Factors affecting consumers’ repurchase intention toward skin care cosmetics: A cross - sectional study in Vietnam

Abstract

The purpose of this study is to investigate the factors affecting consumers' repurchase intentions of skincare cosmetics in Can Tho city. By examining consumers’ knowledge, behaviors, and factors impacting their intention of repeatedly buying skincare products, manufacturers and sales businesses are able to more effectively comprehend their clients' demands. Data were collected from 531 consumers residing in the area of Can Tho city, Vietnam, who had bought skincare products for the second time or more. The survey using the cross-sectional method and a set of questionnaires was based on the theory of planned behavior (TPB) and developed on the 5-point Likert scale. The factors impacting repurchase intentions for skincare merchandise involve attitude, reliability of signal quality, and retailer credibility, with attitude being the most powerful influence, followed by retailer credibility and reliability of signal quality respectively. These factors accounted for 27.5 % of behavior variance. The clarity element of signal quality reliability does not affect behavioral intention. Cosmetic companies can develop policies, improve the quality of their products in order to suit the customer needs, retain potential clients, and enhance their competitive capability.

1. Introduction

Throughout history, humans have been fascinated with the concept of beauty and its enhancement. This fascination has manifested in the development and consumption of cosmetics, multi-billion dollars global industry. Humanity's enduring pursuit of beauty has fueled a booming cosmetics industry, with the global market reaching a staggering 378 billion USD in 2022 and projected to surpass 661 billion USD by 2032 [1]. Following the global trend, the cosmetics business network in Vietnam, especially Can Tho city, is very vibrant. Consumers have a variety of purchasing options, including purchasing directly in-store or through various digital channels such as online platforms and media. In addition, the increasing income of Vietnamese people has significantly increased their ability to purchase cosmetics. For these reasons, the Vietnamese cosmetics market is a potential market with stable revenue growth over the years. Plus, the rising income of Vietnamese individuals has significantly increased their ability to purchase beauty products. For these reasons, the cosmetics market in Vietnam is a potential market with stable revenue growth over the years.

Cosmetics applied to the outermost layer of the skin, the epidermis, are specifically categorized as skincare products. Their effectiveness depends on the types of ingredients and the technology used to prepare them. Skincare products can be presented in different physical states: liquid (solutions or suspensions), solid (powders) or semisolid systems (gels and emulsions). Emulsions are the most frequently used and can be classified, depending on their consistency, as creams or lotions. Creams have a thicker or heavier consistency as compared with lotions. This difference is due to the higher water content of lotions. Skin care products represent a large category of consumer products. Recent decades have seen the emergence of cosmeceuticals, or products that have therapeutic effects capable of affecting the condition of the skin beyond the duration of application [2].

Understanding beauty and cosmetics consumption necessitates a multi-disciplinary approach, drawing from fields such as psychology, sociology, and marketing. A 2018 article explores the concept of the “beauty ideal,” a culturally constructed standard that creates a need to conform [3]. This ideal, often heavily influenced by media portrayals, exerts pressure on individuals to achieve that standard, leading to increased cosmetics consumption. Here's where a customer-centric approach becomes crucial. Brands need to understand not just the pressure to conform, but also the nuances of these beauty ideals across different cultures and demographics. Failing to prioritize customer needs can lead to missed opportunities. This multifaceted field offers a rich avenue for research, with profound implications for understanding human behavior, industry trends, and the cultural significance of beauty. By delving into the interplay of psychology, social factors, and marketing forces, we can illuminate the ever-evolving world of beauty and cosmetics consumption.

By prioritizing a customer-centric approach, the cosmetics industry can not only cater to existing needs but also anticipate and fulfill emerging ones. This scientific exploration of beauty and cosmetics consumption delves into the intricate tapestry of psychological drivers, social influences, and marketing strategies. By examining these interconnected themes, we gain valuable insights into human behavior, unveil the drivers of lucrative industries, and illuminate the cultural significance of beauty throughout history. Ultimately, a deep understanding of customer needs remains the cornerstone for unlocking the true potential of the ever-evolving beauty landscape. Thus, what is the priority and long-term choice of customers among many types of skincare cosmetics on the market, with quality and prices ranging from low to high? How can manufacturers perceive clients’ favorite and reliable products? To answer these questions, a number of studies have been conducted to measure the level of factors affecting the consumption of skin care cosmetics by consumers of various ages, genders, and regions. A study by Robertson et al. (2021) delves into the psychological motivations behind cosmetics use. It uncovers that individuals leverage makeup to project a desired identity, fulfilling a need for self-expression. People use cosmetics to manipulate their outward presentation to coincide with their internal self-concepts [4]. Furthermore, Dimitrov et al. (2023) underscores the importance of a customer-centric approach in a globalized beauty market. It discloses how beauty perceptions and cosmetics use vary significantly across different cultures. The investigation highlights the influence of societal norms and cultural backgrounds on consumer behavior [5].

Therefore, the central question guiding this study is: ‘What factors influence consumers' intention to repurchase skincare products in Can Tho City?’. Building on this research, the present study focuses on Can Tho City, Vietnam, and aims to identify the factors influencing consumers' intention to repurchase skincare products.

2. Literature review

2.1. Underpinning theory

The Theory of Planned Behavior (TPB), developed by Icek Ajzen [6], offers a robust framework for predicting and explaining purchase intentions related to beauty products. This theory builds upon the foundation of the Theory of Reasoned Action (TRA), expanding its scope to encompass perceived behavioral control. As of April 2020, the theory of planned behavior (TPB; Ajzen, 1991, 2012) has been subject to empirical scrutiny in more than 4,200 papers referenced in the Web of Science bibliographic database, rendering it one of the most applied theories in the social and behavioral sciences [7].

A thematic treemap analysis reveals that the TPB has received broad attention in areas such as the health sciences, environmental science, business and management, and educational research, fulfilling George Miller's “giving psychology away” request in an ideal sense. The TPB posits that attitudes, subjective norms, and perceived behavioral control interact to influence a consumer's behavioral intention, which is the strongest predictor of their actual purchase behavior (presented in Fig. 1). A consumer with a positive attitude towards a product, strong subjective norms to use it, and a high perception of control over purchasing it is highly likely to buy the product. Intention is thus assumed to be the immediate antecedent of behavior. To the extent that perceived behavioral control is veridical, it can serve as a proxy for actual control and contribute to the prediction of the behavior in question. By understanding these core constructs and their interaction, manufacter and researchers can gain valuable insights into consumer decision-making for beauty products. The TPB provides a framework for developing targeted marketing strategies and product offerings that resonate with consumer attitudes, social pressures, and perceptions of control.

