From Browsing to Buying: Determinants of Impulse Buying Behavior in Mobile Commerce

Abstract

Mobile commerce has transformed the retail landscape, yet the determinants of impulse buying behavior in this environment remain understudied, particularly in emerging markets. This research investigates the factors influencing impulse buying in mobile commerce in Chile using the Stimulus–Organism–Response framework. A quantitative cross-sectional study collected data from 451 mobile shoppers via an online survey. Structural equation modeling with PLS-SEM revealed that eight of the thirteen hypothesized relationships were significant. Mobile application factors (visual appeal and portability) positively influenced hedonic and utilitarian values. Among personal factors, economic well-being, family influence, and credit card use directly impacted impulse buying, while time availability did not. Hedonic value strongly influenced impulse buying behavior, but utilitarian value showed no significant effect. Contrary to expectations, the COVID-19 pandemic negatively impacted impulse buying. These findings extend theoretical understanding of mobile impulse buying determinants and provide practical insights for mobile commerce developers and marketers to enhance their platforms and strategies.

1. Introduction

The rapid development of wireless technology and the high penetration rate of mobile devices have transformed mobile commerce (m-commerce) into one of the most popular shopping channels [1,2]. This platform allows consumers to satisfy their purchasing needs quickly, as many sites require only an email and a password to complete transactions with a single click [3]. While shopping has become integral to daily life, many purchases are unplanned or sudden, associated with strong desires and feelings of joy and excitement [4,5]. This phenomenon, impulse buying, is significantly facilitated by internet access and mobile applications. According to Hausman [6], impulsive purchases trigger between 30% and 50% of all retail sales, and 90% of consumers make impulsive purchases occasionally.

Impulse buying is a growing global phenomenon that has evolved alongside new commerce modalities, particularly online sales. In 2024, global e-commerce sales reached $6.09 trillion worldwide, representing approximately 20% of total retail sales, with projections reaching $6.56 trillion by 2025 [7]. More than 72% of online shoppers have impulsively bought an item due to an advertised discount, with the average consumer spending $281.75 per month on impulse purchases in 2024 [8]. In the United States, more than 35% of consumers reported they were most likely to make impulse purchases online in 2023 [9].

Despite extensive research in traditional retail and desktop e-commerce contexts, impulse buying in mobile commerce exhibits distinct empirical gaps that require specialized investigation. Current literature demonstrates a cross-platform research deficiency, where most studies examine traditional retail and desktop e-commerce settings with insufficient exploration of mobile-specific behavioral patterns [10]. Additionally, methodological limitations persist as existing impulse buying measurement instruments were designed for traditional retail or desktop contexts and fail to capture unique mobile behavioral patterns, creating significant measurement gaps in the field [11,12].

The general objective of this research is to identify and contrast the variables that influence impulse buying behavior on mobile commerce platforms, while also analyzing the effect of the COVID-19 pandemic on impulse purchases. Following the Stimulus–Organism–Response (S-O-R) theoretical framework, which has been successfully applied to understand consumer behavior in e-commerce contexts [13]. This study examines how various stimuli in the mobile shopping environment affect consumers’ internal states, ultimately leading to impulsive purchase responses. Specifically, this study aims to: (1) identify variables influencing impulse buying in mobile commerce; (2) analyze relationships between variables proposed as antecedents of impulsive behavior in mobile commerce; (3) analyze the effect of impulse buying; (4) test the relationships and hypotheses indicated in the proposed model.

Based on the Stimulus–Organism–Response (S-O-R) framework, this research proposes an integrated model examining multiple determinants of impulse buying behavior in mobile commerce [14,15,16]. The model considers two categories of stimuli: mobile application factors (visual appeal, portability, and task-relevant information) and personal factors (economic well-being, family influence, time availability, and credit card use). These stimuli are proposed to influence impulse purchases directly and indirectly through the mediating effects of hedonic and utilitarian value perceptions. Additionally, trust is examined as a critical factor in the mobile shopping context, where transactions occur without physical interaction with products or salespeople. The model also incorporates the COVID-19 effect as a unique situational factor that may have unprecedentedly altered impulse buying patterns. This comprehensive approach integrates technological, personal, and situational variables, providing a more holistic understanding of impulse buying behavior in the mobile commerce environment.

2. Theoretical Background

2.1. Impulse Buying Behavior in Mobile Commerce

Contemporary studies recognize that impulse buying is a multifaceted construct encompassing conative, visceral, and cognitive dimensions, including rapid decision-making, on-the-spot actions, and compelling psychological urges [17]. Contemporary studies recognize that individual traits do not solely determine impulse buying, but are also influenced by a complex interplay of internal and external factors, including psychological characteristics such as sensation-seeking, varying motives (hedonic and utilitarian), consumer resources (time and money), and marketing stimuli [18] https://link.springer.com/article/10.1007/s11747-019-00670-w, accessed on 3 September 2025. Additionally, the digital transformation of commerce has fundamentally altered impulse buying patterns, with the convenience and automation of online and mobile shopping making impulsive purchases increasingly accessible, while targeted advertisements and social media presence further facilitate hasty purchase decisions [19].

Specifically, existing research lacks an examination of mobile-unique environmental factors, including portability effects, location-based triggers, time-fragmented usage patterns, and device-switching behaviors that fundamentally distinguish mobile commerce experiences from those on desktops. Mobile commerce and traditional e-commerce demonstrate substantial behavioral and technological differences that warrant distinct theoretical consideration. Mobile environments promote intuitive, heuristic processing, while desktop environments favor analytical, deliberative thinking, creating fundamentally different decision-making pathways, affecting impulse buying behavior [20]. Empirical evidence reveals that mobile platforms show 16.7% lower conversion rates (2.0% vs. 2.4% desktop) and 47.6% lower average order values ($104.83 vs. $155), while exhibiting 43.2% higher add-to-cart rates, suggesting different engagement patterns rather than inferior performance [21]. Touch interface interactions create different gesture patterns and psychological ownership effects compared to desktop pointing-and-clicking behaviors, with mobile users showing 67% more navigation difficulties due to screen size limitations that affect cognitive processing during purchase decisions [11]. These platform-specific characteristics necessitate dedicated theoretical frameworks for understanding impulse buying in mobile contexts.

