CB-SEM compared with PLS-SEM

A Primer on Partial Least Squares
Structural Equation Modeling (PLS-SEM)

CB-SEM compared with PLS-SEM

UTAUT Model

Reference:

Baishya, K., & Samalia, H. V. (2020). Extending Unified Theory of Acceptance and Use of Technology with Perceived Monetary Value for Smartphone Adoption at the Bottom of the Pyramid. International Journal of Information Management, 51, 102036.

Sample Size:

n= 590

Legend:

Performance expectancy (PE), Effort expectancy (EE), Subjective influence (SI), Perceived monetary value (PMV), Facilitating conditions (FC), Behavioral intention (BI), Use behavior (UB)

Standardized coefficients:

Measurement model relations		CBSEM results using the lavaan package					PLS results using SmartPLS
PE =~		ML	GLS	ULS	WLS	DWLS	PLS	PLSc	Sumscores
	PE1	0.888	0.897	0.889	0.920	0.888	0.908	0.844	0.911
	PE2	0.853	0.865	0.875	0.900	0.876	0.896	0.885	0.892
	PE3	0.840	0.867	0.812	0.908	0.815	0.875	0.752	0.886
	PE4	0.851	0.863	0.838	0.921	0.836	0.902	0.941	0.894
EE =~
	EE1	0.847	0.889	0.879	0.893	0.878	0.875	0.852	0.874
	EE2	0.858	0.885	0.850	0.888	0.846	0.883	0.805	0.885
	EE3	0.830	0.834	0.808	0.838	0.810	0.871	0.835	0.872
	EE4	0.833	0.826	0.826	0.878	0.825	0.872	0.857	0.870
	EE5	0.829	0.830	0.818	0.845	0.818	0.868	0.847	0.868
SI =~
	SI1	0.885	0.937	0.907	0.911	0.903	0.928	0.970	0.918
	SI2	0.863	0.897	0.890	0.943	0.887	0.904	0.797	0.911
	SI3	0.855	0.858	0.794	0.911	0.793	0.910	0.835	0.914
FC =~
	FC1	0.822	0.812	0.841	0.864	0.836	0.882	0.832	0.857
	FC2	0.902	0.900	0.891	0.928	0.890	0.916	0.867	0.891
	FC3	0.383	0.416	0.441	0.393	0.439	0.514	0.470	0.612
	FC4	0.900	0.909	0.858	0.903	0.858	0.916	0.859	0.891
PMV =~
	PMV1	0.910	0.896	0.905	0.882	0.910	0.945	0.877	0.947
	PMV2	0.873	0.889	0.860	0.985	0.858	0.949	0.906	0.947
BI =~
	BI1	0.853	0.845	0.879	0.849	0.880	0.912	0.891	0.910
	BI2	0.897	0.879	0.896	0.855	0.899	0.927	0.899	0.926
	BI3	0.884	0.892	0.861	0.894	0.859	0.922	0.842	0.925
UB =~
	UB1	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Structural model relations		CBSEM results using the lavaan package					PLS results using SmartPLS
BI ~		ML	GLS	ULS	WLS	DWLS	PLS	PLSc	Sumscores
	PE	0.242	0.236	0.247	0.160	0.248	0.231	0.242	0.228
	EE	0.176	0.234	0.193	0.210	0.196	0.179	0.180	0.179
	SI	0.222	0.177	0.220	0.221	0.225	0.201	0.213	0.200
	PMV	0.226	0.276	0.242	0.164	0.235	0.214	0.233	0.214
UB ~
	FC	0.386	0.370	0.401	0.184	0.404	0.369	0.389	0.365
	BI	0.321	0.327	0.330	0.401	0.329	0.302	0.298	0.304

We analyzed the significance of all solutions except of those with negative variances. For CBSEM we used the p-value, for PLS the 95% percentile bootstrap confidence intervals. All results are significant on the 5% probability of error level.