Fig. 2.

The research model.

Fig. 1.

The theory of planned behavior model.

Note. This study's model was developed based on TPB theory in order to determine whether the following factors affected the customer's intention to buy cosmetic skin care products again. Fig. 2 shows the research model of repurchase intention of customers.

These variables are defined as follows:

• •Attitude: This refers to the consumer's overall evaluation of the skincare product, encompassing their feelings and beliefs about its effectiveness, quality, and value. A positive attitude suggests a favorable predisposition towards repurchasing the product.
• •Brand Credibility: This variable assesses the consumer's perception of the brand's trustworthiness, expertise, and reliability. High brand credibility fosters a sense of confidence in the brand and its products, positively influencing repurchase intentions.
• •Clarity of Product Description: This factor examines the extent to which product information is clear, understandable, and comprehensive. A clear and informative product description allows consumers to make informed decisions, leading to greater satisfaction and potential repurchase.
• •Trust in Retailer This variable evaluates the consumer's level of trust in the retailer from whom they purchased the skincare product. Trust in the retailer's reliability, fairness, and customer service significantly impacts repurchase intentions as it builds confidence in the purchasing process and product authenticity.
• •Intention: This refers to the consumer's conscious plan or willingness to repurchase the skincare product in the future. It signifies a positive inclination towards repeating the purchase based on their evaluation of the product and related factors.
• •Behavior: This variable captures the actual repurchase behavior of the consumer. It indicates whether the consumer has actually repurchased the product after their initial purchase, serving as a concrete measure of their intentions.

Hypothesis 1

The customer's attitude makes a positive perception of customers who want to repurchase the products in Can Tho city

Hypothesis 2

Skincare products having the clarity of product description make a positive perception of customers who want to repurchase the products in Can Tho city.

Hypothesis 3

Skincare products having the brand credibility make a positive perception of customers who want to repurchase the products in Can Tho city.

Hypothesis 4

Trust in retailer make a positive perception of customers who want to repurchase the products in Can Tho city.

3. Subjects and methods

3.1. Subjects

Data were collected from 531 consumers, who had bought skincare products for the second time or more.

3.2. Study participants

Individuals residing in Can Tho city who have purchased skincare cosmetics for the second time or more between November 1st, 2022 and November 31st, 2023. The research employed a team of student volunteers who underwent a comprehensive training program. This program equipped them with the necessary skills and knowledge to effectively approach designated target audiences. The training encompassed ethical considerations for participant recruitment and consent, effective communication techniques for data collection, and standardized protocols for recording and storing the gathered information. Utilizing these protocols, the volunteers approached members of the target audiences within pre-defined settings. They then engaged in structured interactions, collecting data through surveys, interviews, or focus group discussions depending on the specific research design. All collected data was meticulously recorded using standardized formats and securely stored in a designated database management system to ensure data integrity and facilitate subsequent analysis. Data was gathered throughout the daytime hours across all weekdays. The sampling locations encompassed diverse settings relevant to beauty product consumption behaviors. These included university campuses, allowing access to a population of young adults with potentially high beauty product engagement. In addition, data was collected within cosmetic stores, providing a context where consumers are actively interacting with and considering beauty products for purchase. Finally, shopping malls were included as locations with a high concentration of retail outlets offering beauty products, potentially capturing a broader range of consumer demographics and shopping habits.

3.3. Selection criteria

Participants aged 16 years or older, who purchased skin care products for the second time or more within any district of Can Tho city during the period 2022-2023, voluntarily agreed to participate in the survey.

3.4. Exclusion criteria

The study excluded individuals who had never purchased a skin care product, first-time skin care product customers, purchasers under 16 years of age, and those unable to answer the survey questions, do not provide complete answers in the survey or do not agree to participate. In addition, individuals who have never purchased skincare products or have not repurchased skincare products for the second time or more within the survey area were also excluded.

Structured survey questionnaires were conveniently delivered to 531 individuals residing in the area of Can Tho city, Vietnam throughout December 2022 to October 2023. The survey was conducted at 4 universities, 32 local cosmetic stores and 3 shopping centers in nine districts of Can Tho City.

3.5. Design of survey questionnaires and scale

The questionnaire includes 3 sections:

In the first part, respondents are queried about how frequently they use skin care products in order to filter out inappropriate responses.

The second section focuses on the participants' demographic variables, including age, gender, education level, employment status, monthly income, and details about their place of residence. Following that, participants are asked general information questions about skin care cosmetics such as their monthly expenditure on skincare products, usage frequency, preference for natural ingredients or quicker-acting yet potentially less safe chemical-specific treatment ingredients, purchasing locations for skincare products, and the expected duration for skincare items to show effectiveness.

The final part of the questionnaire is the main part. Participants will be asked about their attitudes, signal quality (including clarity and trustworthiness), their intention to make repeat purchases of skincare cosmetics, and the credibility of retailers.

To evaluate these factors, the 5-point Likert scale, which is a relatively popular measurement scale in scientific research questionnaires, was used to validate the opinions, behaviors, and perceptions of each individual. In this study, responses to the questions are scored in the following order: 1 = “Strongly disagree”, 2 = “Disagree”, 3 = “Partially agree”, 4 = “Agree”, 5 = “Strongly agree”. The author conducted preliminary research using qualitative methods and interview techniques. This involved engaging with a group of 5 experts from state health management agencies in Vietnam, lecturers at universities, as well as pharmacists from Can Tho Dermatology Hospital. Additionally, initial interviews were conducted with over ten customers.

We discussed, supplemented, removed inappropriate scales and adjusted the content of each comment to reinforce the research objectives and customers in Vietnam. After that, a preliminary survey was conducted involving 100 customers. To obtain high-quality outcomes, a good research study with relevant experimental design and accurate performance is required. Analyzing its feasibility prior to performing the main study (also known as the full study or large-scale main trial) can be very beneficial for this purpose. A pilot study is the first step of the entire research protocol and is often a smaller-sized study assisting in planning and modification of the main study.

