Factors affecting customer satisfaction with life insurance service quality: A case study in Hanoi

: This study aims to identify and evaluate factors that affect customer satisfaction in Hanoi with insurance services at a life insurance company. With 259 questionnaire survey samples, the research team used SPSS statistical software to test the research hypotheses. Research results show that 3/5 factors of service quality, including Assurance, Tangibles, and Responsiveness, positively impact customer satisfaction, while 2/5 factors, Reliability and Empathy, are not statistically significant enough to conclude the relationship affecting customer satisfaction. From the research results, the authors have proposed solutions to improve the quality of insurance services to increase customer satisfaction and promote the sustainable development of insurance businesses.


Introduction
With today's developments in quality of life, health consciousness has become more important to many people, expanding perspectives on life insurance.According to information from the National Institute for Finance (2024), Vietnam currently has about 12% of the population participating in life insurance, while life insurance participation rates in some developed Southeast Asian countries such as Malaysia and Singapore reached 75% and 80%, respectively (US Department of Treasury, 2017).The number of people participating in life insurance is low for many reasons, including the reasons stemming from the businesses providing life insurance services, such as Inaccurate advice, poor contract management, and limited financial capacity, leading to customer dissatisfaction and lack of trust in life insurance services.
Insurance service quality is essential to customer satisfaction, creating trust, and keeping customers loyal to businesses providing life insurance services.Therefore, research to improve insurance service quality and customer satisfaction is necessary.Many factors measure insurance service quality; which factors have an influence, and to what extent do these factors affect customer satisfaction?To answer that question, the authors researched factors affecting customer satisfaction with the quality of life insurance services.The research was conducted within Hanoi, and the surveyed subjects had been using life insurance services.From the research results, the authors have proposed solutions to improve the quality of insurance services to increase customer satisfaction and promote the sustainable development of insurance businesses.

Basis of theory, model, and research hypotheses 2.1. Basic theory 2.1.1. Customer satisfaction
Richard L. Oliver (2010, p.8) states, "Satisfaction is the consumer's fulfillment response.It is a judgment that a product or service feature, or the product of service itself, provided (or is providing) a pleasurable level of consumption-related fulfillment, including levels of under-or over-fulfillment..." Philip Kotler and Kevin Lane Keller (2006, p.144) define Satisfaction as "a person's feeling of pleasure or disappointment which resulted from comparing a product's perceived performance or outcome against his/ her expectations."Customer perceived value has been defined as "the difference between the prospective customer's evaluation of all the benefits and all the costs of an offering and the perceived alternatives" (Kotler and Keller, 2006, p.141) Customer satisfaction is their desire for a perceived difference between known experience and expectations.That is the customer's known experience when using a service and the results after the service is provided.(Parasuraman et al., 1988).
According to Hansemark and Albinsson (2004), "Customer satisfaction is a customer's overall attitude toward a service provider, or an emotional response to differences between what the customer anticipated before and what they perceive, with respect to the fulfillment of some need, goal or desire." Thus, customer satisfaction is reflected in the psychology and emotions of customers.Customers feel more about products and services after using them than before using them, receiving advice, or being introduced by relatives or acquaintances.In other words, there is a comparison between expectations about products and services and actual perceptions.Can be understood: Level of Satisfaction = Actual feeling -Expectation.That is the premise for forming an assessment of the level of Satisfaction with a product or service.

Service quality
According to Christian Gronroos (1984), service quality is evaluated in two aspects: technical quality (Technical Service Quality -TSQ) and skill quality (Functional Service Quality -FSQ).Technical service quality results from the interaction process between businesses and customers in which companies provide services and customers receive those services.Functional service quality represents the enterprise's service implementation process, reflecting how the service is provided.Parasuraman et al. (1988) affirmed that service quality is an overall long-term assessment; service quality is the gap between customers' perceptions and expectations when using the service.