2.2. The Stimulus–Organism–Response (S-O-R) Model

The theoretical foundation of this research is the Stimulus–Organism–Response (S-O-R) model, developed by Mehrabian and Russell [22] and grounded in environmental psychology. This model focuses on the impact of stimuli on three emotional states—pleasure, arousal, and dominance—providing the theoretical basis for understanding the effect of environmental elements on consumer behavior. In online shopping environments, the S-O-R theory has proven to be an appropriate theoretical lens for understanding how web-based stimuli influence online consumer behavior [23,24,25], with recent studies successfully applying it to mobile commerce contexts [13,15].

The S-O-R model consists of three key components: First, the Stimulus is defined as the actual product or the extrinsic attributes of the product and the purchasing context, such as the shopping environment, sales, and other shoppers [26]. In mobile commerce, stimuli include mobile application factors (visual appeal, portability, information) and personal factors (economic well-being, family influence, time availability, credit card use). Secondly, an Organism refers to consumers’ internal evaluations, and the judgment is generated when interacting with different stimuli [22,27]. It has been detailed as “previous experiences, knowledge, beliefs, attitudes, predispositions, intentions, values, cognitive networks, schemes, scripts, motives, the individual’s personality, feelings” [28]. Trust, hedonic value, utilitarian value, and COVID-19 perception act as organizational elements in this research. Responses represent the results and user decisions based on cognitive, affective, or emotional reactions and include an approach or avoidance behavior [29,30]. In the shopping context, the response has two perspectives: the implicit need to buy and the actual implicit buying behavior [31]. In our research, impulse buying behavior is the response.

2.3. Antecedents of Impulse Buying in Mobile Commerce

Based on the S-O-R framework and previous literature, we propose a model that examines multiple determinants of impulse buying behavior in mobile commerce. The model considers stimuli, organism, and response factors within the mobile commerce context, integrating mobile application factors, personal factors, and value perceptions as key antecedents.

2.3.1. Mobile-Specific Extensions to S-O-R Framework

This study extends the traditional S-O-R framework by incorporating mobile-specific environmental factors that create unique stimulus conditions not present in desktop or offline contexts [11,20]. These mobile-unique extensions include: (1) portability-enabled contextual purchasing triggers that allow consumers to shop spontaneously across different physical locations and temporal contexts [32,33]; (2) touch-interface psychological effects that enhance emotional engagement through haptic feedback, creating more substantial psychological ownership compared to traditional point-and-click interfaces [34,35]; (3) location-awareness capabilities that provide real-time, contextually relevant stimuli based on geographic positioning [1,36]; (4) seamless payment integration that reduces purchase friction through one-touch purchasing systems [37,38].

These constraints create what we term “mobile-mediated impulse facilitation”—a phenomenon where technological limitations paradoxically enhance impulsive behaviors by reducing cognitive load and accelerating decision-making processes [5,39]. Furthermore, mobile commerce introduces novel organism-level mediators, including device attachment, mobility context, and fragmented attention patterns that interact with traditional value perceptions to create hybrid evaluation processes unique to the mobile context, thereby extending existing theoretical understanding of impulse buying mechanisms [2,15].

2.3.2. Mobile Application Factors

Visual Appeal

Visual appeal utilizes vibrant colors, crisp displays, aesthetically pleasing screen layouts, and captivating imagery to draw users in. This becomes particularly crucial in digital environments where visual interaction is the primary means of user engagement [24,35,40]. Mobile applications’ visual appeal can substantially affect evaluation and enjoyment, with recent research demonstrating significant relationships between visual elements and impulse buying behaviors [34,39,41]. Therefore:

H1: 

Visual appeal directly and positively impacts hedonic value in online impulse buying.

H2: 

Visual appeal directly and positively impacts utilitarian value in online impulse buying.

Portability

Portability represents the key characteristic that mobile users appreciate, referring to the physical convenience of mobile devices that enables ubiquitous access to shopping applications [11,32,33]. This capability allows users to access shopping platforms anytime and anywhere, subject only to network connectivity, thereby increasing exposure to purchase stimuli and reducing traditional time and space limitations [36,39,42]. As connection opportunities increase, portability enables consumers to find information or make purchases immediately when desires arise. Therefore:

H3: 

Portability has a direct and positive impact on hedonic value in online impulse buying.

H4: 

Portability directly and positively impacts utilitarian value in online impulse buying.

Task-Relevant Information

Task-relevant information in mobile commerce refers to the degree to which mobile applications provide convenient, accurate, and contextually appropriate content that facilitates consumers’ shopping objectives within the constraints of mobile interfaces [34,43,44]. Unlike desktop environments, where extensive product details can be displayed, mobile applications must balance information richness with screen size limitations and user attention spans [11,20,26]. However, the effectiveness of task-relevant information in mobile impulse buying contexts may be reduced, as the spontaneous nature of impulse purchases and mobile-specific factors such as simplified interfaces may alter traditional information-purchase relationships. Therefore:

H5: 

Task-relevant information has a direct and positive impact on online impulse buying.

2.3.3. Personal Factors

Economic Well-Being

Economic Well-being represents consumers’ financial capacity and perceived economic health, serving as a fundamental determinant of purchasing behavior [45,46,47]. Consumers with higher economic well-being demonstrate greater willingness to engage in impulsive spending, as improved financial resources reduce purchase-related anxiety and increase discretionary spending capacity [48,49,50]. Research consistently shows that money availability serves as a relevant element in the impulse buying process, with consumers being more impulsively willing to spend during periods of economic prosperity. Therefore:

H6: 

Economic well-being has a direct and positive impact on online impulse buying.