A pilot study provides necessary information not only for calculating the sample size, but also for assessment of all other aspects of the main study, minimizing unnecessary effort from the researchers and participants, as well as the dissipation of research resources. In order for the pilot study to play its role, factors introduced in the text must be clearly defined before proceeding with the pilot study, and demonstrate a high level of completion. Furthermore, a pilot study provides valuable information, not only for the researcher's main study, but also for other similar studies; therefore, it is crucial to include complete information on the feasibility of the study [8].

Subsequently, we conducted an Exploratory Factor Analysis (EFA) followed by Confirmatory Factor Analysis (CFA) and Structural Equation Modeling (SEM) to evaluate the convergent and discriminant validity of the scale and proceeded to develop the official scale.

3.6. Sample size

Performing EFA requires a sufficiently large sample size, determined by the minimum size and the number of measured variables included in the analysis. According to Carpenter (2018), the minimum observed/measured variable ratio is 5:1, meaning that one measurement variable requires at least 5 observation samples [9]. The optimal ratio, however, is 20:1. In this study, the adjusted scale included 27 questions, hence requiring a minimum sample size of 135 and an optimal size of 540. We gathered a total of 531 samples for this study, meeting the appropriate sample size conditions.

3.7. Data analysis

Cronbach's Alpha test, invented by Lee Cronbach (1951) is remarkably popular in scientific research for assessing the internal consistency of a scale, ensuring that its constituent items measure a single underlying construct [10]. The value α, varying between 0 and 1 according to system theory, is proportional to the reliability of the scale. However, if α is at 0.95 or higher, it means that some items are redundant or the same question presented in different forms. This phenomenon is known as overlap repetition in the scale.

On the contrary, if α is low due to a limited number questions and weak relationships between the items (which can be improved by modifying or removing some items). Generally, a coefficient of 0.7 or above is considered acceptable for most research purposes [11]. However, this threshold may vary depending on the specific research context and the number of items in the scale [12]. Cronbach's alpha relatively easy to calculate and interpret, making it a popular choice for researchers. However, it's important to acknowledge its limitations. Alpha is primarily concerned with internal consistency, not necessarily the validity of the scale itself. A high alpha value doesn't guarantee that the scale measures the intended construct [11].

Understanding the underlying dimensions that shape consumer perceptions and behaviors towards beauty products is crucial for researchers and marketers alike. Factor analysis, a suite of statistical techniques, offers powerful tools to explore and confirm these latent constructs. However, two distinct approaches – EFA and CFA – serve different purposes within the research process.

Exploratory Factor Analysis (EFA) acts as a discovery tool. It allows researchers to identify the underlying factors that explain the variance observed in a set of measured variables [11]. Imagine a survey measuring various aspects of consumer attitudes towards a new skincare line (e.g., effectiveness, gentleness, price). EFA would statistically analyze these interrelated variables to potentially reveal a smaller number of underlying factors, such as “perceived efficacy” and “value for money.” Here, EFA is hypothesis-generating, helping researchers identify potential latent constructs to be further investigated [13].

Confirmatory Factor Analysis (CFA), in contrast, takes a hypothesis-testing approach [14]. Researchers leverage established theoretical frameworks or findings from previous EFA studies to develop a specific model of the relationships between latent constructs and measured variables [14]. Continuing with the skincare line example, a researcher might posit a model where “perceived efficacy” and “value for money” influence purchase intention. CFA then statistically evaluates how well this pre-defined model fits the observed data, providing evidence for the validity and reliability of the proposed latent constructs [15].

The choice between EFA and CFA hinges on the research stage. EFA is ideal for initial exploration when the underlying structure of the data is unknown [16]. Conversely, CFA is suited for later stages to confirm a pre-specified model based on existing theory or exploratory findings [17]. Recent applications in prestigious journals showcase the value of both approaches.

By strategically employing EFA and CFA, researchers can gain a deeper understanding of the complex factors shaping consumer behavior towards beauty products. EFA acts as a springboard for discovery, while CFA provides rigorous validation of the identified constructs. This combined approach empowers researchers to develop robust theoretical frameworks and inform effective marketing strategies within the ever-evolving beauty landscape.

CFA is a powerful statistical tool used to assess how well a hypothesized model fits the observed data. In simpler terms, imagine you have a theory about how certain variables relate to each other, and CFA helps you determine if the data confirms your theory [18]. However, evaluating this fit requires delving into some key parameters. Here's a breakdown to demystify these parameters:

Fit Indices: These are the overall gauges of how well your hypothesized model aligns with the data.

Chi-Square (χ²): This statistic assesses the absolute difference between the hypothesized model and the observed data. A low chi-square value is generally desirable, indicating a good fit. However, chi-square can be sensitive to sample size, so other fit indices are often considered alongside it [15].

Root Mean Square Error of Approximation (RMSEA): This index represents the average error of approximation between the model and the data. A low RMSEA value (ideally below 0.08) suggests a good fit.

Tucker-Lewis Index (TLI) & Comparative Fit Index (CFI): These indices compare the fit of your hypothesized model to a baseline model with no relationships between variables. Values above 0.9 for both TLI and CFI indicate a good model fit [15]. Imagine these as measures of relative improvement compared to a basic model with no structure.

Factor Loadings: These coefficients represent the strength of the relationship between each observed variable and its underlying latent construct (unobserved factor) in your model.

Standardized Factor Loadings: These values typically range from −1 to 1. A high positive loading (above 0.5) indicates that the variable is strongly related to the factor it's supposed to measure [19]. Conversely, a low loading suggests a weak relationship, potentially requiring further investigation or model adjustments.

Error Variances: These represent the amount of unexplained variance in each observed variable. Ideally, error variances should be low, indicating that most of the variation in the variable is explained by the underlying factors in your model.

Unlike Confirmatory Factor Analysis (CFA), EFA doesn't rely on pre-defined models.

Eigenvalues: These represent the amount of variance explained by each extracted factor. Factors with eigenvalues greater than 1 are generally considered to be meaningful contributors to the overall variance in the data [20].

Variance Explained: This reflects the total percentage of variance explained by all the extracted factors. Ideally, you want a high cumulative percentage (often aiming for over 60 %) to ensure the extracted factors capture most of the relevant information in the data [21].

Factor Loadings: Similar to CFA, EFA utilizes factor loadings to assess the strength of the relationship between each variable and the extracted factors.