Model of Factors Affecting Customer Satisfaction 2.1.3.1. Satisfaction research model by Parasuraman et al. (1988) (SERVQUAL)
The SERVQUAL model was developed by Parasuaman and his colleagues in 1988 to measure customer perceptions of the service quality of the business.Customers have used it through 05 criteria: reliability, responsiveness, assurance, empathy, and tangibles.Specifically: Reliability: evaluates the ability to provide services to ensure timeliness and accuracy of information.
Responsiveness: evaluates the ability to respond and satisfy customer requirements.
Assurance: Demonstrates customer trust through customers' perception of serviceability, professional knowledge, and good communication skills of employees.
Empathy: evaluates customer care and understanding.
Tangibles: evaluated through objective conditions such as facilities, costumes, staff attitudes Unlike the SERVQUAL model, at the scale of the SERVPERF model, service quality is only evaluated through customer perception (Service quality = Perception level).The SERVPERF scale includes 22 questions similar to the SERVQUAL model.However, in the SERVPERF model, the author omits the assessment of customer expectations.
Regarding general assessment, the SERVPERF model is considered more convenient, as it has a shorter questionnaire.However, this model needs to reflect the relationship between customer expectations and service quality.

American customer satisfaction index (ACSI) model
Author Fornell built the American Customer Satisfaction Index (ACIS) model, which was first published in 1996.The American Customer Satisfaction Index (ACIS) model measures results based on three criteria.: -Expectations: Evaluated through the customer's perception of the product compared to the product that the customer expected.
-Perceived quality: The customer's service quality assessment during or after service use.
-Perceived value: customer's assessment of the value received when using the customer's product; customer satisfaction depends on the perceived value of goods and services.
The above criteria have an impact on customer satisfaction with service quality, thereby affecting customer loyalty to the business.According to the definition, loyalty is expressed by the intention to continue buying and recommending products and services to others.The opposite of loyalty is complaints, which appear when the product does not meet the customer's expectations.

Research model and hypothesis 2.2.1 Research model
Based on a theoretical overview and research models on service quality affecting customer satisfaction, the research team built a research model with factors according to the SERVQUAL model.This model has been proven reliable in many studies from developed countries (USA, UK) or developing countries (India), so choosing this model is feasible and reasonable.In particular, this model is considered highly reliable and accurate for many days.
The proposed research model is as follows:

Hypotheses for the proposed research model include:
Hypotheses for the proposed research model include: Hypothesis H1: The higher the customer's perception of "Reliability," the higher the customer's "Satisfaction" with the quality of insurance services and vice versa.
Hypothesis H2: The higher the customer's perception of "Assurance," the higher the customer's "Satisfaction" with the quality of insurance services and vice versa.
Hypothesis H3: The higher the customer's perception of "Tangibles," the higher the customer's "satisfaction" with the quality of insurance services and vice versa.
Hypothesis H4: The higher the customer's perception of "Empathy," the higher the customer's "Satisfaction" with the quality of insurance services and vice versa.
Hypothesis H5: The higher the customer's perception of "Responsiveness," the higher the customer's "satisfaction" with the quality of insurance services and vice versa.(Source: Referenced and developed by the author) 3. Research method 3.1.Data collection method Comrey and Lee (1992) provided sample sizes with corresponding opinions: 100 = poor, 200 = fair, 300 = good, 500 = very good, 1000 or more = excellent.Some researchers do not give specific numbers but provide the ratio between the number of samples needed and the number of parameters that need to be estimated.For factor analysis, the sample size will depend on the number of variables included.

Observed variables and coding
Based on theory and an overview of research on factors affecting behavioral intention, the following factors (independent variables) were included in the model: " Reliability "; " Assurance "and "Empathy.""Emotion,"; "Responsiveness," and " Tangibles." The survey was built on a 5-point Likert scale with the following indicators: 1. Completely disagree EM-2024-6433 2. Disagree 3. Normal 4. Agree 5. Completely agree After developing the survey questionnaire, the research team conducted a random pilot survey of 15 customers in Hanoi.Preliminary survey results show that opinions agree with the factors included in the model.
The author used the purposive sampling method because a list of customers was available for the survey.The sample size was determined according to the rules of Comrey and Lee (1992) and also referred to the rules of Hoang & Chu (2005).With 27 parameters (observed variables) needing to conduct factor analysis, the minimum number of samples needed is 27 x 5 = 135 observed samples.The surveyed subjects are customers living in Hanoi who have purchased life insurance from a life insurance company in Hanoi city.From the perspective of collecting as many observation samples as possible to ensure the stability of the impact, based on the ability to collect samples, the research team decided the number of ballots to be distributed is n = 300.The ballots were sent to the subjects.Survey subjects by email, zalo, phone calls, and face-to-face meetings.The number of ballots received was 265.After screening, 259 eligible ballots were used by the research team as the database for analysis.