Family Influence

Family Influence refers to the normative and informational impact of close family members on consumers’ purchasing behavior, reflecting the extent to which consumers are guided by family opinions and expectations [51,52,53]. Family members serve as powerful reference groups that shape consumer behavior through both direct recommendations and implicit social pressure, with greater susceptibility to interpersonal influence resulting in a higher tendency toward impulsive purchases [4]. Research demonstrates that the presence of close relationships has a positive impact on customers’ normal buying behavior, particularly when strong emotional bonds exist between the consumer and family members. Therefore:

H7: 

Family influence directly and positively impacts online impulse buying.

Time Availability

Time availability refers to consumers’ perception of temporal resources available for shopping activities, which has been found to affect purchase decisions directly [13,54,55,56]. When consumers perceive adequate time for shopping, they experience greater freedom to explore products beyond their initial shopping intentions, potentially leading to unplanned purchases [44,57,58]. Research indicates that consumers often prefer online commerce platforms because they can purchase during their free time, with having enough time to shop causing consumers to feel greater freedom to explore products not initially on their shopping list. However, mobile commerce’s streamlined interfaces may neutralize traditional time constraint effects. Therefore:

H8: 

Time availability directly and positively impacts online impulse buying.

Credit Card Use

Credit card use facilitates impulse purchases through convenient, secure, and rapid payment processing that reduces purchase friction [37,38]. Credit cards create a perception of “painless” spending by eliminating immediate cash outflow, generating psychological distance from actual monetary expenditure, and encouraging spontaneous purchase decisions [45,50]. This payment method generates increased impulse purchases due to easy access and the consumer’s sense of convenience, as there is no tangible money expenditure, creating a perception of low cost in purchases made. Therefore:

H9: 

The use of Credit cards directly and positively impacts online impulse buying.

2.3.4. Mediating Value Factors

Hedonic Value

Hedonic value refers to the comprehensive assessment of experiential advantages derived from shopping, including elements like entertainment, pleasure, and emotional satisfaction [42,59,60]. In mobile commerce contexts, hedonic value captures the enjoyment and emotional gratification consumers derive from mobile shopping experiences, serving as a key mediator between environmental stimuli and impulsive behavior [39,61]. Research indicates that hedonic value satisfies people’s emotional needs to enjoy shopping and make the experience pleasant, with mobile platforms particularly suited for delivering immediate gratification and entertainment-focused shopping experiences. Therefore:

H10: 

Hedonic value directly and positively impacts online impulse buying.

Utilitarian Value

Utilitarian value represents the functional benefits and practical advantages consumers perceive from mobile shopping, including convenience, efficiency, and task accomplishment [42,44,62]. While utilitarian motivations typically involve planned, rational decision-making, mobile commerce’s convenience and accessibility may enable utilitarian value to contribute to impulsive behaviors [39,50]. Research suggests that utilitarian value in mobile contexts encompasses time savings, ease of comparison shopping, and immediate access to desired products, which can unexpectedly trigger spontaneous purchase decisions despite initial rational intentions. Therefore:

H11: 

Utilitarian value directly and positively impacts online impulse buying.

2.3.5. Trust

Trust represents consumers’ confidence in the reliability, security, and integrity of mobile commerce platforms, particularly crucial given the absence of physical product interaction [63,64]. In mobile environments where transactions occur through small screens and digital interfaces, trust becomes essential for reducing perceived risk and facilitating purchase decisions [33,50,65]. Research demonstrates that trust plays a critical role in mobile shopping contexts, where consumers must rely on digital representations and cannot physically inspect products before making a purchase. Therefore:

H12: 

Trust has a direct and positive impact on online impulse buying.

2.3.6. COVID-19 Perception

COVID-19 Effect represents the unprecedented situational influence of the global pandemic on consumer behavior and psychology, theoretically grounded in Event Systems Theory [66,67,68]. The pandemic altered consumption patterns through economic uncertainty, increased digital adoption, and changed lifestyle priorities, potentially affecting impulsive purchasing tendencies in mobile commerce contexts [69,70]. Research indicates that COVID-19’s impact operates through multiple psychological mechanisms, including financial anxiety, loss of control, and information overload, creating complex effects on consumer decision-making processes. Therefore:

H13: 

COVID-19 crisis perception directly and positively impacts online impulse buying.

Figure 1 presents the proposed model for analyzing the antecedents of impulsive buying behavior, supported by the stimulus–organism–response model.

3. Materials and Methods

3.1. Instrument Design and Data Collection

This research employed a descriptive-causal approach using a hypothetico-deductive method. A quantitative study with a cross-sectional design was conducted through a structured online questionnaire to validate the hypotheses formulated in the research model. Data were collected between September and December 2020 in Chile.

Regarding scale translation, all English measurement instruments underwent rigorous English-to-Spanish translation, including forward translation, independent back-translation, expert review, and pretesting with Chilean mobile commerce users to ensure cultural appropriateness and comprehension clarity.

Non-probabilistic sampling was used to select participants, targeting individuals over 18 who had purchased through a mobile commerce platform in the previous three months. The sampling strategy prioritized geographic regions with the highest mobile commerce penetration to ensure participants possessed relevant experience with mobile shopping platforms. The survey yielded a total of 503 respondents. Of these participants, 89.7% confirmed having purchased a mobile application during the previous three months, resulting in an adequate sample of 451 valid responses for analysis. This sample size exceeded the minimum of 384 calculated using a confidence level of 95% and a margin of error of 5%.

Table 1. Socio-demographic data of the sample.