Standardized Factor Loadings: These values, typically ranging from −1 to 1, indicate how strongly each variable “loads” onto a particular factor. A high positive loading (above 0.4) suggests the variable is strongly associated with that factor, while a low loading indicates a weaker relationship [21].

Communalities: This value represents the proportion of the variance in a particular variable that is explained by the extracted factors. A high communality (above 0.5) suggests the variable is well-explained by the factors, while a low communality indicates the variable may require further investigation or exclusion [21].

4. Results

A total of 1,674 potential participants were approached for inclusion in the study. Of these, 531 individuals provided complete data through survey participation, resulting in a response rate of 31.7 % (531/1674). Monthly income data were collected in Vietnamese Dong (VND), the local currency. After that, these values were converted to US Dollars (USD) using the prevailing exchange rate of 1 USD = 22,690 VND (recorded in January 2022).

4.1. Cronbach's alpha coefficients

4.1.1. Independent component

In general, the result table indicates that among the independent components, retailer credibility had the greatest current CA coefficient (0.975) compared to the other two components: attitude (0.934) and signal quality (0.927), which also showed a fairly good level of reliability.

4.1.2. Dependent component

After conducting the Cronbach's Alpha test, two factors (A6 and TIR3) were eliminated from the original scale due to lack of reliability (Tables 2 and 5). 18 factors in the independent components and 7 factors in the dependent component which fully met the requirements were retained (the total variable correlation coefficient >0.3 and CA coefficient after eliminating variables smaller than the current coefficient).

Table 2.

Internal reliability of the attitude component.

Variable-total correlation
Survey variable	Corrected Item - Total Correlation coefficient	CA coefficient when eliminating item
A1. Using skin care cosmetics regularly improves skin problems.
	0.924	0.907
A2. Skin care cosmetics improve my appearance.
	0.888	0.911
A3. Skin care cosmetics slow down the aging process.
	0.869	0.913
A4. Skin care cosmetics can aggravate existing medical conditions on the skin.
	0.920	0.907
A5. Using skin care cosmetics is often costly.
	0.919	0.908
A6. Using skin care cosmetics helps conceal skin imperfections.
	0.411	0.975
Cronbach's Alpha (CA): 0.934

Note: In Table 2, item A6, which involves using skin care cosmetics to help conceal skin imperfections, was removed because its CA coefficient after the variable removal (0.975) was greater than the current CA coefficient (0.934). The remaining 5 items were retained because they met the required conditions.

Table 5.

Internal reliability components trust in the retailer.

Variable-total correlation
Survey variable	Corrected Item - Total Correlation coefficient	CA coefficient when eliminating item
TIR1. Retailers provide complete, consistent, and accurate information about their products.
	0.942	0.968
TIR2. The retail facility is fully capable of satisfying my needs when I come to buy skin care cosmetics.
	0.945	0.967
TIR3. The retailer guarantees the safety of skin care cosmetics
	0.773	0.985
TIR4. Retailers have built trust with enough potential and quality.
	0.948	0.967
TIR5. Retailers have provided information about cosmetics in a safe, transparent and honest manner.
	0.950	0.967
TIR6. Retailers that have clearly conveyed product information as well as origin, without any contradictions can be reliable.
	0.954	0.966
Cronbach's Alpha (CA): 0.975

Note:Table 5 displayed internal reliability components trust in the retailer, in the component of retailer credibility, item TIR3, which pertains retailer ensuring the safety of skin care cosmetics, was excluded because its CA coefficient after variable elimination (0.985) exceeded the current CA coefficient (0.975), while 5 items left were kept of the signal quality component, no item was eliminated as their CA coefficient after variable elimination was lower than the current CA coefficient.

4.2. Exploratory factor analysis (EFA)

4.2.1. EFA analysis for subsection factors in independent components

With an extracted variance value exceeding 70 % (93.104 %), a KMO coefficient of 0.886 (0.5 ≤ KMO ≤1), a significant Bartlett test (Sig. < 0.05), the extraction of 04 groups of factors at Eigenvalue = 1.146 (which is > 1), all meeting the conditions, we proceeded to conduct EFA analysis for the subsection factors in the independent components.

EFA resulted in a refined structure with four distinct factors. Subsections were classified based on their correlation with each factor. Factor A exhibited the strongest loadings, ranging from 0.893 to 0.942. The CLA1 factor also demonstrated high loadings, with values of 0.888 and 0.902, respectively. Both the CRE and TIR factors displayed factor loadings exceeding 0.9, indicating a very strong association between the subsections and their underlying factors. Notably, all variables retained after EFA possessed factor loadings exceeding the customary threshold of 0.5. This high level of loading signifies the statistical significance and practical relevance of these variables within the identified factors.

4.2.2. EFA analysis for subsection elements in dependent components

The suitability of the data for EFA was confirmed through several indicators. Kaiser-Meyer-Olkin (KMO) statistic yielded a value of 0.866, exceeding the recommended threshold of 0.8 for good sampling adequacy [22]. Additionally, Bartlett's test of sphericity produced a statistically significant result (Sig. = 0.000), signifying the presence of significant correlations between the variables, a prerequisite for EFA [23].

The EFA yielded an eigenvalue of 6.602, surpassing the benchmark of 1.0, suggesting the presence of a single dominant factor that explained a substantial portion of the variance in the observed variables [20]. Furthermore, the total variance extracted reached 86.6 %, comfortably exceeding the customary 50 % threshold considered satisfactory for EFA [15].

Subsequently, the EFA was conducted using the observed variables associated with the dependent component, repurchase intention (RPI), aiming to explore the underlying dimensions influencing consumers' willingness to re-purchase the product.

A closer examination of the EFA results for RPI reveals that all subsections converged into a single factor, eliminating the need for label modifications (refer to Table 8 for details). Furthermore, the analysis indicates a strong correlation between the subsections and the factor, as evidenced by all factor loadings exceeding 0.5. Notably, the survey variable RPI5 exhibited the highest factor loading (0.965), signifying a particularly significant correlation with the RPI factor. Conversely, RPI3 displayed the lowest factor loading (0.897), although it still surpassed the recommended threshold.

Table 1.

The characteristics of study sample.

