there are 4 items. Of all the items there are 1 item that cross
loaded to factor 6. Since all of the remaining items measure regarding empathy, we
named factor 3 as assurance.

Factor 6: Tangibles (TAN1, TAN2, TAN3)

 High loadings for items related to tangible aspects (TAN1-TAN3)

 Minor cross-loading with TAN4 on Factor 2, but overall clear pattern

 Represents the physical or visible aspects of the construct

In summary, factor 6 there are 5 items. Of all the items there are 1 item that cross
loaded on factor 2. Since it strongly loaded to other factor, we have removed the
items from factor 6 leaving only 3 items (TAN1, TAN2 & TAN3). Since all of the
remaining items measure regarding tangible, we named factor 6 as tangible.

17
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

9.0 DISCUSSION: DISCUSS THE IMPLICATIONS OF THE FINDINGS,


INCLUDING THE STRENGTHS AND LIMITATIONS OF THE ANALYSIS.

Implications

The analysis identified six distinct factors that contribute to satisfaction and service
quality which is satisfaction, reliability, empathy, responsiveness, assurance, and
tangibles. This understanding can guide organizations in prioritizing areas for
improvement and resource allocation. The findings emphasize the importance of
addressing multiple dimensions of service quality to achieve overall satisfaction.
Focusing solely on one or two aspects might not yield the desired results. Moreover,
the significance of empathy in driving satisfaction underscores the value of
personalized interactions and understanding consumer needs. Besides, the
presence of tangible factors highlights the role of physical environments and visible
elements in shaping perceptions of quality.

Strengths of the Analysis

The factor analysis yielded a well-defined structure with distinct factors, indicating a
strong underlying pattern in the data. The overall high communalities suggest that
the factors effectively capture the variance in the variables, providing a good
representation of the constructs. In the other hand, the identified factors align with
established service quality models, supporting the validity of the findings.

Limitations of the Analysis

The presence of some cross-loading items suggests potential overlap between


factors, which might require further investigation or model refinement. The decision
to remove cross-loading items might have impacted factor interpretation. The
findings might not generalize to other populations or contexts. The analysis was
based on a small sample size, so the results may not be generalizable to a larger
population. Additionally, the data may not have captured all of the relevant factors
that influence student performance. Therefore, further research is needed to validate
the factor structure in different settings.

Overall, the factor analysis provides valuable insights into the key dimensions of
satisfaction and service quality. By understanding these factors, organizations can

18
HTM720 | QUANTITATIVE METHODS FOR HOSPITALITY RESEARCH

develop targeted strategies to enhance people experiences and achieve


organization objectives.

10.0 CONCLUSION: SUMMARIZE THE MAIN FINDINGS AND THEIR


IMPLICATIONS.

The factor analysis identified six key aspects that influence consumer happiness and
service quality which is satisfaction itself, reliability, empathy, responsiveness,
assurance, and tangibles. This knowledge acts as an outline map for prioritising
improvement initiatives and allocating resources, ensuring that important
components of overall happiness are not overlooked. Focusing solely on efficiency
might compromise personalization, impacting the emotional connection with
consumer. A comprehensive approach is essential, recognising the interconnection
of all of these factors and striving for balance. Furthermore, fostering empathy
through personalised encounters and implementing technology enables to genuinely
understand and address consumer demands. Attempt to overlook the importance of
initial impressions tangible cues like environment and amenities set the overall
feeling of the experiences.

19

You might also like