Unraveling the Influential Mechanisms of Social Commerce Overloads on User Disengagement: The Buffer Effect of Guanxi

Figures & data

Table 1 Related Studies on User Disengagement (UD) in the SC Context

Authors	Findings
O’Brien et al^Citation17	The work on UD is sparse, and UD is a means to address problematic technology use, often involving ceasing use or withdrawing time or energy of users.
Yan et al^Citation74	Employing ad blockers positively impacts quantity and variety of news consumption.
Vatavwala et al^Citation16	UD involves dormancy, termination, and acrimonious exit.
Matthes et al^Citation18	UD is the results of negative interactions with technologies, and involves pausing or ceasing apps or taking a break from devices.
Tripathi et al^Citation19	UD involves the more willful and active withdrawal of users, and consumer skepticism positively impacts advertising disengagement.
Chellasamy et al^Citation75	No studies have thoroughly investigated the construct, antecedents and outcomes of UD.

Figure 1 The screenshots of user disengagement.

Figure 2 Research model.

Table 2 Demographics Information (N = 406)

Characteristics	Frequency	Percentage (%)
Gender
Male	204	50.2
Female	202	49.8
Age
18–25	158	38.9
26–35	150	36.9
36–45	86	21.2
>46	12	3.0
Education
≤High school/professional high school	36	8.9
College	91	22.4
University degree	237	58.4
≥Graduate degree	42	10.3
Monthly disposable Incomes(RMB)
<1000	88	21.7
1001–3000	82	20.2
3001–5000	87	21.4
5001–7000	93	22.9
>7001	56	13.8
Usage experience with WeChat
<1 year	53	13.0
1–2 years	91	22.4
3–4 years	118	29.1
>4 years	144	35.5

Table 3 Confirmatory Factor Analysis Result of All the Constructs

Variable	Item	Loading	CR	AVE	CA
Information overload (IO)	IO1	0.781	0.843	0.641	0.842
	IO2	0.813
	IO3	0.808
Social overload (SO)	SO1	0.773	0.801	0.573	0.801
	SO2	0.770
	SO3	0.726
Communication overload (CO)	CO1	0.794	0.837	0.632	0.834
	CO2	0.855
	CO3	0.730
Inferences of manipulative intent (IMI)	IMI1	0.795	0.810	0.588	0.810
	IMI2	0.791
	IMI3	0.711
Compulsive perception (CP)	CP1	0.779	0.892	0.622	0.891
	CP2	0.837
	CP3	0.818
	CP4	0.727
	CP5	0.779
Reactance (PR)	PR1	0.712	0.845	0.522	0.845
	PR2	0.726
	PR3	0.735
	PR4	0.719
	PR5	0.721
Guanxi (GX)	GX1	0.833	0.957	0.710	0.957
	GX2	0.817
	GX3	0.832
	GX4	0.852
	GX5	0.822
	GX6	0.855
	GX7	0.861
	GX8	0.859
	GX9	0.853
Neglecting behavior (NE)	NE1	0.790	0.836	0.629	0.832
	NE2	0.828
	NE3	0.760
Blocking behavior (BL)	BL1	0.812	0.834	0.626	0.832
	BL2	0.812
	BL3	0.747

Abbreviations: CA, Cronbach’s α; CR, composite reliability; AVE, average variance extracted.

Table 4 Discriminant Validity of Measures Used in Measurement and Correlation Matrix

	Mean	1	2	3	4	5	6	7	8	9
IO	5.700	0.801
SO	5.463	0.632	0.757
CO	5.376	0.588	0.600	0.795
IMI	5.502	0.674	0.642	0.642	0.767
CP	5.551	0.656	0.660	0.630	0.684	0.789
PR	5.433	0.662	0.676	0.592	0.676	0.655	0.722
GX	2.969	−0.499	−0.523	−0.463	−0.522	−0.565	−0.566	0.843
NE	5.521	0.582	0.622	0.592	0.711	0.666	0.685	−0.501	0.793
BL	5.553	0.648	0.628	0.564	0.635	0.675	0.671	−0.584	0.603	0.791

Notes: The numbers on the diagonal row were the square root of Average Variance Extracted (AVE). Off diagonal values were the correlations between the constructs.

Abbreviations: IO, information overload; SO, social overload; CO, communication overload; IMI, inferences of manipulative intent; CP, compulsive perception; PR, reactance; GX, guanxi; NE, neglecting behavior; BL, blocking behavior.

Table 5 Hypothesis Test Results

Hypothesis	Path Relationship	Standardized Coefficient	Result
H1a	IO→IMI	0.367***	∏
H1b	IO→CP	0.292***	∏
H2a	SO→IMI	0.405***	∏
H2b	SO→CP	0.421***	∏
H3a	CO→IMI	0.235***	∏
H3b	CO→CP	0.217**	∏
H4a	IMI→PR	0.683***	∏
H4b	CP→PR	0.221**	∏
H6a	IMI→NE	0.493*	∏
H6b	IMI→BL	0.250***	∏
H8a	CP→NE	0.194**	∏
H8b	CP→BL	0.362***	∏
H10a	PR→NE	0.240*	∏
H10b	PR→BL	0.309**	∏

Notes: *p<0.05, **P<0.01, ***P<0.001, ∏: the hypothesis was supported.