Characteristics of the study sample		Number (n = 531)	Frequency (%)
Gender	Male	135	25.4
	Female	396	74.6
Age	From 16 to 18 years old	34	6.4
	From 18 to 35 years old	342	64.4
	From 35 to 50 years old	146	27.5
	Over 50 years old	9	1.7
Education	High school	116	21.8
	Intermediate/college	83	15.6
	University	271	51.0
	After university	58	10.9
Career	Student	188	35.4
	Laborer	29	5.5
	Healthcare worker	122	23.0
	Office worker	60	11.3
	Private enterprise owner/entrepreneur/small business	84	15.8
	Teacher	11	2.1
	Housewife	19	3.6
	Others	18	3.4
Location	Rural area	133	25.0
	Suburban area	102	19.2
	City area	296	55.7
Monthly estimated amount you spend on skin care cosmetics	8.8$ - 22.03$	263	49.5
	22.03$ - 44.07$	193	36.3
	44.07$ - 220.36$	72	13.6
	>220.36$	3	0.6
Frequency of using your skin care cosmetics	Sometimes/month or fewer	42	7.9
	Sometimes/week	107	20.2
	Daily	329	62.0
	Above 2 times/day	51	9.6
	Others	2	0.4
Between choosing natural ingredients and quick-acting, less safe chemical-physical treatment ingredients, which one would you prefer to use?	Natural cosmetic	457	86.1
	Less safe and fast-acting cosmetic that contains specific chemical and physical ingredients	74	13.9
Where do you usually buy skin care cosmetics?	Pharmacy/pharmacy counter	114	21.5
	Hospital	26	4.9
	Hand good	39	7.3
	Social media/Internet	90	16.9
	Cosmetic store	254	47.8
	Others	8	1.5
How long does it take to see improvement since using skincare cosmetics?	2 weeks	103	19.4
	1 months	190	35.8
	2 months	152	28.6
	3 months	72	13.6
	Others	14	2.6

Note.Table 1 indicated that female buyers of skin care cosmetics accounted for a big chunk (396 people, 74.6 %) compared to their male counterparts (135 people, 25.4 %). The majority of survey participants fell in the age group of 18–35 (342 people, 64.4 %), followed by those aged 35 to 50 (146 people, 27.5 %). The remaining participants were from 16 to 18 years old (34 people, 6.4 %) or 50 years old or older (9 people, 1.7 %). A significant portion of shoppers held a university degree (271 people, approximately 51 %). Other categories, such as company owners, business people, office workers, and laborers, had lower purchase rates. The average monthly income per capita in Can Tho was around $236 [24]. People with a monthly income of less than $132 were the most frequent buyers of cosmetics (173 people, accounting for 32.6 %) and the lowest purchasers were those with an income was over $440 (76 people, 14.3 %). Moreover, residents in the city center used more skin care cosmetics (296 people, 55.7 %) than those in rural areas (133 people, 25 %) and suburbs (102 people, 19.2 %). In terms of the usage frequency, using skin care cosmetics on a daily basis was the highest percentage with 62 %, while using them several times a week was 20.2 %. There is an apparent distinction between using cosmetics with natural ingredients (over 86 %) and less-safe quick-acting chemical-physical treatment ingredients (under 14 %).

Table 3.

Internal reliability of the charity component.

Variable-total correlation
Survey variable	Corrected Item - Total Correlation coefficient	CA coefficient when eliminating item
The clarity
CLA1. The clarity of media has a positive impact on my trust in skin care cosmetics
	0.972	0.907
CLA2. Information on skin care cosmetics is clear and easy to understand.
	0.972	0.911
Cronbach's Alpha (CA): 0.986

Note: According to Table 3, due to the required conditions, the items CLA1 and CLA2 were retained.

Table 4.

Internal reliability of the reliability component.

Variable-total correlation
Survey variable	Corrected Item - Total Correlation coefficient	CA coefficient when eliminating item
The reliability
CRE1. Skin care cosmetics are trustworthy products.
	0,933	0,976
CRE2. I trust online retailers when buying skin care products.
	0,939	0,976
CRE3. I trust buying skin care cosmetics at reputable cosmetic stores.
	0,937	0,976
CRE4. I believe in the quality of skin care cosmetics at pharmacy counters and pharmacies.
	0,944	0,975
CRE5. I trust the brand's authenticity in terms of ingredients, origin, …
	0,922	0,977
CRE6. Promoted information about skin care cosmetics and effectiveness when using are consistent.
	0,922	0,977
Cronbach's Alpha (CA): 0.980

Note: As shown in Table 4, due to the required conditions, the 6 items CRE1, CRE2, CRE3, CRE4, CRE5, and CRE6 were retained.

Table 6.

Internal reliability of the repeat purchase intention component.

Variable-total correlation
Survey variable	Corrected Item - Total Correlation coefficient	CA coefficient when eliminating item
RPI1. I will repeatedly buy skin care cosmetics because the brand is safe and reputable.
	0.883	0.971
RPI2. I will repeat purchases of skin care cosmetics because the brand is famous and frequently featured in the media.
	0.888	0.971
RPI3. I will repurchase skin care cosmetics when I find them effective and suitable for me.
	0.860	0.973
RPI4. I am willing to recommend my relatives/friends, and the community to the skin care cosmetics that I have bought repeatedly.
	0.880	0.972
RPI5. I am willing to re-buy the skin care cosmetics that I trust at a higher price even if another company sells them at a lower price.
	0.951	0.967
RPI6. I will prioritize purchasing skin care cosmetics from retailers that I trust. The skin care cosmetics that I am using are worthy of my repurchase.
	0.951	0.967
RPI7. The skin care cosmetics that I am using are worthy of my repurchase.
	0.922	0.969
Cronbach's Alpha (CA): 0.974

Note: As per Tables 6, in the repeat purchase intention component, no item was eliminated because they all met the necessary criteria.

Table 7.

Rotated component matrix for the independent variable.

Rotated Component Matrixa
	Component
	1	2	3	4
CRE4	0.941
CRE3	0.937
CRE2	0.936
CRE1	0.933
CRE6	0.915
CRE5	0.913
TIR6		0.945
TIR5		0.943
TIR2		0.937
TIR1		0.933
TIR4		0.929
A1			0.942
A4			0.927
A5			0.926
A2			0.900
A3			0.893
CLA1				0.902
CLA2				0.888

Note. Consistent with the specifications outlined in Table 7, the initial four-factor structure of the survey variables remained stable after EFA. All six observed variables from the “Reliability” factor converged into a single factor, necessitating no label modification. Likewise, the five variables associated with the “Retailer Credibility” factor and the five variables from the “Attitude” factor each converged into distinct factors, retaining their original labels. Finally, the two variables from the “Clarity” factor converged into a single factor, again requiring no reassignment of labels.