Variable	Category	Frequency	Percentage
Age range	18–24 years	209	46.3%
	25–34 years	152	33.7%
	35–44 years	59	13.1%
	45+ years	31	6.9%
Gender	Female	288	63.9%
	Male	163	36.1%
Education level	High school	72	16.0%
	Technical education	73	16.2%
	University	293	65.0%
	Postgraduate	13	2.8%
Employment status	Student	189	41.9%
	Employee	172	38.1%
	Self-employed	63	14.0%
	Unemployed	27	6.0%
Geographic location	Northern zone	324	71.8%
	Central zone	92	20.4%
	Southern zone	35	7.8%

presents the demographic characteristics of the sample. Most respondents (46.3%) were between 18 and 24 years old, and 63.9% were female. Regarding education level, 65.0% had a university education. As for employment status, 41.9% were students. Geographically, 71.8% of respondents were from the northern zone of Chile. The high concentration of participants from north Chile (71.8%) reflects a purposive sampling approach targeting regions with the highest internet penetration rates (94%) and mobile device adoption. This concentration, while potentially limiting generalizability to rural or less digitally connected areas, enhances internal validity by focusing on Chile’s most mobile-commerce-active population and controlling for regional infrastructure variations. The sample’s high educational attainment (65% university education) suggests economic homogeneity that reduces potential confounding effects from socioeconomic disparities. Also, the survey revealed that clothing and accessories were the most frequently purchased category, followed by food, beverages, and tobacco. Pedidos Ya, Falabella, and Mercado Libre were the most used Chilean mobile applications among participants.

3.2. Measurements

The research model includes several variables related to impulse buying behavior in mobile commerce. All variables were measured using a 5-point Likert scale, where 1 = “Strongly Disagree”, 2 = “Disagree”, 3 = “Neutral”, 4 = “Agree”, and 5 = “Strongly Agree”.

Table 2. Variables, Abbreviations, and Measurement Scales.

Variable	Items	Source
Visual Appeal (AT)	3	Parboteeah et al. [26]
Portability (P)	3	Okazaki & Mendez [71]
Task-Relevant Information (IAT)	3	Parboteeah et al. [26]
Economic Well-being (BE)	3	Rosenstone et al. [49]
Family Influence (IF)	3	Badgaiyan & Verma [45]
Time Availability (DT)	3	Beatty & Ferrell [46]
Credit Card Use (UTC)	3	Badgaiyan & Verma [45]
Hedonic Value (VH)	4	Park et al. [34]
Utilitarian Value (VU)	5	Park et al. [34]
Trust (CON)	3	Yi & Jai [63]
COVID-19 Effect (EC)	4	Lins & Poeschl [72]
Impulse Buying (CPI)	5	Park et al. [34]

summarizes all variables, their abbreviations, the number of items, and the source of each measurement scale.

3.3. Data Analysis

The data analysis employed structural equation modeling (SEM), utilizing the partial least squares approach (PLS-SEM) through SmartPLS 3.0 software. This analytical framework was selected due to its capacity to maximize variance explanation and facilitate the examination of complex models containing multiple interconnected latent constructs [73,74].

The analytical procedure was executed through a two-phase approach. Initially, the measurement model underwent evaluation using reliability measures (Cronbach’s alpha and composite reliability coefficients) [75], as well as convergent validity assessment (average variance extracted) and discriminant validity verification (Fornell–Larcker criterion [76] and heterotrait–monotrait ratio [77]). Subsequently, the structural model was examined by analyzing path coefficients, t-statistics, p-values, and coefficient of determination (R²) to establish the statistical significance of the proposed relationships and assess the model’s explanatory capacity [74]. The statistical significance of path coefficients was determined through bootstrapping procedures that utilized 5000 resamples. Furthermore, the model’s predictive validity was evaluated using the Stone-Geisser Q² statistic via the blindfolding technique [73,78].

4. Results

4.1. Measurement Model Analysis

The data analysis was performed utilizing the partial least squares (PLS) methodology, which proves suitable for examining intricate associations among latent constructs. This approach enables the explanation of empirical data while facilitating predictive analysis, serving as a significant component in scholarly investigation. PLS demonstrates value through its ability to quantify the influence of one variable upon another [73,79].

PLS was chosen for its mathematical flexibility, which does not establish rigid premises in data distribution, measurement scale, or sample size [80]. SmartPLS software version 3.0 was used to calculate the indicators suggested by Hair et al. [74]. The measurement model was evaluated for convergent validity and individual indicator reliability, followed by scale reliability assessment, discriminant validity analysis through variable correlations, and discriminant validity analysis through cross-loadings.

4.1.1. Convergent Validity and Indicator Reliability

Table 3. Reliability and Validity of Variables.

Construct	Cronbach’s Alpha	rho_A	Composite Reliability	Average Variance Extracted (AVE)
Visual Appeal (AT)	0.978	0.978	0.986	0.958
Economic Well-being (BE)	0.942	0.954	0.963	0.895
Impulse Buying (CPI)	0.914	0.916	0.936	0.746
Trust (CON)	0.903	0.911	0.940	0.839
Time Availability (DT)	0.826	0.853	0.895	0.740
COVID-19 Effect (EC)	0.896	1.056	0.926	0.760
Family Influence (IF)	0.894	0.914	0.933	0.823
Task-Relevant Information (IAT)	0.932	0.938	0.957	0.880
Portability (P)	0.914	0.919	0.946	0.853
Credit Card Use (UTC)	0.853	1.029	0.897	0.744
Hedonic Value (VH)	0.893	0.898	0.926	0.759
Utilitarian Value (VU)	0.889	0.895	0.919	0.696

presents the reliability and validity measures for all research constructs examined in this study. Construct reliability was evaluated using two established criteria: Cronbach’s alpha coefficient and composite reliability index. As noted by Chin [81], composite reliability represents a more suitable measure than Cronbach’s alpha for PLS analyses, given that it does not presuppose equal weighting across all indicators.

The composite reliability coefficients ranged from 0.895 to 0.986, indicating robust reliability since all values surpass the established threshold of 0.6 [80]. Correspondingly, Cronbach’s alpha coefficients varied from 0.826 to 0.978, which are deemed satisfactory as they exceed the recommended minimum threshold of 0.707 [74,82]. Convergent validity was assessed through the Average Variance Extracted (AVE) criterion, which quantifies the proportion of variance within a construct that can be accounted for by the selected indicators [74,80]. An AVE coefficient of 0.5 or greater signifies that the construct accounts for more than half of its indicators’ variance [73]. As evidenced in Table 3, AVE coefficients ranged from 0.696 to 0.958, substantiating that each variable accounts for at least 50% of its indicators’ variance. Furthermore, all Dillon–Goldstein’s Rho coefficients exceeded 0.7, confirming reliability [78,79]. Based on these findings, it can be concluded that the measurement model exhibits adequate validity and reliability.