Abbreviations: IO, information overload; SO, social overload; CO, communication overload; IMI, inferences of manipulative intent; CP, compulsive perception; PR, reactance; NE, neglecting behavior; BL, blocking behavior; NS, non-significant.

Figure 3 Structural model result (*p<0.05, **P<0.01, ***P<0.001, NS: non-significant).

Table 6 Testing of the Mediating Effects (SmartPLS Bootstrapping Approach)

Hypothesis	Mediation Path	Effect	T-values	P-values	Results
H5a	IO → IMI → PR	0.148	5.694	0.000	∏
H5b	SO → IMI → PR	0.109	4.242	0.000	∏
H5c	CO → IMI → PR	0.123	5.178	0.000	∏
H5d	IO → CP → PR	0.110	4.858	0.000	∏
H5e	SO → CP → PR	0.112	4.762	0.000	∏
H5f	CO → CP → PR	0.096	4.812	0.000	∏
H7a	IO → IMI → NE	0.122	5.294	0.000	∏
H7b	IO → IMI → BL	0.064	2.893	0.004	∏
H7c	SO → IMI → NE	0.090	4.083	0.000	∏
H7d	SO → IMI → BL	0.047	2.655	0.008	∏
H7e	CO → IMI → NE	0.101	4.646	0.000	∏
H7f	CO → IMI → BL	0.053	3.089	0.002	∏
H9a	IO → CP → NE	0.071	4.200	0.000	∏
H9b	IO → CP → BL	0.103	4.620	0.000	∏
H9c	SO → CP → NE	0.072	3.988	0.000	∏
H9d	SO → CP → BL	0.104	4.454	0.000	∏
H9e	CO → CP → NE	0.062	3.854	0.000	∏
H9f	CO → CP → BL	0.090	4.814	0.000	∏
H11a	IMI → PR → NE	0.127	5.564	0.000	∏
H11b	IMI → PR → BL	0.140	5.387	0.000	∏
H11c	CP → PR → NE	0.107	4.657	0.000	∏
H11d	CP → PR → BL	0.118	5.035	0.000	∏

Note: ∏, the hypothesis was supported.

Abbreviations: IO, information overload; SO, social overload; CO, communication overload; IMI, inferences of manipulative intent; CP, compulsive perception; PR, reactance; NE, neglecting behavior; BL, blocking behavior.

Table 7 Interaction Moderation Analysis

Hypotheses	Std. β	T-value	Bootstrap 5000 Times, Bias-Corrected, 95% CI	Effect	Results
H12a: IMI × GX → NE	−0.043*	−2.208	[−0.0047, −0.0815]	Negative effect	Supported
H12b: IMI × GX → BL	−0.056**	−2.620	[−0.0139, −0.0973]	Negative effect	Supported
H12c: CP × GX → NE	−0.012^ns	−0.536	[0.0314, −0.0550]	-	Unsupported
H12d: CP × GX → BL	−0.048*	−2.216	[−0.0054, −0.0912]	Negative effect	Supported

Notes: *p<0.05, **P<0.01, Std. β, standardized regression coefficient; Negative effect, a significant negative relationship.

Abbreviations: IMI, inferences of manipulative intent; CP, compulsive perception; NE, neglecting behavior; BL, blocking behavior; GX, guanxi; ns, not significant.

O’Brien HL, Roll I, Kampen A, Davoudi N. Rethinking (dis)engagement in human-computer interaction. Comput Hum Behav. 2022;128:107109. doi:10.1016/j.chb.2021.107109

 Web of Science ®Google Scholar

Yan S, Miller KM, Skiera B. How does the adoption of ad blockers affect news consumption? J Market Res. 2022;00222437221076160.

 Web of Science ®Google Scholar

Vatavwala S, Kumar B, Sharma A, Billore A, Sadh A. Customer disengagement in business-to-business markets: a framework for analysis. Ind Market Manag. 2022;105:114–130. doi:10.1016/j.indmarman.2022.05.018

 Web of Science ®Google Scholar

Matthes J, Karsay K, Hirsch M, Stevic A, Schmuck D. Reflective smartphone disengagement: conceptualization, measurement, and validation. Comput Hum Behav. 2022;128:107078. doi:10.1016/j.chb.2021.107078

 Web of Science ®Google Scholar

Tripathi S, Jain V, Pandey J, Merchant A, Ambika A. When consumers tune out advertising messages: development and validation of a scale to measure advertising disengagement. J Advert Res. 2022;62(1):3–17. doi:10.2501/JAR-2021-020

 Web of Science ®Google Scholar

Chellasamy A, Almeida SM, Prakasha G, Nagarathinam A, Rangasamy S. Social media disengagement among gen y and gen z: a quantitative investigation. Med Literacy Acad Res. 2022;5(5):230–252.

 Google Scholar