Analysis of research data
All survey forms will be processed using the analytical techniques of IBM SPSS 26.0 software.Descriptive statistics: Descriptive statistical methods aim to describe a data sample's characteristics, including variables' averages and frequencies.To supplement the quantitative analysis results and clarify the impact level of the factors, the research team performed descriptive statistics to evaluate them.This assessment is based on the distance and average value of the calculated factors.Check the reliability of the scale (Cronbach's Alpha): According to Nguyen Dinh Tho (2013), a scale is considered to have good reliability when the Cronbach's Alpha coefficient >= 0.6, but should not exceed 0. 95 to avoid duplicate measurements.Variables with adjusted total variable correlation coefficients below 0.3 will be eliminated (Nguyen Dinh Tho, 2013), quoted from Nunnally & Bernstein (1994).Factor Analysis EFA: EFA evaluation often uses the KMO (Kaiser-Meyer-Olkin) coefficient, which evaluates the appropriateness of factor analysis.If KMO is high (from 0.5 to 1), factor analysis is considered appropriate; Conversely, if KMO is lower than 0.5, factor analysis may not be suitable for the data.Variables with Factor Loading below 0.5 are often eliminated, and stopping criteria include an Eigenvalue more significant than one and a total variance extracted greater than 50% (Gerbing & Anderson, 1988).Correlation analysis: Pearson correlation measures the strength and direction of the linear relationship between pairs of continuous variables.Denoted r, the correlation coefficient typically ranges from -1 to +1.The coefficient sign indicates the relationship's direction, while its magnitude indicates the strength of the relationship.Linear regression analysis: Linear regression predicts the relationship between two variables by assuming a linear association.It searches for the optimal path to minimize the difference between expected and actual values.It can be extended to apply to multiple independent variables and logistic regression for binary classification problems.
If the F statistical value has Sig < 0.05, then the hypothesis H0 is rejected, indicating that the set of independent variables in the model can explain the variation of the dependent variable.The ANOVA test selects the optimal model: Sig ≤ 0.05: Model is suitable.
Tolerance or VIF can be used to detect multicollinearity, with a threshold of VIF > 10 or > 2 being a sign of multicollinearity.Evaluate model fit using Adjusted R Square instead of R Square.Finally, check the assumption of no correlation between residuals using Durbin -Watson or Scatter plot.Regression analysis: Linear regression predicts the relationship between two variables by assuming a linear association between them.It searches for the optimal path to minimize the difference between expected and actual values.It can be extended to apply to multiple independent variables and logistic regression for binary classification problems.
If the F statistical value has Sig < 0.05, then the hypothesis H0 is rejected, indicating that the set of independent variables in the model can explain the variation of the dependent variable.The ANOVA test selects the optimal model: Sig ≤ 0.05: Model is suitable.Barlett's test results show a correlation between the variables in the population (Sig = 0.000), and the KMO coefficient = 0.879 proves that factor analysis to group variables together is appropriate.All observed variables have standard Factor loading coefficients greater than 0.5.Value of total variance extracted = 72.59%> 50%: satisfactory; then it can be said that these five factors explain 72.59% of the variation in the data.

Correlations
The results of the Cronbach's Alpha test and EFA factor analysis show six factors with 25 observed variables in the research model.The Pearson correlation coefficient method is performed to evaluate the correlation between factors in the model.The results of correlation analysis from Table 4 show that no factor is eliminated because the Sig between each independent variable and the dependent variable is less than 0.05.Thus, all independent variables have a linear correlation with the dependent variable.