4.1.2. Discriminant Validity Analysis

Discriminant validity represents the difference between one variable and other variables in the proposed model. As a necessary condition for this validity, the shared variance between a variable and its indicators should be greater than the shared variance with other model variables [83]. Additionally, correlations between different factors in the model should not exceed 0.8 [84].

Table 4. Discriminant Validity—Fornell–Larcker Criterion.

Variable	AT	BE	CPI	CON	DT	EC	IF	IAT	P	UTC	VH	VU
AT	0.979
BE	0.316	0.946
CPI	0.343	0.321	0.864
CON	0.570	0.427	0.360	0.916
DT	0.090	0.155	0.235	0.179	0.860
EC	0.296	−0.030	0.074	0.268	0.139	0.872
IF	0.255	0.223	0.547	0.256	0.260	0.199	0.907
IAT	0.636	0.322	0.316	0.649	0.097	0.229	0.159	0.938
P	0.617	0.336	0.229	0.568	0.065	0.298	0.133	0.683	0.924
UTC	0.205	0.183	0.466	0.208	0.193	0.096	0.414	0.184	0.159	0.862
VH	0.460	0.246	0.562	0.435	0.194	0.201	0.378	0.403	0.364	0.366	0.871
VU	0.514	0.365	0.298	0.593	0.245	0.232	0.171	0.503	0.491	0.126	0.449	0.834

presents the discriminant validity assessment using the Fornell–Larcker criterion, where the diagonal values represent the square root of AVE for each construct.

As shown in Table 4, the square root values of AVE (diagonal values) are greater than the correlations between variables, indicating that correlations between constructs do not exceed the square root value of the AVE of each construct, thus meeting the criterion for discriminant validity [77,85]. Additionally, correlations between different factors do not exceed the value of 0.8. Further discriminant validity assessment was conducted using the Heterotrait–Monotrait (HTMT) ratio criterion, which requires values below 0.9 [86].

Table 5. Heterotrait–Monotrait Ratio.

Variable	AT	BE	CPI	CON	DT	EC	IF	IAT	P	UTC	VH	VU
AT
BE	0.327
CPI	0.365	0.345
CON	0.607	0.462	0.397
DT	0.104	0.178	0.266	0.210
EC	0.314	0.081	0.088	0.291	0.159
IF	0.276	0.243	0.593	0.287	0.289	0.225
IAT	0.669	0.343	0.343	0.709	0.109	0.252	0.177
P	0.650	0.360	0.252	0.624	0.088	0.335	0.148	0.741
UTC	0.258	0.209	0.455	0.265	0.205	0.120	0.416	0.229	0.224
VH	0.499	0.264	0.623	0.483	0.220	0.219	0.440	0.419	0.400	0.390
VU	0.547	0.399	0.333	0.664	0.291	0.253	0.196	0.554	0.540	0.193	0.501

shows the HTMT values ranging from 0.088 to 0.741, further confirming the measurement model’s discriminant validity.

4.2. Structural Model Evaluation and Hypothesis Testing

After validating the measurement model, we evaluated the structural model to answer two key questions posed by Falk and Miller [86]; 1. How much variance in endogenous variables is explained by their predictor variables? 2. To what extent do exogenous variables contribute to the explained variance of endogenous variables? To address these questions, we examined Pearson’s Correlation (R²), the Stone-Geisser Indicator (Q²), the Cohen Indicator (f²), and Path Coefficients, calculating the PLS Algorithm and performing resampling with the Bootstrapping method. Additionally, we assessed the standardized root mean square residual (SRMR).

4.2.1. Pearson’s Correlation (R²) Evaluation

The Pearson correlation coefficient indicates the quality of the model. Specifically, it shows the variance in a construct explained by predictor variables [81].

Table 6. Pearson’s Correlation.

Construct	R Square	Adjusted R Square
Hedonic Value (VH)	0.228	0.225
Utilitarian Value (VU)	0.313	0.310
Impulse Buying (CPI)	0.505	0.495

displays the R² values obtained from the PLS Algorithm result. The results indicate that the model explains 22.8% of Hedonic Value, 31.3% of Utilitarian Value, and 50.5% of Impulse Buying. This suggests that the model effectively captures the relationships between the study variables.

4.2.2. Stone–Geisser Indicator (Q²)

The StoneGeisser indicator (Q²) analyzes the model’s predictive quality, indicating how closely the model approximates what was expected [87]. The evaluation criterion suggests that the value obtained should be greater than zero [88]. A perfect model would have Q² = 1, reflecting a reality without errors. As shown in

Table 7. Stone–Geisser Indicators (Q²).

Construct	Stone–Geisser (Q2)
Impulse Buying (CPI)	0.370
Hedonic Value (VH)	0.170
Utilitarian Value (VU)	0.210

, all indicators have Q² values greater than zero, guaranteeing the predictive relevance of the model.

4.2.3. Standardized Root Mean Square Residual (SRMR)

Subsequently, the standardized root mean square residual (SRMR) was calculated to assess model fit. According to Henseler et al. [83], SRMR serves as a goodness-of-fit measure for PLS-SEM, helping to prevent incorrect model specification. Values below 0.08 are considered indicative of adequate model fit. As presented in

Table 8. Standardized Root Mean Square Residual.

Fit Measure	Saturated Model	Estimated Model
SRMR	0.058	0.084
d_ULS	3.024	6.340
d_G	1.310	1.420
Chi-square	3438.239	3590.218
NFI	0.808	0.800

, the proposed research model yielded an SRMR value of 0.058, thereby confirming that the model demonstrates a satisfactory fit to the data.

4.2.4. Path Coefficients and Hypothesis Testing

Finally, it analyzed whether the regressions proposed in the model are significant.

Table 9. Summary of Results for Hypothesis Testing.

Hypothesis	Independent Variable	Dependent Variable	Beta	t-Statistic	p-Value	Result
H1	Visual Appeal (AT)	Hedonic Value (VH)	0.393	7.270	0.000	Accepted *
H2	Visual Appeal (AT)	Utilitarian Value (VU)	0.340	5.890	0.000	Accepted *
H3	Portability (P)	Hedonic Value (VH)	0.122	2.121	0.000	Accepted *
H4	Portability (P)	Utilitarian Value (VU)	0.281	4.850	0.031	Accepted **
H5	Task-Relevant Information (IAT)	Impulse Buying (CPI)	0.068	1.406	0.168	Rejected
H6	Economic Well-being (BE)	Impulse Buying (CPI)	0.091	2.060	0.035	Accepted **
H7	Family Influence (IF)	Impulse Buying (CPI)	0.318	6.654	0.000	Accepted *
H8	Time Availability (DT)	Impulse Buying (CPI)	0.042	1.019	0.313	Rejected
H9	Credit Card Use (UTC)	Impulse Buying (CPI)	0.181	4.593	0.000	Accepted *
H10	Hedonic Value (VH)	Impulse Buying (CPI)	0.321	6.097	0.000	Accepted *
H11	Utilitarian Value (VU)	Impulse Buying (CPI)	0.000	0.002	0.998	Rejected
H12	Trust (CON)	Impulse Buying (CPI)	0.037	0.654	0.516	Rejected
H13	COVID-19 Effect (EC)	Impulse Buying (CPI)	−0.100	2.394	0.017	Rejected ***

Source: * Accepted at significance level p < 0.01. ** Accepted at significance level p < 0.05. *** Rejected as a positive relationship was proposed, but the result was negative.

presents this analysis, where Beta represents the path coefficients, indicating how much a change in the independent variable affects the dependent variable. The significance level can be determined through p-values or t-values.

Path coefficients, p-values, and t-statistics were employed to evaluate hypothesis acceptance or rejection, as illustrated in Figure 1 and Table 9. Path coefficient magnitudes indicate the strength of relationships between constructs, with values approaching +1 representing strong positive associations. Among the thirteen hypotheses examined, eight received support while five were rejected.

The following hypotheses were accepted based on significant statistical evidence. H1 received support, demonstrating that Visual Appeal (AT) positively influences Hedonic Value (VH) (β = 0.393, p < 0.001, t = 7.270). H2 was confirmed, showing that Visual Appeal (AT) positively affects Utilitarian Value (VU) (β = 0.340, p < 0.001, t = 5.890). H3 gained acceptance, indicating that Portability (P) positively impacts Hedonic Value (VH) (β = 0.122, p < 0.001, t = 2.121). H4 was supported, revealing that Portability (P) positively influences Utilitarian Value (VU) (β = 0.281, p = 0.031, t = 4.850).

Regarding direct effects on impulse buying behavior, several hypotheses received empirical support. H6 was accepted, confirming that Economic Well-being (BE) positively affects Impulse Buying (CPI) (β = 0.091, p = 0.035, t = 2.060). H7 gained support, demonstrating that Family Influence (IF) positively impacts Impulse Buying (CPI) (β = 0.318, p < 0.001, t = 6.654). H9 was confirmed, showing that Credit Card Use (UTC) has a positive influence on impulse buying (CPI) (β = 0.181, p < 0.001, t = 4.593). H10 received acceptance, indicating that Hedonic Value (VH) positively affects Impulse Buying (CPI) (β = 0.321, p < 0.001, t = 6.097).

Conversely, five hypotheses lacked statistical support. H5 was rejected, as Task-Relevant Information (IAT) did not demonstrate a significant positive impact on Impulse Buying (CPI) (β = 0.068, p = 0.168, t = 1.406). H8 was not supported, indicating that Time Availability (DT) does not significantly influence Impulse Buying (CPI) (β = 0.042, p = 0.313, t = 1.019). H11 was rejected due to the absence of a significant relationship between Utilitarian Value (VU) and Impulse Buying (CPI) (β = 0.000, p = 0.998, t = 0.002). H12 lacked support, as Trust (CON) did not significantly impact Impulse Buying (CPI) (β = 0.037, p = 0.516, t = 0.654). Finally, H13 was rejected because COVID-19 Effect (EC) demonstrated a significant negative rather than positive relationship with Impulse Buying (CPI) (β = −0.100, p = 0.017, t = 2.394).

These results reveal that mobile application characteristics (visual appeal and portability) significantly enhance hedonic and utilitarian values. Personal factors such as economic well-being, family influence, and credit card usage directly affect impulse buying behavior, whereas time availability demonstrates no significant influence. While hedonic value substantially impacts impulse buying, utilitarian value exhibits no measurable effect. Additionally, contrary to theoretical expectations, the COVID-19 effect was found to display a negative association with impulse buying behavior in the mobile commerce context.

5. Discussion and Conclusions

5.1. General Discussion

The main objective of this research was to contribute to the understanding of impulse buying in mobile commerce and identify the variables that influence consumers’ impulse buying behavior in Chile. The study found significant relationships between various variables in the research model, supporting eight of thirteen hypotheses.

The findings demonstrate that mobile application factors—specifically visual appeal and portability—significantly influence hedonic and utilitarian values. Visual appeal showed a strong positive impact on hedonic value (β = 0.393, p < 0.01) and utilitarian value (β = 0.340, p < 0.01), supporting the findings of Zheng et al. [39] who identified that visual elements in mobile applications influence shopping value perceptions. Similarly, portability positively influenced hedonic value (β = 0.122, p < 0.01) and utilitarian value (β = 0.281, p < 0.05), confirming that the ability to access mobile shopping anywhere increases exposure to stimuli, thereby enhancing both hedonic and utilitarian motivations [39].

However, task-relevant information did not significantly impact impulse buying (β = 0.068, p = 0.168), contradicting previous findings by Parboteeah et al. [26] who suggested that effective task-relevant cues could considerably increase the likelihood and magnitude of online impulse purchases. This discrepancy may suggest that, in the mobile commerce context, consumers rely less on detailed information when making impulse decisions than in desktop e-commerce environments. The non-significant effect of task-relevant information (H5) suggests that in mobile commerce, consumers may prioritize hedonic triggers over rational, task-oriented cues when making impulsive decisions. Prior research has shown that detailed product information and task-relevant content play a stronger role in desktop or traditional online shopping, where decision-making involves more deliberation [26,89]. However, in m-commerce environments, simplified interfaces, visual appeal, and emotional gratification dominate the decision process, reducing the influence of informational content. Zheng et al. [39] further demonstrated that hedonic browsing exerts a significantly stronger effect on impulse buying than utilitarian browsing, supporting the idea that emotional and experiential elements outweigh task-relevant information in mobile shopping environments.

The lack of significance of time availability (H8) challenges prior research suggesting that time pressure can increase consumers’ likelihood of making impulsive purchases. For example, Liu et al. [58] found that time pressure significantly amplifies impulsive buying behavior, especially for hedonic consumption, by reducing consumers’ capacity for deliberation. Similarly, Sun et al. [90] demonstrated that time constraints intensify emotional responses and perceived value, thereby fostering impulse buying tendencies. However, in the mobile commerce context, our results indicate that the design of mobile shopping applications—characterized by simplified navigation and one-click purchasing—neutralizes the role of time availability. In this environment, impulsive purchases can occur regardless of perceived temporal resources, suggesting that structural efficiency in m-commerce platforms substitutes for the moderating role of time pressure observed in other retail settings.

Among personal factors, economic well-being (β = 0.091, p < 0.05), family influence (β = 0.318, p < 0.01), and credit card use (β = 0.181, p < 0.01) all demonstrated significant positive relationships with impulse buying. These findings align with Badgaiyan and Verma [45], who found that consumers are more willing to spend impulsively when they have good economic well-being, and Baker et al. [51], who affirmed that family influences an individual’s impulsive buying behaviors. Credit card use showed a significant positive effect, supporting previous research suggesting that the painless spending sensation associated with credit cards encourages impulse purchases.

Contrary to expectations, time availability did not significantly impact impulse buying (β = 0.042, p = 0.313), which contradicts previous studies suggesting that more available time increases the likelihood of impulse purchases [54,55]. This unexpected finding might indicate that in the mobile commerce context, impulse purchases can occur quickly regardless of perceived time availability, possibly due to the streamlined nature of mobile shopping apps.

Regarding value factors, hedonic value showed a strong positive impact on impulse buying (β = 0.321, p < 0.01), confirming Kim and Eastin’s [61] assertion that hedonic value perception is an important antecedent to impulse buying behavior. Surprisingly, utilitarian value showed no significant effect on impulse buying (β = 0.000, p = 0.998), suggesting that functional and rational aspects of mobile shopping may not trigger impulsive behavior.

Trust did not demonstrate a significant relationship with impulse buying (β = 0.037, p = 0.516), contradicting the findings of Kauffman et al. [65]. This finding suggests that consumers may not consider trust a key indicator when making impulsive purchases in mobile commerce, possibly because established mobile platforms have already achieved a baseline level of trust.

Finally, contrary to the hypothesized positive relationship, the COVID-19 effect significantly negatively impacted impulse buying (β = −0.100, p < 0.05). This contradicts Ahmed et al. [91], who found that the pandemic positively affected impulse buying. This discrepancy might be explained by the fact that this study examined various products, not just essential goods, which might not necessarily experience positive impacts during the pandemic.

Reconciling Contradictory COVID-19 Effects

Previous studies have generally reported a positive relationship between COVID-19 and impulse buying.

Past studies [67,69,70], emphasizing panic buying and increased online shopping during the early pandemic phase. However, our results revealed a negative effect (β = −0.100, p < 0.05), highlighting that, in the Chilean context, pandemic-induced uncertainty and financial restrictions outweighed consumers’ tendency to engage in impulsive consumption.

One possible explanation is that, although the pandemic increased exposure to digital channels, it also generated greater economic and financial uncertainty, which may have led consumers to be more cautious with their spending on non-essential goods. This aligns with studies showing that perceived risk and the need for financial control during health crises reduce the propensity for impulsive spending [70]. In this sense, the pandemic not only functioned as a catalyst for e-commerce but also as a restrictive context for impulsive consumption. This finding appears to contradict studies such as Ahmed et al. [91] and Iyer et al. [28], which reported positive effects. However, this discrepancy can be explained by critical moderating factors that have not been adequately considered in previous research.

The most critical moderator is the nature of the products studied. Studies reporting positive effects of COVID-19 predominantly focused on essential products, personal protection items, and daily necessities where what appears as “impulse” actually reflects rational stockpiling behavior disguised as impulsivity [26]. In contrast, our study covers diverse categories dominated by non-essential products, particularly clothing and accessories, which represent genuinely hedonistic purchases. During health and economic crises, consumers redirect their limited resources from hedonic products to basic needs, explaining the observed negative effect. This differentiation suggests that COVID-19 does not have a uniform impact on impulsive behavior, but rather acts as a differential moderator depending on the consumer’s hierarchy of needs.

The temporal phase of data collection constitutes a second crucial moderator. The studies that reported positive effects collected data during the initial phase of the pandemic (March–June 2020), characterized by panic behavior and massive stockpiling. Our study was conducted during the adaptation phase (September–December 2020), when consumers had adjusted their behavior patterns to the new economic and health reality. During this later phase, prolonged economic uncertainty and mobility restrictions acted as inhibitors of hedonic impulsive behavior. In contrast, the initial phase was characterized by reactive hoarding responses that superficially appeared impulsive but actually reflected a rational, preventive logic.

Chile’s specific socioeconomic context during the pandemic provides a third explanatory moderator. Chile experienced severe economic contractions, with unemployment rates reaching 13.1% in July 2020, significantly higher than in countries where previous studies were conducted. In contexts of severe economic restriction, consumers adopt resource conservation strategies that prioritize basic needs over impulsive hedonic gratification. Additionally, strict lockdown measures in Chile limited exposure to impulse buying stimuli and reduced opportunities for social consumption, factors that traditionally facilitate impulsive behavior. This combination of economic pressure and social restrictions created an environment inhibiting impulsive behavior, contrasting with more economically stable contexts where previous studies were conducted.

These findings suggest an integrative theoretical model where COVID-19 acts as a contextual variable that differentially moderates impulsive behavior according to a hierarchy of products and needs. At the base of this hierarchy, essential products experience positive effects due to preemptive stockpiling behavior. At the intermediate level, convenience products show neutral or slightly negative effects. At the top level, purely hedonic products (such as those predominant in our study) experience significant adverse effects due to the reallocation of resources toward basic needs. This model reconciles the apparent contradictions in the literature and provides a predictive framework for understanding the impact of crises on consumption behavior according to the product, temporal, and socioeconomic context (

Table 10. Comparison of COVID-19 and Impulse Buying Studies.

Research	Context	Time	Products	COVID-19 Effects	Results
Ahmed et al. [91]	USA	USA March–June 2020	Essential/PPE	Positive (+)	Reactive stockpiling
Iyer et al. [28]	Multi-country	March–May 2020	Miscellaneous	Positive (+)	Initial panic phase
Present research	Chile	September–December 2020	Non-essential	Negative (−)	Economic restriction

).

5.2. Theoretical and Practical Implications

5.2.1. Theoretical Implications

This study contributes theoretically through three explicit refinements of the traditional S-O-R model for mobile contexts. First, it redefines stimuli from physical environmental elements to specific digital factors: visual appeal as an interactive interface (β = 0.393), portability as temporal-spatial ubiquity (β = 0.122), and technological enablers of payment (β = 0.181). Second, it simplifies organism processing from the original tripartite model (pleasure-arousal-dominance) to a hedonic-dominant model where only hedonic value (β = 0.321) influences impulsive responses, eliminating utilitarian considerations (β = 0.000). Third, it specifies mobile impulsive responses characterized by technological immediacy and greater social influence (β = 0.318) than cognitive influence.

The findings resolve three fundamental inconsistencies in the literature. The absence of an effect of utilitarian value contradicts Kim and Eastin [61], who explained it by the fact that mobile impulsive decisions are purely hedonic, not functional. The lack of significance of time availability contradicts Bell et al. [54], reflects that in m-commerce, impulsive decisions occur instantaneously, regardless of perceived time. The negative effect of COVID-19 (β = −0.100) contradicts Ahmed et al. [91]. Still, it is theoretically reconciled through moderation by product type: positive effects for essential products (as observed in previous studies) versus negative effects for non-essential hedonic products (as observed in our study).

This research proposes a hierarchical model where COVID-19 differentially moderates impulsive behavior: essential products (positive effect due to stockpiling), convenience products (neutral effect), and hedonic products (negative effect due to resource reallocation). This model reconciles apparent contradictions and provides a predictive framework for crisis effects based on product, temporal, and socioeconomic context.

5.2.2. Practical Implications

The highest coefficient for visual appeal (β = 0.393) requires specific implementations: warm color palettes (reds/oranges) for impulsive products, micro-animations of 300–500 ms duration, tactile elements of at least 44 pt, and a flat visual architecture with a 3:2:1 hierarchy for titles:subtitles:text. Portability (β = 0.281) is optimized through intelligent geolocation, contextual notifications, offline mode, and cross-device synchronization. Payment enablers (β = 0.181) require one-click integration (Apple Pay, Google Pay), single-screen checkout, and buy-now-pay-later options.

The strongest effect (family influence β = 0.318) requires specific features, including native sharing with enriched previews via WhatsApp, shared family wishlists with discount notifications, group buying functionality, and social proof elements with demographic filters (“parents recommend”). Implement family points systems and group savings challenges.

COVID-19 findings indicate the need for adaptive systems that modify strategies according to the economic context: during a crisis, emphasize value and need over hedonic drive; during recovery, gradually reactivate impulse stimulation elements. Develop automatic economic context detection capabilities to adjust messaging and product presentation.

Prioritize investments based on effect sizes: (1) Visual appeal (β = 0.393): 40% of the UX/UI budget; (2) Family influence (β = 0.318): 30% in social features; (3) Hedonic value (β = 0.321): 20% in gamification and experiential elements; (4) Payment enablers (β = 0.181): 10% in checkout optimization. This allocation maximizes ROI based on empirical evidence.

5.3. Limitations and Future Research Directions

This study has several limitations that should be acknowledged. First, there were challenges related to the measurement instrument. The survey was lengthy and not presented in an engaging format, which may have affected participation rates and response quality. Additionally, the measurement scales were originally in English and had to be translated and adapted, potentially causing interpretation issues.

Second, the geographical distribution of respondents was uneven, with 71.8% from northern Chile, 22% from central Chile, and only 6.2% from southern Chile. This imbalance means the results may not fully represent the entire Chilean population, as regional differences in consumer behavior exist.

Third, the study was conducted during the COVID-19 pandemic, which created unique circumstances for data collection and may have influenced consumer behavior in ways that would not be present under normal conditions. The online-only data collection method, necessitated by the pandemic, may also have introduced biases.

Regarding the COVID-19 effect variable, the analysis is limited as a direct positive relationship was proposed, but a direct negative relationship was found. This might be because the study examined various product categories, not just essential goods, which the pandemic may have affected differently.

Future research should address these limitations by expanding the study to achieve more balanced geographical representation across Chile. The research model could be enhanced by exploring new relationships between existing variables and incorporating additional antecedents of impulse buying in mobile commerce. The COVID-19 effect could be examined as a moderating variable rather than a direct predictor, potentially yielding more nuanced insights.

Further studies could also investigate cross-cultural differences in mobile commerce impulse buying, comparing Chile with other Latin American countries or regions globally. Longitudinal research would be valuable to understand how impulse buying patterns in mobile commerce evolve, particularly as the effects of the pandemic diminish.