Table 8.

Rotated component matrix for the dependent variable.

Component Matrixa
	Component
	1
RPI5	0.965
RPI6	0.962
RPI7	0.944
RPI2	0.918
RPI1	0.914
RPI4	0.911
RPI3	0.897

Note. The rotated component matrix results revealed a unidimensional structure for the observed variables. All seven variables converged into a single factor group, with each variable boasting a factor loading coefficient exceeding 0.5. This outcome, along with the EFA analysis conducted on both independent and dependent variables, suggests the suitability of all extracted factors for subsequent correlation and regression analyses.

4.3. Correlation between independent components and dependent components and Confirmatory Factor Analysis test results

Multivariate regression analysis was conducted using the “Enter” method.

Detailed results of the survey are provided in Table 9, Table 10, Table 11. CFA yielded satisfactory fit indices, including TLI = 0.985, CFI = 0.989, GFI = 0.932 (all exceeding 0.9), RMSEA = 0.051 (below 0.08), and Chi-square/df = 2.368 (less than 3) [20]. These values suggest good model fit for real data and support the unidimensionality of the scales, as proposed by Steenkamp and Van Trijp (1991). The factor rotation matrix further confirms this notion, as all observed variables (except RPI3) converged into a single factor with loadings exceeding 0.5. Due to its slightly lower loading of 0.497 (refer to Table 8), RPI3 was excluded from the model (see Table 12).

Table 9.

Durbin-Watson summary.

Variable	R	R2	Adjusted R2	Error estimation	Change Statistics
					Sig, F Change
1	0.525a	0.275	0.270	0.59207	0.000
Durbin-Watson: 1.716

Table 10.

Variance analysis.

Variable	Sum of squares	df	Mean squared	F	Sig
Variation due to regression	70.071	4	17.518	49.973	0.000
Variation due to residuals	184.389	526	0.351
Total	254.460	530

Table 11.

Regression coefficient.

Variable	Unstandardized coefficients		Standardized coefficients beta	t Sig.	Sig, cumulative statistic	Cumulative statistic
	Unstandardized coefficients	Standardized coefficients				Variable	Unstandardized coefficients
(Constant)	1.513	0.184		8.240	0.000	1.153	1.874
Aaa	0.283	0.043	0.307	6.632	0.000	0.199	0.367
CLAaa	−0.038	0.041	−0.043	−0.947	0.344	−0.118	0.041
CREaa	0.169	0.039	0.178	4.342	0.000	0.092	0.245
TIRaa	0.236	0.041	0.243	5.790	0.000	0.156	0.316

Note. Examining fit indices in Table Ap5 reveals a non-significant effect of the clarity factor on purchase intention (Sig >0.05). This suggests that clarity, as a latent variable in the model, does not statistically influence consumers' decisions to purchase skincare cosmetics. In contrast, the attitude factor emerges as the most influential variable, exhibiting a positive standardized coefficient of 0.307. This coefficient indicates a moderate positive effect [14], signifying that more positive attitudes towards skincare products lead to a stronger intention to purchase them. Conversely, the standardized coefficient of −0.043 for the clarity factor suggests a negligible negative effect [14]. While negative, the small magnitude implies a close to zero influence on purchase intention. Refer to Table 11 for a detailed overview of all standardized coefficients.

^aa: average.

Table 12.

Confirmatory factor analysis results.

CMIN/DF	GFI	CFI	TLI	RMSEA	PCLOSE	p	AGFI
2.368	0.932	0.989	0.985	0.051	0.401	0.000	0.902

The results of the Confirmatory Factor Analysis (CFA) test were presented in Fig. 3.

Fig. 3.

CFA test result.

Table 13 presents evidence supporting the distinctiveness of the identified concepts. Average Variance Extracted (AVE) values for each concept exceed the corresponding Maximum Shared Variance (MSV), indicating that more variance is explained by a concept itself than by its shared variance with other concepts [14]. Furthermore, the strong correlations between the concepts further bolster their distinctiveness, as highly correlated concepts would tend to share more variance.

Table 13.

Results of testing convergent validity and discriminant validity.

	CR	AVE	MSV	MaxR(H)	CRE	RPI	TIR	A	CLA
CRE	0.99	0.943	0.122	1.932	0.971
RPI	0.97	0.846	0.15	0.986	0.298	0.92
TIR	0.984	0.925	0.149	0.986	0.349	0.386	0.962
A	0.969	0.864	0.366	1	0.206	0.387	0.304	0.93
CLA	0.986	0.973	0.366	1.026	0.293	0.248	0.325	0.605	0.986

The scale also demonstrates satisfactory convergent validity and reliability. Convergent validity is reflected in the AVE values ranging from 0.943 to 0.973, all surpassing the 0.5 threshold recommended by Hair et al. (2019). These values indicate that a substantial proportion of the variance in the observed variables is captured by their underlying constructs. Additionally, composite reliability (CR) values span from 0.99 to 0.986, exceeding the recommended threshold of 0.7 [14]. This signifies high internal consistency and reliability within each construct.

In line with Table 14 and Table 15’s findings, we come to the conclusion that:

Hypothesis 1

Attitude has a positive impact on consumers' repurchase intention of skincare products in Can Tho city.

Hypothesis 2

Clarity of signal quality has a positive impact on consumers' repurchase intention of skincare products in Can Tho city.

Hypothesis 3

Credibility of signal quality has a positive impact on consumers' repurchase intention of skincare products in Can Tho city.

Hypothesis 4

Trust in retailer does not have a positive impact on consumers' repurchase intention of skincare products in Can Tho city.

Table 14.

Standardized regression weights.

	Estimate
RPI ← A	0.165
RPI ← CLA	0.260
RPI ← CRE	0.322
RPI ← TIR	−0.083

Table 15.

Regression weights.

			Estimate	S.E.	C.R.	P	Hypothesis
RPI	←	A	0.17	0.041	4.187	***	Accepted
RPI	←	CLA	0.264	0.043	6.18	***	Accepted
RPI	←	CRE	0.322	0.048	6.74	***	Accepted
RPI	←	TIR	−0.079	0.046	−1.733	0.083	Rejected

4.4. Structural equation modeling (SEM)

Unlike other statistical techniques that only allow estimating the partial relationship of each pair of factors (elements) in the classical model (measurement model), measuring direct as well as indirect effects, including measurement error and residual correlation, the SEM model allows flexibility in finding the most suitable model among the proposed models with the confirmatory factor analysis (CFA) technique. Bacon (1997) emphasized that SEM applications typically use sample sizes of 200–400 to fit models with a minimum of 10–15 observed variables [25].

Fig. 4.

SEM test result.

Note. The fit indices in Fig. 4 indicate a well-fitting model for the data. The Good Fit Index (GFI) of 0.932 exceeds the recommended threshold of 0.9, signifying a good match between the model and the observed data. The Root Mean Square Error of Approximation (RMSEA) of 0.051 falls within the acceptable range (<0.08), demonstrating a low level of error. Additionally, the chi-square value divided by degrees of freedom (2.368) is lower than the recommended cut-off of 3, suggesting a good fit.

Both the Comparative Fit Index (CFI) of 0.989 and the Tucker-Lewis Index (TLI) of 0.985 are exceptionally high, exceeding the commonly desired value of 0.9. These high values provide strong evidence that the model effectively captures the relationships between the latent variables in your theoretical framework.

The results of our structural equation modeling (SEM) analysis paint a very positive picture. Fit indices, including the Good Fit Index (GFI) of 0.932 and the chi-square value divided by degrees of freedom (2.368), suggest a strong convergence between the hypothesized model and the observed data [15]. Furthermore, the Root Mean Square Error of Approximation (RMSEA) of 0.051 falls well within the acceptable range, signifying a low level of error in the model's approximation of real-world relationships [25]. Particularly noteworthy are the exceptionally high values obtained for the Comparative Fit Index (CFI) of 0.989 and the Tucker-Lewis Index (TLI) of 0.985. These values significantly exceed the commonly desired threshold of 0.9, providing robust evidence that the model effectively captures the interrelationships between the latent variables within our theoretical framework [26].

In this study, there was a large gender gap, with women accounting for 74.6 % of the survey population. Besides, this difference in the ratio of men and women also appeared in Oberoi's (2018) study in Delhi, where the ratio was 78 % female and 22 % male participants [27].

In terms of age groups, individuals aged 18 to 35 account for the highest proportion (64.4 %), while those aged 50 and over has the lowest proportion (1.7 %). These results showed that young people are more interested in purchasing and also using skin care cosmetics than older age groups, as demonstrated by the similarity between the results of this study and those public findings, such as Di Palma's (2015) result [28].

Consumers with low income (under $132.22 per month or being supported by family) constitute the largest group (32.6 %), mirroring findings from Lim Kah Boon's 2020 research [29]. This similarity could be attributed to the comparable economic realities of developing economies, where a significant portion of the population falls within lower income brackets. Additionally, the high participation rate of students (35.4 %) aligns with a study conducted in Bangkok, where students comprised 44.5 % of the sample [30]. This finding might be linked to the cultural emphasis on education in Southeast Asia and the increasing consumerism among young adults, particularly students residing in urban areas with greater access to skincare products and trends.

The present study was conducted throughout Can Tho city, with the highest percentage of respondents residing in the inner city (55.7 %), followed by those in rural areas (25 %) and finally in the suburbs (19.2 %). As can be seen, the growing economy is fueling the expansion of cosmetics trading networks and beauty services, making them more accessible for people. However, monthly spending on skin care products among Can Tho city residents remains relatively low; with the largest proportion (49.5 %) falling between $8.8 and $22.03, and only 0.6 % spending more than $22.0. To elaborate, our team highlights that individuals prioritize purchasing products with affordable pricing while still guaranteeing quality, and that the majority of participants in this study have lower incomes. Furthermore, skin care cosmetics are not considered non-essential items, purchasing decisions are influenced by personal financial situations and specific needs.

The majority of survey participants use skin care cosmetics every day (62 %). Perhaps they believe that regular skin care will increase efficacy. Most individuals prefer natural cosmetics over those that promise quick results but come with a higher risk of side effects. People may opt for products that are both health-conscious and eco-friendly due to the progress of society and an increasing awareness of environmental protection.

Although cosmetics and skin care products are available for purchase on e-commerce platforms, people in Can Tho (47.2 %) still favor cosmetic stores for the desire to directly experience product samples, receive enthusiastic and helpful advice, ensure genuine products, and take advantage of appealing promotions.

After using skincare cosmetics, users often experience improvements. Some notice a difference in about 2 weeks (19.4 %), while others report improvements after 1 month (35.8 %), 2 months (28.6 %), or even 3 months (13.6 %). These results complement a quantitative research study conducted in Delhi [31] that also found a range of timeframes for experiencing noticeable effects from skincare products. While a direct comparison is limited due to potential differences in the specific products used and the demographics of the study populations, the overall trend suggests that the timeframe for observing improvements with skincare products can vary significantly among individuals.

Given its correspondence with market data and its demonstrated validity and reliability, the research model is deemed suitable for further analysis using CB-SEM (composite-based structural equation modeling). This model selection is consistent with recent trends in scientific research, suggesting its contemporary relevance.

5. Discussion

This study examines factors influencing repurchase intention for skincare products in Can Tho city, Vietnam. Interestingly, detailed product information (Clarity) regarding ingredients, usage, and target audience did not significantly impact repurchase decisions. This finding diverges from studies by Yee & Suan (2012) on the younger generation in Malaysia, where product information and design were key factors influencing purchase decisions (β = 0.38, p < 0.05) [32]. This discrepancy suggests potential age or cultural variations in how consumers process and utilize product information. Further research exploring these demographic nuances could be highly valuable.

On the other hand, it evidently identified a positive attitude towards skincare as the strongest predictor of repurchase intention (β = 0.307), echoing findings from Pathmaperuma & Fernando's (2018) research in Sri Lanka [33]. While this consistency suggests a universal appeal associated with enhancing appearance and well-being through skincare, it is crucial to recognize the influence of cultural factors. In both Vietnam and Sri Lanka, societal beauty standards and traditional practices may contribute to positive attitudes towards specific types of skincare products and routines. For example, the emphasis on fair and youthful skin in many Asian cultures could further strengthen the impact of a positive attitude on repurchase intention. Therefore, while the desire for improved appearance and well-being through skincare might be universally present, its expression and influence on consumer behavior are likely shaped by cultural nuances.

Building trust with the seller emerged as a crucial factor, with Trust in Retailer significantly influencing repurchase intention (β = 0.243). In our context, trust in the retailer likely extends to trust in the products they offer due to several factors. Consumers often associate reputable retailers with quality products, reliable information, and a curated selection process. Additionally, positive customer service experiences can further enhance trust in both the retailer and the products they sell. While our study focuses on trust in the retailer, research by Samir et al. (2022) in Bangladesh highlights the importance of perceived benefits in purchase decisions [34]. Although not directly comparable, both studies emphasize the role of consumer perceptions in influencing buying behavior. Future research could explore the interplay between trust and perceived benefits, specifically within the context of skincare product purchases.

Multivariate regression analysis revealed a non-significant effect of the CLA variable on repurchase intention (Sig. = 0.344 > 0.05). This suggests that clarity, as measured by the CLA construct, does not statistically influence consumers' decisions to repurchase skincare products in Can Tho city. Consequently, clarity is excluded from the final repurchase intention model.

5.1. Implications

These findings offer valuable insights for skincare brands and marketing professionals aiming to cultivate customer loyalty and drive repeat purchases.

Firstly, the disconnect between detailed product information (Clarity) and consumer behavior suggests a need for a shift in communication strategies. Instead of overwhelming consumers with comprehensive details, brands should focus on delivering clear, concise, and consumer-centric information. This might involve utilizing visuals, interactive elements, or targeted messaging tailored to specific demographics.

Secondly, the importance of a positive attitude towards skincare (Attitude) as a driver of repurchase intention underscores the value of emotional marketing. Highlighting the potential benefits of skincare products, such as improved appearance and enhanced well-being, can resonate with consumers and motivate them to repurchase.

Finally, the crucial role of Trust in Retailer emphasizes the necessity for building strong customer relationships. Providing complete, accurate, and transparent information, alongside excellent customer service, can foster trust and encourage repeat business. Additionally, partnering with reputable retailers can leverage their established trust to boost brand credibility.

By incorporating these research implications into their marketing strategies, skincare brands can cultivate stronger customer relationships, increase repurchase intention, and ultimately achieve long-term success in the competitive skincare market.

5.2. Limitations

While this study offers valuable insights into repurchase intention for skincare products in Can Tho city, Vietnam, it is essential to acknowledge some limitations that guide future research directions.

• +Sample Specificity: The study's findings are based on data collected in Can Tho city, Vietnam. Generalizability to other demographics, cultures, or regions may be limited. Future research should consider replicating the study with diverse samples to explore potential cultural or regional variations.
• +Self-Reported Data: The study relies on self-reported data from participants, which can be susceptible to biases such as social desirability or memory lapses. Future research could incorporate techniques like observing purchasing behavior or utilizing purchase history data to provide a more objective perspective.
• +Limited Scope of Factors: This study focused on a specific set of factors influencing repurchase intention. Other potential factors, such as brand image, marketing strategies, or social media influence,were not explored. Future research can broaden the scope to investigate the interplay of a wider range of variables.
• +Indirect Effects: While the study suggests potential indirect effects of Trust in Retailer and product safety on repurchase intention, these were not explicitly tested. Future research can utilize bootstrapping techniques or mediation analysis to explore these indirect relationships in greater detail.

By acknowledging these limitations and pursuing the suggested future research directions, we can continue to refine our understanding of consumer behavior in the skincare market. This will ultimately lead to the development of more effective strategies for skincare brands to cultivate customer loyalty and drive repeat purchases in the ever-evolving beauty industry.

6. Conclusion

In conclusion, this study explored the factors influencing repurchase intention for skincare products in Can Tho city, Vietnam. Interestingly, detailed product information did not significantly impact repurchase decisions, suggesting a need for clearer communication strategies. A positive attitude towards skincare emerged as the strongest driver, highlighting the desire for improved appearance and potential health benefits. Building trust with the retailer also proved crucial for repeat purchases. These findings offer valuable insights for skincare brands. Shifting communication to be clear, concise, and consumer-centric, emphasizing the emotional benefits of products, and fostering trust through transparency are key strategies to encourage repurchase intention. In essence, the strong fit indices and exceptionally high CFI and TLI values provide compelling academic evidence that the developed model accurately reflects the interplay between the variables under investigation. This paves the way for further exploration and interpretation of the model's findings. While limitations like sample specificity and self-reported data exist, future research can address these by using diverse samples, objective data collection methods, and exploring a broader range of variables.

Ethics declarations

This study was reviewed and approved by the Ethics Council in Biomedical Research of Can Tho University of Medicine and Pharmacy, with the approval number: [22.038. GV/PCT-HĐĐĐ]. The participants willingly agreed to take part in the study through verbal consent after receiving a thorough explanation of the study's objectives and contents. All participant details will remain confidential and solely used for research. All participants provided informed consent to participate in the study.

Consent for publication

The authors guarantee that the contribution to the work has not been previously published elsewhere.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for profit sectors.

Contribution

In the scientific realm, the results of the study reveal behavioral insights that are crucial for product development, fostering the creation of innovative, science-backed formulations.

In the field of marketing, this knowledge refines strategies, giving brands the capability to customize experiences, establish trust, and nurture enduring customer relationships. As a result, this promotes market sustainability and enhances customer satisfaction.

Understanding these factors can bolster economic viability by guiding businesses on effective strategies, thereby stimulating consumer loyalty and sustaining market growth within the skincare cosmetic industry.

Data availability statement

The data that support the findings of this study are available from the corresponding author Hung Phuc Nguyen (i.e. upon reasonable request).

CRediT authorship contribution statement

Huong V.T.M: Writing – review & editing, Validation, Resources, Methodology, Investigation, Conceptualization. Hung N.P: Writing – review & editing, Validation, Resources, Methodology, Investigation, Conceptualization. Minh N.T.T: Writing – review & editing. Thuy L.K: Resources, Investigation. Duyen L.T.N: Supervision, Project administration, Methodology. Minh T.N: Validation, Methodology, Formal analysis.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.