Regression analysis
The authors performed regression analysis to evaluate the impact of factors affecting customer satisfaction with insurance services.Table 6 shows that the Sig value of the F test = 0.000 < 0.05, so the regression model is meaningful and reflects.The regression analysis model between the dependent and independent variables is appropriate.
The correlation coefficient R is 0.758, reflecting that the variables have a relatively close and proportional relationship.
The coefficient R is 0.758, reflecting that the variable's relationship has a relatively close and proportional correlation.
The adjusted R 2 is 0.566 = 56.6%.Thus, the independent variables included in the regression run affect 56.6% of the change in the dependent variable.The remaining 43.4% is due to variables outside the model and random errors.
Durbin Watson coefficient = 1.882 (range 1 to 3) means that the model does not violate when using the regression method, and the model has no first-order serial correlation.Thus, the regression model satisfies the appropriate evaluation and testing conditions for drawing research conclusions.
The VIF coefficient ranges from 1.365 to 1.739 < 3, so no variables violate the hypothesis of multicollinearity; the model is statistically significant.
The testing significance level (Sig) of the intercept and regression coefficients of the factors Reliability and Empathy is greater than 0.05.Therefore, it can be seen that there is not enough statistical significance to conclude that the two factors, Reliability and Empathy, have an impact on customer satisfaction with the quality of life insurance services.Hypotheses H1 and H4 are rejected The testing significance level (Sig) of the intercept and regression coefficients of the factors Reliability and Empathy is greater than 0.05.Therefore, it can be seen that there is not enough statistical significance to imply that the two factors, Reliability and Empathy, have an impact on customer satisfaction with the quality of life insurance services The observed variables Assurance, Tangibles, Empathy, and Responsiveness impact the dependent variable because the t-test Sig value of each independent variable is less than 0.05; Hypotheses H2, H3, and H5 are accepted.

Conclusion and recommendations 5.1. Conclusions
From the results of regression analysis and assessment of model suitability, the impact relationship between factors in the model is expressed according to the following formula: HL = 0.364 x DB + 0.138 x HH + 0.385 x DU Responsiveness is the factor that has the most decisive influence on customer satisfaction, with a standardized regression coefficient of 0.385.This implies that when other factors remain unchanged if EM-2024-6438 responsiveness increases by 1 point, customer satisfaction with service responsiveness will also increase by 0.385 points.
Next is the factor of assurance, which also has a substantial impact on customer satisfaction, with a standardized regression coefficient of 0.364.This means that when other factors do not change if the Assurance increases by 1 unit, customer satisfaction with the level of insurance service assurance will also increase by 0.364 units.
Finally, there is the Tangibles factor.Although it is seen that the impact is the lowest among the three factors, it can also be seen that this factor contributes to customer satisfaction with a standardized coefficient of 0.138.That is, when other factors do not change if tangible means increase by 1 unit, customer satisfaction with tangible means of insurance services will also increase by 0.138 units.

Recommendations
Firstly, according to research results, Responsiveness is the factor that has the most decisive influence on customer satisfaction.To improve responsiveness, insurance businesses need to focus on improving the digital service experience, Improving products to reflect customer desires, Shortening and clarifying insurance contracts, Enhancing customer care and support, Applying insurance benefits settlement policy, and strictly controlling the response and compensation process.
Second, the Assurance factor also substantially impacts customer satisfaction, so businesses also need to focus on improving service quality in terms of assurance.Employees and consultants must participate in communication courses to increase their ability to negotiate and persuade customers.A 24/7 customer service center will help resolve problems quickly and effectively.Inspection officers need to be responsible, honest, enthusiastic, and experienced.In addition, insurance employees/agents must pay attention to professional behavior, politeness, neat dress, and customer respect.
Third, to improve service quality in tangibles, insurance businesses should refer to and carry out some activities: Upgrading offices and equipment, creating design highlights, creating unique experiences, and ensuring safety and hygiene.
Fourth, the appraisal and compensation work must be carried out accurately, transparently, and carefully, following the corporation's instructions.The goal is to avoid any errors that could lead to customer complaints.At the same time, improving and standardizing the compensation process is also essential to minimize case processing time.During the appraisal and compensation process, customer questions or complaints may arise.To ensure their Satisfaction and trust, relevant officials need to take responsibility for listening and resolving all issues promptly and effectively.This not only helps avoid inconvenience or discomfort to customers but also shows professionalism and respect on the part of the organization.
Fifth, resolving customer inquiries and complaints is also essential and indispensable.It is necessary to ensure that all issues are resolved satisfactorily and promptly and to advise and guide customers in resolving difficulties most effectively.

Table 4 . Factor analysis of EFA result
The EFA method uses Principal axis factoring and Varimax to determine factors affecting customer satisfaction.In the model, there are 27 observed variables.The first results showed that two variables (DB4 and DC1) did not meet the convergence condition, so they were eliminated.The second EFA with 25 observed variables gave the following results: