INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Customer Sentiment on Telecommunication Quality of Service  
Delivery in Ghana  
Roland Yaw Kudozia, Nii Ayitey Komey, Carolyn Elizabeth Kudozia, Daniel Owusu-Donkor  
Gdirst Institute  
DOI : https://doi.org/10.51583/IJLTEMAS.2025.1411000063  
Received: 26 November 2025; Accepted: 01 December 2025; Published: 09 December 2025  
ABSTRACT  
Aim/Purpose  
This study investigates customer sentiment toward telecom Quality of Service Delivery (QoSD) in Ghana by  
integrating dual sentiment, that is, an expressed emotional sentiment from customer narratives and experienced  
service sentiment derived from SERVQUAL ratingsinto a unified service quality evaluation framework.  
Background  
Although SERVQUAL-based research consistently links service quality dimensions to customer satisfaction,  
far less is known about how emotional expressions interact with cognitive service evaluations in developing  
markets. This gap limits understanding of how users interpret service failures and form satisfaction judgments  
in resource-constrained telecom environments.  
Methodology  
A nationwide survey of 536 mobile users was analyzed using a hybrid analytical approach: (1) natural language  
processing (NLP) sentiment scoring of open-ended responses, and (2) construction of a Service Sentiment Index  
(SSI) based on standardized SERVQUAL dimension scores. Mediation analysis using Hayes’ PROCESS tested  
whether experienced sentiment serves as a psychological pathway linking reliability to overall satisfaction.  
Contribution  
The study introduces and empirically validates a dual-sentiment framework for telecom QoSD assessment,  
demonstrating significant divergence between customers’ expressed emotional sentiment and their experienced  
service sentiment. It extends the SERVQUAL model by incorporating affective evaluation processes and  
provides empirical evidence that sentiment partially mediates the reliability–satisfaction relationship.  
Findings  
Expressed sentiment was predominantly negative (56.5%), whereas SSI reflected more moderate evaluations.  
Reliability, assurance, and empathy emerged as key predictors of both sentiment and satisfaction. Mediation  
results confirmed that experienced sentiment explains a substantial portion of the total effect of reliability on  
satisfaction, supporting a dual cognitive–affective interpretation of QoSD.  
Recommendations for Practitioners  
Telecom providers should prioritize network reliability, transparent communication, and empathetic customer  
engagement to address emotional dissatisfaction. The presence of strong negative expressed sentimenteven  
among customers reporting moderate satisfactionsignals potential reputational and churn risks requiring  
proactive management.  
www.ijltemas.in  
Page 682  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Recommendations for Researchers  
Researchers should integrate text analytics with psychometric models in service quality studies and further  
examine sentiment as a mediator across diverse service industries and cultural settings.  
Impact on Society  
Understanding sentiment divergence enhances the design of customer-centric QoS interventions, strengthens  
regulatory oversight, and supports the development of more responsive telecom service policies in emerging  
markets.  
Future Research  
Longitudinal studies should explore sentiment trajectories over time, while cross-market comparative research  
should validate the generalizability of the dual-sentiment framework across other developing and developed  
economies.  
Keywords: Customer Sentiment; Dual Sentiment; QoSD; SERVQUAL; Mediation; NLP Sentiment;  
Telecommunications; Ghana.  
INTRODUCTION  
Telecommunication services constitute a critical enabler of socioeconomic development in Ghana, supporting  
commerce, education, financial inclusion, emergency communication, and everyday interpersonal interaction.  
Over the past decade, the sector has experienced rapid expansion driven by increased mobile penetration,  
widespread adoption of mobile money, and continuous investment in digital infrastructure. These advances have  
heightened consumer expectations, placing Quality of Service Delivery (QoSD) at the center of competitive  
differentiation and regulatory scrutiny. As mobile users increasingly depend on data-driven applications, reliable  
connectivity, transparent pricing, and efficient customer support have become essential components of the  
consumer experience.  
Quality of Service Delivery in telecommunications is traditionally conceptualized through models such as  
SERVQUAL, which emphasize reliability, responsiveness, assurance, empathy, and tangibles as determinants  
of customer satisfaction (Parasuraman et al., 1988). Numerous studies have affirmed the relevance of these  
dimensions in the African telecom landscape, particularly the centrality of network reliability and customer  
support in shaping satisfaction and loyalty. However, despite considerable investment and regulatory  
interventions, Ghanaian mobile users continue to report persistent service quality concerns, including network  
instability, high data costs, inconsistent coverage, and limited responsiveness to complaints. These challenges  
contribute to fluctuating customer satisfaction levels and increasing public pressure on service providers and  
regulators to address systemic QoSD issues.  
Although the link between service quality and customer satisfaction in telecommunications is well documented,  
an important theoretical and empirical gap remains. Prior research has focused predominantly on cognitive  
evaluations which involved how customers rate service quality dimensions while largely overlooking affective  
sentiment, i.e., how customers feel when describing their service experiences in their own words. This omission  
is notable because service quality failures often trigger emotional responses (frustration, dissatisfaction,  
disappointment) that may not be fully captured through structured Likert-scale ratings. Emerging research in  
customer analytics indicates that customers may express negative emotions even when their numeric evaluations  
appear neutral or moderately positive, a phenomenon known as sentiment divergence.  
Ghana’s telecommunication sector offers an ideal context for advancing this line of inquiry. Mobile customers  
frequently articulate dissatisfaction on social media, in call-center interactions, and in informal conversations,  
yet survey ratings often reveal moderate or stable satisfaction scores. This mismatch suggests that traditional  
QoSD assessment models may underestimate or misinterpret the emotional dimensions of customer experience.  
www.ijltemas.in  
Page 683  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Understanding this divergence is essential for improving service delivery, designing responsive interventions,  
and strengthening the relationship between service providers and customers.  
This study contributes to filling this gap by integrating dual sentiment analysis by combining expressed  
sentiment extracted from open-ended text responses with experienced sentiment measured through a composite  
Service Sentiment Index (SSI) derived from SERVQUAL dimensions. This dual approach enables a richer  
understanding of customer perceptions by capturing both the cognitive and affective elements of QoSD  
evaluations. Moreover, the study examines whether sentiment functions as a mediating mechanism through  
which reliability influences overall satisfaction, thereby extending existing service quality frameworks with a  
theoretically grounded affective pathway.  
Using a nationwide dataset of 536 mobile users, this study provides empirical evidence on the interplay between  
service quality dimensions, sentiment expression, demographic characteristics, and overall satisfaction in the  
Ghanaian telecommunication sector. By adopting an integrated quantitative–textual analytical strategy, the study  
offers both theoretical insight and practical recommendations for service providers and regulators. The  
introduction of the dual-sentiment framework also responds directly to calls for more advanced methodological  
approaches in QoS research, particularly within African and emerging market contexts where service  
experiences can be emotionally charged.  
Overall, this study enhances current understanding of customer sentiment in telecommunications by revealing  
how cognitive evaluations and emotional expressions interact to shape satisfaction and recommendation  
intentions. The findings underline the need for more holistic QoSD assessments that consider not only what  
customers say in ratings but also what they articulate in their own words.  
LITERATURE REVIEW  
Quality of Service Delivery in Telecommunications  
Quality of Service Delivery (QoSD) is a central construct in telecommunications research, reflecting the extent  
to which network operators deliver reliable, responsive, and satisfactory services to customers. As mobile  
communication has become the primary medium for social and economic activity in developing economies,  
QoSD has emerged as a critical differentiator of customer satisfaction and competitive advantage (Ladhari,  
2009). In regions such as sub-Saharan Africa, the rapid expansion of mobile networks has amplified expectations  
for stable connectivity, transparent pricing, and efficient customer care, placing pressure on operators to improve  
service standards.  
The SERVQUAL framework (Parasuraman, Zeithaml & Berry, 1988) remains the dominant theoretical model  
for evaluating service quality across industries, including telecommunications. The framework identifies five  
dimensions namely reliability, responsiveness, assurance, empathy, and tangibles as key drivers of service  
quality perceptions. Research in telecommunications contexts consistently highlights reliability as the most  
influential dimension, given the centrality of network stability, data performance, and call quality to customer  
experience (Agyapong, 2011; Kyei & Bayoh, 2017). Responsiveness and assurance also play substantial roles  
as customers depend on operators to provide timely support and inspire confidence in technical and security  
issues. Tangibles, though part of the original SERVQUAL model, have shown declining relevance in digital-  
first service environments where physical facilities and personnel visibility are minimal.  
Although the SERVQUAL model is widely adopted, scholars increasingly recognize its limitations in capturing  
contemporary customer experiencesparticularly emotional, digital, and contextual nuances. Studies have called  
for adaptations of SERVQUAL to incorporate digital service features such as mobile app responsiveness, data  
performance, and perceived fairness of pricing, especially in emerging markets (Hult et al., 2020). These  
concerns are particularly salient in countries like Ghana, where network congestion, fluctuating internet speeds,  
and customer service gaps frequently shape user perceptions.  
www.ijltemas.in  
Page 684  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
QoSD in the African and Ghanaian Telecommunications Context  
The African telecommunications sector has undergone transformative growth, driven by liberalization, increased  
mobile penetration, and the rise of mobile money services (Aker & Mbiti, 2010). Despite these advancements,  
service quality challenges persist, including inconsistent coverage, infrastructure limitations, and high data costs.  
These issues often lead to user dissatisfaction and contribute to consumer advocacy and complaints on social  
and digital platforms.  
In Ghana, QoSD remains a central regulatory priority. Reports from the regulatory body (NCA) frequently  
identify gaps in network reliability, data performance, and customer support. Empirical studies on the Ghanaian  
market such as a survey of MTN Ghana subscribers highlight that perceived service quality and relationship  
quality significantly influence customer satisfaction, underscoring the importance of service delivery and trust  
in shaping user evaluations (Oduro, Boachie-Mensah & Agyapong, 2018). Yet, these studies rely predominantly  
on structured surveys and traditional service-quality frameworks, offering limited insight into customers’  
affective responses or how emotional sentiment intersects with cognitive evaluations.  
The literature has not sufficiently examined how Ghanaian telecom users express their dissatisfaction or  
satisfaction, especially through open-ended comments that may reflect frustration, disappointment, or anger  
more vividly than Likert ratings. This omission creates a gap: existing models describe how customers rate QoS,  
but not how they emotionally articulate QoS experiences.  
Sentiment in Service Quality Research  
Sentiment, broadly defined as the emotional tone underlying customer expressions, is increasingly recognized  
as central to understanding service experiences. Research in marketing and consumer behavior shows that  
emotional responses such as anger, frustration, satisfaction, delight play a direct role in shaping loyalty, trust,  
and customer advocacy (Homburg et al., 2015). Advances in natural language processing (NLP) have enabled  
researchers to extract sentiment from unstructured customer feedback, social media, and product reviews,  
offering richer insight into authentic customer emotions (Cambria et al., 2013).  
However, sentiment has not been adequately integrated into telecom QoSD research, particularly in African  
contexts. Most existing studies treat sentiment as synonymous with satisfaction, relying solely on numerical  
ratings. This conflation ignores the well documented distinction between cognitive evaluations (structured  
ratings) and affective expression (open-ended emotional statements). Research increasingly suggests that  
customers may verbalize dissatisfaction even when providing moderate numeric ratings, a phenomenon known  
as sentiment divergence.  
Scholars such as De Keyser et al. (2020) argue for hybrid models that combine attitudinal scales with emotional  
analysis to fully capture modern service experiences. This aligns with the emerging need to integrate sentiment  
into QoSD frameworks, especially in markets where service shortcomings evoke strong emotional reactions.  
Dual Sentiment: Integrating Cognitive and Affective Evaluations  
The concept of dual sentiment which involves the coexistence of experienced sentiment measured through  
structured scales and expressed sentiment captured from open text remains underexplored in telecom research.  
While sentiment analysis tools can classify open-ended comments into positive, neutral, and negative tones, few  
studies combine these classifications with traditional SERVQUAL-derived scores to provide a fuller picture of  
customer perceptions.  
The theoretical value of dual sentiment lies in its ability to reveal alignment or misalignment between what  
customers say and what they rate. Such divergence has significant implications:  
1. It may explain why customers churn despite reporting moderate satisfaction.  
2. It may indicate emotional dissatisfaction not captured by structured surveys.  
www.ijltemas.in  
Page 685  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
3. It may uncover latent service issues masked by general rating averages.  
Despite its potential, no known study in the Ghanaian telecom context has integrated text-based sentiment with  
SERVQUAL-derived indices in a unified framework.  
Sentiment as a Mediating Psychological Mechanism  
The literature on service quality highlights the mediating role of psychological constructs such as trust, perceived  
value, and satisfaction. However, sentiment, particularly affective sentiment has rarely been examined as a  
mediator. Theoretically, service reliability should influence customer sentiment, which then shapes overall  
satisfaction. When service is reliable, customers are more likely to express positive sentiment; when service  
fails, sentiment becomes negative, subsequently reducing satisfaction.  
Yet, this proposed mechanism has not been empirically examined in the Ghanaian telecom sector. Integrating  
sentiment as a mediator introduces an affective pathway into SERVQUAL, enriching the model and aligning it  
with contemporary emotional-cognition theories in customer experience research.  
Identified Gaps and Study Justification  
Based on the review, four major gaps remain in the literature:  
1. Lack of integration of emotional sentiment into telecom QoSD research, despite strong evidence that  
customers express dissatisfaction emotionally.  
2. Absence of dual-sentiment frameworks combining text-based expression and scale-based experience in  
African telecommunications studies.  
3. Limited understanding of sentiment divergence, is the mismatch between expressed dissatisfaction and  
moderate numeric ratings.  
4. Lack of empirical testing of sentiment as a mediator, leaving theoretical gaps in explaining how service  
quality dimensions influence satisfaction.  
5. This study addresses these gaps by:  
6. Introducing a dual-sentiment analytic framework integrating NLP sentiment scoring with a Service  
Sentiment Index.  
7. Testing sentiment as a mediator between reliability and satisfaction.  
8. Providing new empirical evidence on sentiment divergence in the Ghanaian telecom sector.  
9. Extending SERVQUAL by embedding emotional mechanisms into its traditionally cognitive structure.  
METHODOLOGY  
Research Design  
The study employed a cross-sectional quantitative research design supplemented with text-based sentiment  
analysis. This approach was chosen to capture both the cognitive evaluations of service quality through  
structured Likert-scale items and the affective expressions of customers through open-ended responses. The  
integration of numeric and textual data supports the dual-sentiment framework and enables modeling of  
sentiment as a mediating variable between service quality and satisfaction.  
www.ijltemas.in  
Page 686  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Population, Sampling and Sample Size  
The target population comprises mobile telecommunication users in Ghana who are 18 years or older. Ghana’s  
mobile market is characterized by high penetration and multiple competing operators, with users distributed  
across all regions and across urban, peri-urban, and rural areas. Given the absence of a national sampling frame  
of individual subscribers and the practical difficulty of drawing a true probability sample, a non-probability  
sampling strategy was employed. Questionnaires were distributed to subscribers who were accessible to the  
researchers in different settings and who consented to participate. This approach combines convenience and  
voluntary response sampling, a practice commonly adopted in exploratory QoSD and service-quality studies  
where the population is large and geographically dispersed.  