TAM Model

Reference:

Anderson, C., Al-Gahtani, S.S. and Hubona, G.S. (2011), "The Value of TAM Antecedents in Global IS Development and Research", Journal of Organizational and End User Computing, Vol. 23 No. 1, pp. 18-37.

Sample Size:

n= 1,190

Legend:

Attitude (ATT), Behavioral intention to use (BI), Perceived ease of use (PEOU), Perceived usefulness (PU), Actual system use (USE)

Standardized coefficients:

Measurement model relations		CBSEM results using the lavaan package					PLS results using SmartPLS
BI =~		ML	GLS	ULS	WLS	DWLS	PLS	PLSc	Sumscores
	BI1	0.711	0.695	0.804	0.695	0.791	0.855	0.877	0.831
	BI2	0.771	0.753	0.683	0.753	0.628	0.831	0.591	0.867
	BI3	0.869	0.846	0.807	0.846	0.753	0.890	0.849	0.884
PEOU =~
	EOU1	0.69	0.625	0.695	0.625	0.694	0.765	0.696	0.777
	EOU2	0.778	0.734	0.766	0.734	0.764	0.827	0.780	0.823
	EOU3	0.837	0.802	0.842	0.802	0.842	0.863	0.830	0.853
	EOU4	0.754	0.707	0.759	0.707	0.762	0.809	0.792	0.801
	EOU5	0.723	0.635	0.724	0.635	0.717	0.789	0.681	0.801
PU =~
	USEF1	0.739	0.741	0.634	0.741	0.603	0.808	0.811	0.808
	USEF2	0.806	0.793	0.649	0.793	0.616	0.847	0.796	0.849
	USEF3	0.855	0.856	0.659	0.856	0.615	0.880	0.817	0.88
	USEF4	0.839	0.835	0.665	0.835	0.627	0.866	0.821	0.865
	USEF5	0.815	0.820	0.646	0.820	0.601	0.860	0.812	0.86
ATT =~
	ATT1	0.877	0.878	0.876	0.878	0.886	0.900	0.912	0.895
	ATT2	0.884	0.896	0.887	0.896	0.895	0.905	0.916	0.899
	ATT3	0.870	0.861	0.867	0.861	0.874	0.897	0.892	0.893
	ATT4	0.840	0.901	0.822	0.901	0.825	0.880	0.831	0.886
	ATT5	0.840	0.900	0.791	0.900	0.780	0.878	0.751	0.888
USE =~
	USE1	0.787	0.778	0.638	0.778	0.654	0.800	0.702	0.783
	USE2	0.729	0.720	0.602	0.720	0.589	0.761	0.641	0.756
	USE3	0.592	0.614	0.763	0.614	0.705	0.772	0.667	0.771
	USE4	0.458	0.468	0.557	0.468	0.551	0.664	0.581	0.689
Structural model relations:		CBSEM results using the lavaan package					PLS results using SmartPLS
BI ~		ML	GLS	ULS	WLS	DWLS	PLS	PLSc	Sumscores
	ATT	0.160	0.189	-0.102	0.543	-0.149	0.172	0.174	0.170
	PU	0.466	0.558	0.867	0.281	0.949	0.401	0.461	0.390
PU ~
	PEOU	0.509	0.577	0.748	0.611	0.757	0.445	0.500	0.444
ATT ~
	PEOU	0.285	0.333	0.153	0.615	0.036(n.s.)	0.253	0.272	0.250
	PU	0.264	0.196	0.400	0.212	0.522	0.267	0.276	0.268
USE ~
	ATT	0.288	0.347	0.354	0.761	0.311	0.271	0.316	0.276
	BI	0.189	0.081(n.s.)	0.219	-0.322	0.265	0.134	0.164	0.124

(n.s.) = nonsignificant

We analyzed the significance of all solutions except of those with negative variance. For CBSEM we used the p-value, for PLS the 95% percentile bootstrap confidence intervals. All results are significant on the 5% probability of error level. The only exceptions are the BI to USE relationship in the GLS results and the PEOU to ATT relationship in the DWLS results of CBSEM.