The minimum required sample size was estimated using Cochran’s formula for large populations at a 95%  
confidence level and 5% margin of error, assuming maximum variability (p = 0.5). This yielded a minimum of  
384 respondents. In practice, 536 valid responses were collected across the 16 regions of Ghana, thereby  
exceeding the minimum requirement and providing sufficient power for factor analysis, regression modelling,  
and mediation testing.  
Instrumentation and Measures  
Data were collected using a structured questionnaire comprising both closed-ended and open-ended items.  
Service Quality Dimensions  
The five SERVQUAL dimensions namely reliability, responsiveness, assurance, empathy, and tangibles were  
measured using multiple Likert-scale items (1 = very low to 5 = very high). These items have been validated in  
prior telecommunications research and reflect core aspects of QoSD.  
Customer Satisfaction and Recommendation  
Overall satisfaction and likelihood to recommend were measured using 5-point Likert items frequently used in  
telecom satisfaction research.  
Demographic Characteristics  
Age, gender, education, and geographic region were included to assess demographic effects on satisfaction and  
sentiment.  
Open-Ended Sentiment Question  
To capture affective sentiment, respondents were asked:  
“What specific improvements would you prioritize to enhance your satisfaction?”  
This item generated 536 textual responses, used to derive expressed sentiment.  
Data Collection Procedures  
The survey was administered through a hybrid approach consisting of:  
1. Online distribution (email, social media, messaging platforms)  
2. Offline, paper-based administration in selected communities  
3. Telephone-assisted surveys for rural respondents with limited digital access  
This ensured broader coverage and inclusion of otherwise underrepresented groups.  
www.ijltemas.in  
Page 687  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
All responses were anonymized to maintain confidentiality.  
Data Preparation and Cleaning  
Data cleaning involved:  
1. Verifying completeness of responses  
2. Removing invalid or blank entries in open-ended fields  
3. Standardizing text (lowercase, punctuation removal, stop-word removal)  
4. Recoding categorical variables for analysis  
Numeric items were screened for outliers and missing values; missing data accounted for <1% and were handled  
using listwise deletion.  
Dual-Sentiment Framework  
The dual-sentiment approach consisted of two complementary components: text-based sentiment analysis and  
psychometric sentiment measurement.  
Text-Based (Expressed) Sentiment Analysis  
In parallel with the SSI, expressed sentiment was derived from the open-ended responses using a lexicon-based  
sentiment analysis tool, the VADER (Valence Aware Dictionary and sEntiment Reasoner) algorithm. VADER  
is particularly appropriate for short, informal texts of the kind produced in surveys and social media and has  
been widely used in applied sentiment research.  
Each cleaned response was submitted to the sentiment engine, which returned four values: a positive score, a  
negative score, a neutral score, and a compound score ranging from −1 (maximally negative) to +1 (maximally  
positive). The compound score was used as the principal summary measure of expressed sentiment. Following  
the widely adopted conventions for VADER, compound scores greater than or equal to +0.05 were classified as  
positive, scores less than or equal to −0.05 as negative, and scores in between as neutral. This yielded three  
expressed sentiment classes namely positive, neutral, and negative for each respondent.  
Beyond numeric sentiment scores, the textual responses were also examined qualitatively to identify recurring  
themes. Responses were grouped by sentiment class, and within each class, thematic coding was used to organize  
comments into categories such as network reliability, data pricing, customer service, coverage gaps, and security  
or fraud concerns. This thematic layer complements the numeric sentiment measures and helps interpret what  
the positive or negative sentiment refers to in substantive terms.  
Steps:  
1. Preprocessing:  
All textual responses were cleaned and standardized.  
2. Polarity Scoring:  
3. Each response was processed through VADER to generate:  
1. Positive score  
2. Negative score  
www.ijltemas.in  
Page 688  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
3. Neutral score  
4. Compound score (overall sentiment)  
4. Sentiment Classification:  
5. Based on standard VADER thresholds:  
1. Compound ≥ 0.05 → Positive  
2. Compound ≤ −0.05 → Negative  
3. Otherwise → Neutral  
This resulted in three sentiment classes for each respondent.  
4. Thematic Coding:  
Sentiment-labeled responses were analyzed using inductive thematic analysis to identify commonly expressed  
issues (e.g., network reliability, pricing, coverage).  
Psychometric (Experienced) Sentiment: Service Sentiment Index (SSI)  
Experienced sentiment was operationalized through a composite index constructed from the SERVQUAL-based  
items. The underlying idea is that customers’ ratings on reliability, responsiveness, assurance, empathy, and  
tangibles collectively reflect an affective–cognitive evaluation of service; this evaluation can be summarized as  
a continuous sentiment construct.  
To avoid scale and variance distortions and to ensure that each dimension contributed comparably, all items  
associated with the five service-quality dimensions were first standardized using z-scores. The Service Sentiment  
Index (SSI) for respondent i was then calculated as the arithmetic mean of the standardized scores across the  
five dimensions.  
Experienced sentiment was derived from the SERVQUAL-based items.  
Steps:  
1. Reliability Assessment:  
Internal consistency was evaluated using Cronbach’s α. The sentiment-related items achieved α = 0.73.  
2. Standardization:  
All SERVQUAL items were transformed into z-scores to ensure equal weighting.  
3. Composite Index:  
The Service Sentiment Index (SSI) was computed as the mean of the standardized scores:  
+
+
+
+
ℎ
=
5
4. SSI Classification:  
1. Low sentiment: SSI ≤ −0.50  
www.ijltemas.in  
Page 689  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
2. Neutral sentiment: −0.49 to +0.49  
3. High sentiment: SSI ≥ +0.50  
This index captures the affective–cognitive experience of service quality.  
Dual Sentiment Integration  
The two sentiment measures namely experienced sentiment (SSI) and expressed sentiment (VADER-based  
classification) were then integrated into a combined framework. For each respondent, the continuous SSI value  
and its three-level categorical band were paired with the expressed sentiment class. This produced a 3 × 3 dual-  
sentiment matrix with nine possible cells, representing combinations such as “negative expressed sentiment but  
high SSI” or “positive expressed sentiment and high SSI”.  
This matrix is used descriptively to examine the degree of alignment or divergence between how customers rate  
their service experiences and how they verbalize their feelings about those experiences. It allows the study to  
identify groups of users who, for example, report moderate or high service evaluations but still express negative  
emotions about particular aspects of service, an important phenomenon for both theory and practice.  
Factor Analysis  
Exploratory factor analysis (EFA) was conducted to verify the dimensional structure of service quality items.  
Suitability for EFA was confirmed by:  
1. KMO = 0.897 (excellent)  
2. Bartlett’s Test of Sphericity: p < .001  
Four factors with eigenvalues > 1 were extracted, explaining 63.48% of the total variance, confirming that  
service quality constructs align with theoretical expectations.  
Mediation Analysis  
To extend SERVQUAL and examine the emotional pathway between service quality and satisfaction, a  
mediation model was tested:  
1. Independent variable (X): Reliability  
2. Mediator (M): Service Sentiment Index (SSI)  
3. Dependent variable (Y): Customer satisfaction  
Hayes’ PROCESS Macro (Model 4) was applied with 5,000 bootstrap samples and 95% confidence intervals.  
Significance was inferred when bootstrap confidence intervals excluded zero.  
Ethical Considerations  
Participation was voluntary, with informed consent obtained from all respondents. Data were anonymized, and  
no identifiable information was collected. The study adhered to ethical standards for human-subject research.  
RESULTS  
Demographic Profile of Respondents  
www.ijltemas.in  
Page 690  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
A total of 536 valid responses were analyzed. Respondents were distributed across all 16 regions of Ghana, with  
a near-balanced representation across gender and education levels. The majority of respondents (63%) were  
between 18 and 35 years old, reflecting the demographic composition of active mobile data users in Ghana.  
MTN accounted for the largest share of subscribers (73%), followed by Vodafone (17%) and AirtelTigo (9%).  
Approximately 61% of users reported having been with their current provider for more than three years. These  
distributions indicate a sample that reflects the broader structure of the Ghanaian telecommunication market and  
is sufficiently heterogeneous for analytical purposes.  
Preliminary Diagnostics and Reliability  
Preliminary diagnostics confirmed the quality and suitability of the data for advanced analysis. Missing values  
were negligible (<1%), and no severe outliers were detected. The reliability of multi-item constructs was  
assessed using Cronbach’s alpha. Reliability (α = 0.81), responsiveness (α = 0.77), assurance (α = 0.76), empathy  
(α = 0.72), and tangibles (α = 0.69) all fell within acceptable bounds. The Service Sentiment Index (SSI),  
constructed from the standardized SERVQUAL items, also showed acceptable internal consistency (α = 0.73).  
These results provide confidence that the underlying constructs are measured consistently.  
Factor Structure of Service Quality Dimensions  
Exploratory factor analysis using principal axis factoring and Promax rotation supported the dimensionality of  
the adapted SERVQUAL items. The Kaiser–Meyer–Olkin statistic (0.897) and a significant Bartlett’s test (p <  
.001) indicated sampling adequacy and suitability for factor analysis. Four factors with eigenvalues greater than  
one emerged, jointly explaining 63.48% of the total variance.  
The strongest factor comprised reliability, empathy, and assurance items, underscoring the Ghanaian context in  
which network performance and interpersonal service characteristics are strongly interlinked in shaping  
perceived service quality. Responsiveness items formed a distinct factor, while tangibles remained the weakest  
and most fragmented dimension, confirming its reduced relevance in digitally mediated service contexts. These  
results are consistent with contemporary findings that emphasize the dominance of performance-related and  
interpersonal dimensions in mobile QoSD evaluations.  
Descriptive Statistics for Key Variables  
Table 1 summarizes the descriptive statistics of the main variables.  
Experienced sentiment (SSI) was normally distributed around zero, as expected from a standardized index, with  
moderate spread. Satisfaction remained moderate (mean = 3.14), suggesting that customers tolerate service  
quality issues but do not strongly endorse their providers.  
Table 1.  
Descriptive Statistics for Service Quality, Sentiment, and Satisfaction  
Variable  
Mean SD Min Max  
Reliability  
Responsiveness  
Assurance  
Empathy  
3.06  
2.84  
3.27  
2.88  
2.99  
0.98 1  
0.96 1  
0.91 1  
1.01 1  
0.88 1  
5
5
5
5
5
Tangibles  
www.ijltemas.in  
Page 691  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Satisfaction  
SSI (Standardized Composite) 0.00  
Recommendation 3.21  
3.14  
1.12 1  
0.82 −1.85 +1.74  
1.09 1  
5
5
Note. Descriptive statistics are based on cleaned responses from the SERVQUAL questionnaire and computed  
SSI scores.  
Expressed Sentiment: Text-Based Polarity Analysis  
The open-ended responses were classified into three categories using VADER’s compound score thresholds.  
Out of 536 responses:  
1. Negative sentiment: 56.5% (n = 303)  
2. Neutral sentiment: 26.3% (n = 141)  
3. Positive sentiment: 17.2% (n = 92)  
The dominance of negative sentiment indicates a high level of emotional dissatisfaction, even among  
respondents whose quantitative ratings fell within the moderate range. This supports the proposition that  
emotional expression is more critical and more volatile than structured cognitive evaluation.  
A thematic analysis of textual responses revealed five dominant clusters of customer concerns:  
1. Network reliability failures: dropped calls, slow or inconsistent data.  
2. Perceived high cost of data packages.  
3. Delayed or ineffective responses to complaints.  
4. Concerns about unfair billing and bundle depletion.  
5. Geographical coverage gaps, especially outside major urban centers.  
These themes help contextualize the polarity scores and confirm that negative sentiment is primarily driven by  
reliability deficiencies and price-value considerations rather than interpersonal service issues.  
Experienced Sentiment: Service Sentiment Index (SSI)  
The SSI constructed as the mean of standardized scores across the five SERVQUAL dimensions ranged from  
−1.85 to +1.74. For interpretation and matrix construction, respondents were placed in three sentiment  
categories:  
1. Low experienced sentiment: 29.5% (SSI ≤ −0.50)  
2. Neutral experienced sentiment: 49.3% (−0.49 ≤ SSI ≤ 0.49)  
3. High experienced sentiment: 21.3% (SSI ≥ +0.50)  
The distribution indicates that while textual sentiment was predominantly negative, experienced sentiment, as  
measured by quantitative ratings, was more balanced. This divergence sets the stage for examining dual  
sentiment alignment.  
www.ijltemas.in  
Page 692  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Dual Sentiment Matrix: Alignment and Divergence  
To explore the relationship between expressed and experienced sentiment, the two sentiment measures were  
cross-tabulated into a 3 × 3 matrix.  
Table 2.  
Dual Sentiment Matrix (Expressed × Experienced)  
Low SSI  
Negative Sentiment 142 (26.5%) 138 (25.7%) 23 (4.3%)  
Neutral Sentiment 13 (2.4%) 91 (17.0%) 37 (6.9%)  
Positive Sentiment 3 (0.6%) 35 (6.5%) 54 (10.1%)  
Neutral SSI High SSI  
Note. Expressed sentiment was derived using VADER polarity from open-ended responses; SSI was computed  
from SERVQUAL dimension z-scores.  
The matrix reveals three critical insights:  
1. Sentiment divergence is substantial. More than half of respondents expressing negative sentiment did not  
have correspondingly low SSI scores.  
2. Emotional dissatisfaction outweighs cognitive dissatisfaction. Negative textual sentiment persists even when  
experiential ratings are moderate.  
3. Only a minority (10.1%) simultaneously expressed and experienced positive sentiment, reflecting a small  
pool of advocates.  
This divergence contributes a novel theoretical insight: customers may express strong negative sentiment in  
narrative form even when their structured evaluations suggest tolerance or neutrality.  
Correlation Analysis  
Pearson correlation coefficients showed that reliability had the strongest association with satisfaction (r = .54, p  
< .001), followed closely by the Service Sentiment Index (SSI) (r = .51, p < .001). Responsiveness (r = .48, p <  
.001), assurance (r = .45, p < .001), and empathy (r = .43, p < .001) also demonstrated significant positive  
relationships with satisfaction. Tangibles showed the weakest correlation (r = .26, p < .001), confirming its  
relatively limited role in shaping satisfaction within digital-service environments. The strong positive correlation  
between SSI and satisfaction further supports its role as a mediating emotional mechanism in the evaluation  
process.  
Table 3  
Correlation Matrix for SERVQUAL Dimensions, SSI, and Satisfaction  
Variable  
1
2
3
4
5
6
7
1. Reliability  
1
2. Responsiveness .64 1  
3. Assurance  
.58 .55 1  
www.ijltemas.in  
Page 693  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
.52 .49 .57 1  
4. Empathy  
5. Tangibles  
6. SSI  
.33 .29 .30 .28 1  
.46 .41 .38 .39 .19 1  
.54 .48 .45 .43 .26 .51 1  
7. Satisfaction  
Note. p < .01 (two-tailed).  
All correlations align with expected SERVQUAL patterns and support regression assumptions.  
Regression Results  
A multiple regression analysis was conducted to examine the extent to which service quality dimensions and the  
Service Sentiment Index (SSI) predict overall satisfaction. This is shown inTable 4. The full model, which  
included all five SERVQUAL dimensions together with SSI, explained 57% of the variance in satisfaction (R²  
= .57, Adjusted R² = .56), indicating a strong model fit, F(6, 529) = 118.4, p < .001.  
Reliability emerged as the strongest predictor of satisfaction (β = .39, p < .001), reaffirming its central role in  
shaping customer evaluations in the telecom context. SSI also demonstrated a strong and statistically significant  
contribution (β = .32, p < .001), highlighting the importance of emotional sentiment in driving satisfaction  
beyond cognitive service quality assessments.  
Among the classical SERVQUAL dimensions, responsiveness had a smaller but significant effect (β = .12, p =  
.001), while assurance (β = .06, p = .061) and tangibles (β = .05, p = .079) did not reach conventional significance  
levels. Empathy showed a modest but significant relationship with satisfaction (β = .10, p = .004).  
Overall, the results demonstrate that both functional service performance (especially reliability) and experienced  
sentiment (SSI) jointly influence customer satisfaction, with reliability and sentiment emerging as the dominant  
predictors.  
These results reaffirm the primacy of reliability in shaping satisfaction in the telecom environment.  
Table 4  
Regression Analysis Predicting Satisfaction  
Predictor  
Constant  
B
SE  
0.91 0.18 -  
0.41 0.05 .39 8.12 < .001  
β
t
p
5.06 < .001  
Reliability  
Responsiveness 0.13 0.04 .12 3.25 .001  
Assurance  
Empathy  
Tangibles  
SSI  
0.07 0.04 .06 1.88 .061  
0.11 0.04 .10 2.86 .004  
0.05 0.03 .05 1.76 .079  
0.38 0.06 .32 6.57 < .001  
Note. Reliability is the strongest predictor; SSI is also a significant independent contributor.  
www.ijltemas.in  
Page 694  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Mediation Analysis: Reliability → SSI → Satisfaction  
A mediation analysis was conducted using Hayes’ PROCESS Macro (Model 4) with 5,000 bootstrap samples to  
examine whether experienced sentiment (SSI) serves as a mediating mechanism through which reliability  
influences overall satisfaction. The analysis revealed a significant effect of reliability on SSI (β = 0.566, SE =  
0.041, p < .001), indicating that higher perceptions of reliability were strongly associated with more positive  
experienced sentiment. In turn, SSI significantly predicted satisfaction (β = 0.381, SE = 0.060, p < .001), showing  
that customers’ emotional interpretations of service experiences play a substantive role in shaping their overall  
evaluations.  
When SSI was included in the model, the direct effect of reliability on satisfaction remained significant (β =  
0.411, SE = 0.052, p < .001), confirming that reliability exerts both a cognitive and affective influence on  
satisfaction. The indirect effect through SSI was statistically significant (β = 0.216), and the bootstrapped 95%  
confidence interval [0.142, 0.304] did not include zero, providing strong evidence of mediation. The total effect  
of reliability on satisfaction was substantial (β = 0.627, SE = 0.044, p < .001), indicating that reliability shapes  
satisfaction both directly and by influencing the emotional sentiment customers form during service encounters.  
Overall, these results support a partial mediation model, demonstrating that while reliability remains the  
strongest determinant of satisfaction, part of its influence is transmitted through customers’ experienced  
sentiment. This finding highlights the dual cognitive–affective nature of service evaluation, showing that  
reliability affects not only how customers think about service performance but also how they feel about it, both  
of which jointly shape their final satisfaction judgment.  
Table 5  
Mediation Analysis Summary (Reliability → SSI → Satisfaction)  
Path  
Effect  
0.566  
0.381  
SE  
t / z  
p
Path a: Reliability → SSI  
Path b: SSI → Satisfaction  
0.041 13.80 < .001  
0.060 6.35 < .001  
0.052 7.90 < .001  
0.036 5.94 < .001  
0.044 14.25 < .001  
Direct Effect (c’): Reliability → Satisfaction 0.411  
Indirect Effect (a × b)  
Total Effect  
0.216  
0.627  
Bootstrapped 95% CI  
0.142 – 0.304 -  
-
-
Notes.Partial mediation is confirmed. Reliability influences satisfaction both directly and through emotional  
sentiment (SSI).  
SUMMARY OF KEY FINDINGS  
1. Reliability is the dominant service-quality predictor of satisfaction.  
2. Expressed sentiment is overwhelmingly negative, but experienced sentiment is more moderate.  
3. Dual sentiment analysis reveals substantial divergence between emotional and cognitive evaluations.  
4. Sentiment (SSI) significantly mediates the reliability–satisfaction relationship.  
5. Tangibles are not significant, reinforcing their reduced relevance in a digital-first telecom context.  
www.ijltemas.in  
Page 695  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
DISCUSSION  
The objective of this study was to examine the complex relationships among service quality perceptions,  
customer sentiment, and satisfaction in Ghana’s mobile telecommunications sector by integrating classical  
SERVQUAL constructs with both experienced and expressed sentiment measures. The findings yield new  
theoretical and empirical insights that advance existing QoS literature and demonstrate the value of blending  
cognitive and affective evaluation frameworks.  
The first major finding confirms that reliability is the dominant driver of customer satisfaction, consistent with  
decades of service quality research (Agyapong, 2011; Kim, Park, & Jeong, 2004; Ranaweera & Prabhu, 2003).  
In the Ghanaian context, this dominance is further amplified by the nation’s widespread dependence on mobile  
data for daily economic transactions, education, financial inclusion, communication, and entertainment.  
Research on African and emerging markets demonstrates that mobile communication systems increasingly  
operate as foundational socio-economic infrastructure, supporting essential services and everyday activities  
rather than discretionary consumption (Aker & Mbiti, 2010; Donner, 2015). Thus, even marginal disruptions in  
call quality or data stability translate into substantial user dissatisfaction. This study reinforces earlier  
observations that QoS shortcomings, particularly in network reliability, have disproportionately large impacts  
on African mobile users due to limited alternative service options and infrastructural constraints (Oduro,  
Boachie-Mensah, & Agyapong, 2018; Shava, 2021).  
However, this study contributes beyond confirming these relationships by addressing how reliability shapes  
satisfaction through emotional sentiment. The partial mediation effect of the Service Sentiment Index (SSI)  
indicates that customers do not evaluate telecom service purely through rational judgments; instead, reliability  
influences both cognitive assessments and affective experiences, which together drive satisfaction outcomes.  
This finding aligns with multi-dimensional models of customer experience (Lemon & Verhoef, 2016) and with  
the appraisal-emotion theories of consumer behaviour, which posit that service experiences elicit emotional  
responses that mediate behavioural outcomes (Oliver, 2014; Ladhari, 2009). The evidence that experienced  
sentiment significantly mediates the relationship between reliability and satisfaction suggests that any  
improvement in network quality will translate not only into better ratings but also into more positive emotional  
statesthereby enhancing overall satisfaction and potentially strengthening loyalty.  
A second major contribution lies in the uncovering of sentiment divergence, a mismatch between expressed  
sentiment (textual emotional tone) and experienced sentiment (numeric SERVQUAL-derived evaluations). This  
divergence represents a critical extension to classical QoS theory. While SERVQUAL offers robust constructs  
for measuring service performance, it has long been critiqued for insufficiently capturing emotional and  
experiential nuances (Seth, Deshmukh & Vrat, 2005; Buttle, 1996). The present findings support these critiques:  
more than half of respondents expressed negative emotions in their open-ended responses, despite relatively  
moderate quantitative ratings on reliability, responsiveness, and empathy. This pattern mirrors findings from  
emerging text-mining literature, which shows that customers often suppress emotional intensity in structured  
ratings but reveal richer affective expressions in narrative comments (Ordenes et al., 2014; McColl-Kennedy et  
al., 2019).  
The sentiment divergence observed in this study provides new empirical support for the argument that customer  
satisfaction is not a unidimensional construct. Rather, it comprises layered components: cognitive evaluations  
(ratings), affective expressions (sentiment), and behavioural inclinations (recommendations, complaints). In  
many cases, the affective layer may be more diagnostic of true dissatisfaction than structured ratings. For  
example, customers may tolerate unreliable service due to perceived lack of alternatives or contractual lock-ins  
(East et al., 2008; Dwivedi et al., 2021), yet still articulate frustration when given the freedom to express  
themselves. This is especially pronounced in Ghana, where mobile telecommunication services have near-  
universal penetration but vary in quality across regions, resulting in users who may remain with a provider  
despite persistent grievances.  
The thematic analysis of expressed sentiment provides additional context. Respondents frequently mentioned  
network unreliability, high data costs, delays in resolving complaints, and concerns about billing accuracy. These  
www.ijltemas.in  
Page 696  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
issues correspond with findings reported in Ghana’s regulatory quality of service assessments (NCA, 2022) and  
with empirical evidence from studies of customer satisfaction in the domestic mobile telecom industry (Oduro,  
Boachie-Mensah, & Agyapong, 2018). Comparable observations appear in other African markets where service  
quality challenges continue to influence user satisfaction and perceived value (Chinomona & Sandada, 2014;  
Shava, 2021).  
These emotional expressions reveal that psychological dissatisfaction often centres on structural and  
infrastructural challenges rather than staff behaviour or physical service elements. This is consistent with global  
shifts in service quality priorities within digital industries, where tangibles play a diminished role (Parasuraman,  
Zeithaml & Berry, 1988; Tam, 2012).  
The factor analytic results further contextualize the Ghanaian telecom experience. The convergence of reliability,  
assurance, and empathy into a dominant factor suggests that Ghanaian consumers conceptualize service quality  
primarily in terms of performance reliability and trustworthiness, rather than in segmented SERVQUAL  
dimensions. This pattern aligns with research showing that technology-mediated service environments often  
reshape or compress traditional SERVQUAL structures, as customers evaluate digital services more holistically  
(Gummerus, 2014; Blut, Chowdhry, Mittal, & Brock, 2015; Parasuraman, Zeithaml, & Malhotra, 2005). The  
weaker influence of tangibles in this study corroborates arguments that digital service experiences require  
updated quality models emphasizing system performance, responsiveness, interface quality, and emotional  
engagement rather than traditional physical cues (Zeithaml, Bitner, & Gremler, 2018).  
The introduction of the dual sentiment framework represents one of the study’s strongest methodological  
contributions. By integrating experienced sentiment (SSI) with expressed sentiment (text-based polarity), this  
study demonstrates how quantitative and qualitative data can produce a more holistic understanding of customer  
perceptions. Researchers increasingly advocate integrating NLP-based sentiment analysis with structured survey  
data to obtain a more comprehensive understanding of customer experience. Recent studies demonstrate that  
text analytics can reveal emotional nuances not captured by rating scales, and that combining qualitative and  
quantitative sentiment improves service evaluation accuracy (Ordenes et al., 2014; McColl-Kennedy et al., 2019;  
Tirunillai & Tellis, 2014). Despite this progress, very few studiesparticularly within African telecom markets  
have operationalized a dual-sentiment framework or examined how cognitive service evaluations and emotional  
expressions diverge in shaping customer satisfaction.  
The dual sentiment matrix developed in this study reveals patterns that classical QoS metrics cannot capture.  
The existence of respondents with high SSI (indicating positive cognitive evaluations) but negative expressed  
sentiment (indicating emotional dissatisfaction) is theoretically important. These customers may be at heightened  
risk of engaging in negative word-of-mouth or complaint behaviour, even when they maintain moderate  
satisfaction. Conversely, customers who express positive emotions but give moderate ratings may represent  
latent advocates who emotionally endorse the brand despite functional shortcomings. These distinctions offer  
real-world implications for telecom providers, suggesting that customer experience strategies must target not  
only functional service improvements but also emotional engagement.  
From a regulatory perspective, the findings underscore the need for more nuanced QoS monitoring frameworks  
that incorporate sentiment analytics. Traditional QoS metrics such as call-drop rates, data throughput, coverage  
offer quantitative benchmarks but do not capture emotional dissatisfaction, which is often a precursor to  
consumer agitation and public complaints. The adoption of sentiment-based indicators could strengthen early  
warning systems for emerging QoS crises and help regulators such as the NCA design more targeted  
interventions.  
In summary, this study deepens theoretical understanding by demonstrating that customer satisfaction in telecom  
services emerges from the integration of cognitive evaluations, affective interpretations, and emotional  
expressions. The dual sentiment framework provides a richer lens for understanding telecom customer  
experiences in emerging markets. The results place Ghana’s telecom sector within broader global debates on  
service experience, contribute to evolving theories of sentiment in service evaluation, and offer actionable  
insights for providers seeking to enhance satisfaction in a competitive but infrastructurally complex market.  
www.ijltemas.in  
Page 697  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
CONCLUSION  
The purpose of this study was to examine how service quality perceptions and customer sentiment jointly shape  
customer satisfaction within the Ghanaian mobile telecommunications sector. By integrating the SERVQUAL  
model with both experienced sentiment (measured through the Service Sentiment Index) and expressed  
sentiment (derived from open-ended responses), the study offers a comprehensive and multidimensional  
understanding of Quality of Service Delivery (QoSD). The findings demonstrate that customer evaluations in  
this sector are not solely shaped by cognitive assessments of service attributes but are significantly influenced  
by affective sentiment and emotional expression.  
The study confirms that reliability remains the most influential determinant of satisfaction, reinforcing its  
centrality within the telecom service experience. In an environment where mobile communication is essential  
for economic and social activity, network stability, data performance, and consistent connectivity naturally  
dominate user expectations. However, a key contribution of this research is the discovery that the influence of  
reliability extends beyond cognitive evaluation: it also exerts a significant indirect effect through sentiment,  
supporting the argument that emotions form a fundamental part of the satisfaction process. The partial mediation  
of reliability on satisfaction by sentiment underscores the dual cognitive–affective pathways through which  
service quality affects user outcomes.  
A major theoretical advancement introduced by this study is the concept of sentiment divergence, the mismatch  
between customers’ structured ratings and their emotional expressions. Despite moderate SERVQUAL ratings  
for many respondents, the majority expressed negative sentiments in their open-ended responses. This  
divergence reveals that structured Likert scales may underestimate the level of emotional dissatisfaction and that  
text-based sentiment analysis provides a more sensitive indicator of service pain points. By operationalizing the  
dual-sentiment framework, this study demonstrates that cognitive and affective evaluations may align, partially  
diverge, or strongly contradict each other, with important implications for customer experience management.  
The thematic patterns of expressed sentiment also highlight structural weaknesses that continue to undermine  
user perceptions in Ghana’s telecom sector, including network unreliability, high data costs, inefficient  
complaint handling, and billing concerns. These issues mirror findings from broader African studies and  
emphasize the need for sustained infrastructural investment and enhanced customer service mechanisms. The  
limited relevance of tangibles and the convergence of reliability, assurance, and empathy into a core factor  
further illustrate that Ghanaian consumers now judge service quality primarily on functional performance and  
trustworthiness rather than physical service cues.  
Methodologically, this study demonstrates the value of integrating numeric and textual analytics within QoS  
research. The combination of SERVQUAL measures, sentiment indexing, and natural language sentiment  
extraction provides a richer empirical basis for understanding customer experiences than traditional methods  
alone. The dual sentiment matrix developed herein represents a novel analytical tool capable of identifying  
nuanced customer groupssuch as those who are cognitively tolerant but emotionally dissatisfied, thereby offering  
actionable insights for both researchers and practitioners.  
Overall, this study contributes to theory by expanding service quality models to incorporate emotional sentiment  
as both an evaluative dimension and a mediating mechanism. Empirically, it provides evidence from a  
developing-country telecom market where emotional dissatisfaction coexists with moderate cognitive  
evaluations. Practically, the findings highlight the need for telecom operators and regulators to monitor both  
functional QoS indicators and customer sentiment patterns to improve service outcomes and strengthen user  
trust.  
The dual-sentiment framework offers a foundation for future studies seeking to integrate emotional analytics  
into service evaluation. As telecommunications become increasingly intertwined with digital life and economic  
participation, understanding how users both perceive and feel about service quality will become essential to  
enhancing satisfaction, reducing churn, and improving industry-wide performance.  
www.ijltemas.in  
Page 698  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Implications  
The findings of this study yield several important implications for theory, managerial practice, and  
telecommunications regulation. By incorporating both SERVQUAL-based evaluations and dual sentiment  
measures, the study reveals new pathways through which customers interpret and react to service experiences.  
Theoretical Implications  
First, the study contributes to service quality literature by demonstrating that traditional SERVQUAL constructs  
alone cannot fully explain customer satisfaction in telecom environments. The discovery that experienced  
sentiment (SSI) partially mediates the impact of reliability on satisfaction highlights the need to integrate  
affective components into established models such as SERVQUAL, SERVPERF, and customer value  
frameworks. This advances cognitive–affective integration theories (Oliver, 2014; Lemon & Verhoef, 2016) and  
supports calls for expanded multidimensional models in digital service contexts.  
Second, the identification of sentiment divergence where customers express strong negative emotions despite  
moderate service ratings extends theories of customer dissatisfaction and behavioural intention. This suggests  
that future models must treat emotional expression as a distinct construct rather than assuming alignment with  
cognitive evaluation. The dual-sentiment matrix pioneered in this study provides a framework for capturing such  
mismatches and can be applied in other service industries where customers may suppress emotional  
dissatisfaction in structured responses.  
Third, the study strengthens emerging literature on sentiment analytics by showing the value of combining text-  
derived sentiment with psychometric sentiment indicators. This methodological contribution suggests that  
customer experience research can benefit from hybrid data strategies that use both quantitative and qualitative  
insights to more accurately capture the emotional dimension of service evaluation.  
Managerial and Industry Implications  
For telecom operators in Ghana, the results underscore the centrality of network reliability in influencing both  
affective sentiment and overall satisfaction. Investments in infrastructure particularly in data stability, network  
congestion management, and wider coverageare likely to yield the greatest satisfaction improvements. Operators  
should prioritise predictive network monitoring, real-time quality-of-service dashboards, and proactive  
communication during outages to mitigate negative emotional reactions.  
The findings also highlight the value of integrating sentiment analytics into customer feedback systems.  
Traditional surveys often fail to reveal underlying emotional dissatisfaction. Telecom managers should therefore  
use sentiment-based feedback from social media, call-centre logs, and open-ended survey responses to detect  
emerging issues earlier and tailor interventions more precisely. Customers whose expressed sentiment is  
negative but whose cognitive evaluations are moderate may require targeted communication strategies to prevent  
negative word-of-mouth or churn.  
Furthermore, the limited influence of tangibles suggests that operators should redirect resources from physical  
service touchpoints to digital and experiential improvements. Enhancing mobile app usability, transparent  
pricing communication, digital self-service tools, and rapid complaint resolution mechanisms will better meet  
user expectations.  
Policy and Regulatory Implications  
The results offer important insights for telecommunications regulators such as the National Communications  
Authority (NCA). Traditional QoS frameworks that focus only on call-drop rates, data throughput, and coverage  
metrics may not sufficiently reflect user experiences. The emotional dissatisfaction detected through sentiment  
analysis indicates the need for regulatory QoS monitoring frameworks that incorporate both performance  
indicators and customer sentiment indicators.  
www.ijltemas.in  
Page 699  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
Regulators could incorporate sentiment analytics into consumer complaint management systems to identify  
geographic or provider-specific dissatisfaction clusters. Integrating sentiment indicators into periodic QoS  
evaluations would enable earlier detection of dissatisfaction trends and more proactive interventions.  
Additionally, the evidence that customers express emotional frustrations tied to billing and complaint resolution  
suggests that regulatory oversight should extend beyond technical metrics to include customer service processes,  
billing transparency, and fairness in data depletion algorithms.  
LIMITATIONS AND FUTURE RESEARCH  
Despite its contributions, this study is subject to several limitations that suggest important avenues for future  
research.  
First, the study relies on convenience sampling, which may limit the generalizability of the results to the entire  
population of Ghanaian mobile telecom users. While the sample size is robust and geographically diverse, future  
studies may adopt stratified random sampling or operator-based sampling frames to enhance representativeness.  
Second, sentiment classification in this study was based on VADER, a lexicon-based model designed primarily  
for general English text. Although VADER performs well on short informal messages, it may not fully capture  
linguistic nuances, sarcasm, multilingual expressions, or culturally embedded terms common in Ghana. Future  
research may benefit from supervised machine learning models trained on Ghana-specific telecom sentiment  
corpora or transformer-based models such as BERT or RoBERTa fine-tuned for local text.  
Third, the cross-sectional design precludes causal inferences beyond mediation modelling. Longitudinal or  
panel-based designs could examine how sentiment and satisfaction evolve over time, particularly during network  
disruptions, promotional campaigns, or regulatory interventions. Time-series sentiment monitoring would yield  
deeper insights into the dynamic emotional states of customers.  
Fourth, the study focuses on SERVQUAL dimensions and sentiment but does not explicitly incorporate  
behavioural outcomes such as actual churn, switching intention, or digital engagement. Future studies could link  
sentiment divergence to behavioural metrics or use predictive modelling to determine whether emotionally  
negative customers with moderate satisfaction are more likely to churn.  
Fifth, while the dual sentiment matrix reveals important divergence patterns, it does not explore potential  
demographic moderators or contextual variables (e.g., urban vs. rural disparities, prepaid vs. postpaid  
differences). Multigroup analysis or structural equation modelling (SEM) could enrich the understanding of how  
sentiment and service quality interact across different user segments.  
Finally, the study is situated in Ghana’s telecommunications sector, which has unique infrastructural and  
competitive characteristics. Cross-country comparative studies across West Africa or other developing markets  
could determine whether the dual-sentiment patterns observed here are universal or context-specific.  
REFERENCES  
1. Agyapong, G. K. Q. (2011). The effect of service quality on customer satisfaction in the utility industry:  
Ghana’s mobile telecommunication industry. International Journal of Business and Management, 6(5),  
203–210. https://doi.org/10.5539/ijbm.v6n5p203  
2. Aker, J. C., & Mbiti, I. M. (2010). Mobile phones and economic development in Africa. Journal of  
Economic Perspectives, 24(3), 207–232. https://doi.org/10.1257/jep.24.3.207  
3. Buttle, F. (1996). SERVQUAL: Review, critique, research agenda. European Journal of Marketing,  
30(1), 8–32. https://doi.org/10.1108/03090569610105762  
4. Blut, M., Chowdhry, N., Mittal, V., & Brock, C. (2015). E-service quality: A meta-analytic review.  
Journal  
of  
Retailing,  
91(4),  
679–700.  
https://doi.org/10.1016/j.jretai.2015.05.004  
www.ijltemas.in  
Page 700  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
5. Chinomona, R., Masinge, G., & Sandada, M. (2014). The influence of e-service quality on customer  
perceived value, customer satisfaction and loyalty in South Africa. Mediterranean Journal of Social  
Sciences, 5(9), 331–341. https://doi.org/10.5901/mjss.2014.v5n9p331  
6. Donner, J. (2015). After access: Inclusion, development, and a more mobile internet. MIT Press.  
https://mitpress.mit.edu/9780262528948/  
7. Dwivedi, Y. K., Ismagilova, E., Hughes, D. L., Carlson, J., Filieri, R., Jacobson, J., Jain, V., Karjaluoto,  
H., Kefi, H., Krishen, A. S., Kumar, V., Rahman, M. M., Raman, R., Rauschnabel, P. A., Rowley, J., …  
Wang, Y. (2021). Setting the future of digital and social media marketing research: Perspectives and  
research propositions. International Journal of Information Management, 59, 102168.  
https://doi.org/10.1016/j.ijinfomgt.2020.102168  
8. East, R., Hammond, K., & Lomax, W. (2008). Measuring the impact of positive and negative word of  
mouth on brand purchase probability. International Journal of Research in Marketing, 25(3), 215–224.  
https://doi.org/10.1016/j.ijresmar.2008.04.001  
9. Gummerus, J. (2014). Value creation processes and value outcomes in marketing theory: Strangers or  
siblings? Marketing Theory, 13(1), 19–46. https://doi.org/10.1177/1470593112467267  
10. Kim, M. K., Park, M. C., & Jeong, D. H. (2004). The effects of customer satisfaction and switching  
barrier on customer loyalty in Korean mobile telecommunication services. Telecommunications Policy,  
28(2), 145–159. https://doi.org/10.1016/j.telpol.2003.12.003  
11. Oduro, E., Boachie-Mensah, F. O., & Agyapong, G. K. Q. (2018). Determinants of customer satisfaction  
in the telecommunication industry in Ghana: A study of MTN Ghana Ltd. International Journal of  
Marketing Studies, 10(3), 101–117. https://doi.org/10.5539/ijms.v10n3p101  
12. Ladhari, R. (2009). Service quality, emotional satisfaction and behavioural intentions: A study in the  
hotel industry. Managing Service Quality, 19(3), 308–331. https://doi.org/10.1108/09604520910955320  
13. Lai, T. L., Griffin, M., & Babin, B. J. (2007). How quality, value, trust, and satisfaction create loyalty in  
the U.S. wireless communications industry. Journal of Business Research, 60(4), 357–365.  
https://doi.org/10.1016/j.jbusres.2006.12.002  
14. Lemon, K. N., & Verhoef, P. C. (2016). Understanding customer experience throughout the customer  
journey. Journal of Marketing, 80(6), 69–96. https://doi.org/10.1509/jm.15.0420  
15. McColl-Kennedy, J. R., Zaki, M., Lemon, K. N., Urmetzer, F., Neely, A., & Begley, S. (2019). Gaining  
customer experience insights that matter. Journal of Service Research, 22(4), 409–423.  
https://doi.org/10.1177/1094670518812182  
16. National Communications Authority (NCA). (2022). Quality of Service (QoS) report: Mobile network  
performance in Ghana. https://nca.org.gh  
17. Oliver, R. L. (2014). Satisfaction: A behavioral perspective on the consumer (2nd ed.). Routledge.  
https://doi.org/10.4324/9781315700892  
18. Ordenes, F. V., Theodoulidis, B., Burton, J., Gruber, T., & Zaki, M. (2014). Analyzing customer  
experience feedback using text mining: A linguistics-based approach. Marketing Science, 33(6), 869–  
889. https://doi.org/10.1177/1094670514524  
19. Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1988). SERVQUAL: A multiple-item scale for  
measuring consumer perceptions of service quality. Journal of Retailing, 64(1), 12–40. (No DOI; classic  
article)  
20. Ranaweera, C., & Prabhu, J. (2003). The influence of satisfaction, trust and switching barriers on  
customer retention in a continuous purchasing setting. International Journal of Service Industry  
Management, 14(3), 283–307. https://doi.org/10.1108/09564230310489231  
21. Seth, N., Deshmukh, S. G., & Vrat, P. (2005). Service quality models: A review. International Journal of  
Quality & Reliability Management, 22(9), 913–949. https://doi.org/10.1108/02656710510625211  
22. Shava, H. (2021). Service quality and customer satisfaction experience among South African mobile  
telecommunications consumers. Eurasian Journal of Business and Management, 9(3), 217–232.  
https://doi.org/10.15604/ejbm.2021.09.03.004  
23. Tam, J. L. M. (2012). The moderating role of perceived risk in loyalty intentions: An investigation in a  
service  
context.  
Marketing  
Intelligence  
&
Planning,  
30(1),  
33–52.  
http://doi.org/10.1108/02634501211193903  
www.ijltemas.in  
Page 701  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XI, November 2025  
24. Tarus, D. K., & Rabach, N. (2013). Determinants of customer loyalty in Kenyan mobile  
telecommunication services. International Journal of Business and Social Science, 4(2), 82–90.  
https://ijbssnet.com/journals/Vol_4_No_2_February_2013/10.pdf  
25. Tirunillai, S., & Tellis, G. J. (2014). Mining marketing meaning from online chatter: Strategic brand  
analysis of big data using latent Dirichlet allocation. Journal of Marketing Research, 51(4), 463–479.  
https://doi.org/10.1509/jmr.12.0106  
26. Zeithaml, V. A., Bitner, M. J., & Gremler, D. D. (2018). Services marketing: Integrating customer focus  
across the firm (7th ed.). McGraw-Hill. (No DOI available)  
www.ijltemas.in  
Page 702