A Multilevel Approach to Examine Employees’ Loyal Use of ERP Systems in Organizations

Abstract

A successful enterprise resource planning (ERP) system ultimately requires loyal use—proactive, extended use and willingness to recommend such uses to others—by employees. Building on interactional psychology literature and situational strength theory, we emphasize the importance of psychological commitment, in addition to behavioral manifestation, in a multilevel model of loyal use. Our empirical test of the model uses data from 485 employees and 166 information system professionals in 47 large Taiwanese organizations. Individual-level analyses suggest that perceived benefits and workload partially mediate the effects of perceived information quality (IQ) and system quality (SQ) on loyal use. Cross-level analyses show that IQ at the organizational level alleviates the negative effect of an employee’s perceived workload on loyal use; organization-level SQ and service-oriented organizational citizenship behaviors (SOCBs) of internal information systems staff reduce the influence of employees’ perceived benefits. Overall, our findings suggest that IQ, SQ, and SOCBs at the organizational level influence employees’ loyal use in ways different from their effects at the individual level, and seem to affect individuals’ cost–benefit analyses. This study contributes to extant literature by considering the SOCBs of the internal information systems group that have been overlooked by most prior research. Our findings offer insights for managers who should find ways to create positive, salient, shared views of IQ, SQ, and SOCBs in the organization to nourish and foster employees’ loyal use of an ERP system, including clearly demonstrating the system’s utilities and devising viable means to reduce the associated workload.

Key words and phrases:

• information quality
• loyal use
• multilevel analysis
• perceived benefits
• perceived workload
• service-oriented organizational citizenship behaviors
• system quality

Acknowledgment

This research is partially supported by Ministry of Science and Technology in Taiwan (Grant Nos. 102-2410-H-004-197-MY3, 102-2410-H-007-050-MY3, and 95-2416-H-008-039).

Supplemental File

Supplemental data for this article can be accessed on the publisher’s website at http://dx.doi.org/10.1080/07421222.2015.1138373

Notes

1. Acknowledging the subtle differences among continuance of use, continued use, and postadoption use, we consider these usage behaviors mostly interchangeable, rather than attempting to differentiate them explicitly.

2. The original items focus on two indicators of loyal customers: repurchase and recommendation. To fit our study context, we adapted one item to measure an employee’s use of more system features and functionalities in many work tasks, which mimics a consumer’s behavior of purchasing more from a store [80]. The psychological commitment item refers to an employee’s willingness to recommend such system uses to others in the organization. Our use of a single item to measure each respective subdimension of loyal use is in line with Bergkvist and Rossiter [11], who suggest that single-item measures are sufficient if, in the minds of respondents, the object of the construct is concrete and singular, and the attribute of the construct is concrete (i.e., easily and uniformly imagined). Previous research, including [77], similarly employs single items to measure the different subdimensions of a formative construct.

3. An experienced IS researcher, fluent in both English and Chinese, translated the original question items to Chinese. Another IS researcher, also fluent in both languages, translated these items back to English. A panel of three experienced IS researchers and five senior IS managers, knowledgeable about ERP systems, assessed each back-translated item to ensure that the semantics of the original item had been preserved.

4. We assessed nonresponse bias by comparing respondents who completed the survey in the initial response window with those who needed additional time to do so. Chi-square tests and Student’s t-tests indicated no significant between-group differences in gender, age, tenure with the organization, previous experience using the ERP system, or interactions with internal IS staff (largest χ² = 2.24; p > .10). The between-group differences in the mean of each item were not significant either (largest t = .18; p > .10). We followed the same procedure to assess nonresponse bias in the IS staff members’ assessments and noted no significant differences between early and late respondents (largest χ² = .96; p > .10; largest t = 1.65; p > .10). Thus, nonresponse bias did not seem to pose a serious threat.

5. Because loyal use is operationalized as a formative construct, we first performed partial least squares structural equation modeling to estimate the value of loyal use. We used bootstrapping with 200 resamples to test the significance of the respective path coefficients. The weights for the two items measuring loyal use (LU) were .601 and .437. We therefore modeled loyal use as (.601 × LU-1 + .437 × LU-2) in all subsequent analyses.

6. Following Hox [45], we analyzed the effect size of the individual-level model (Model 4 of Table 2) and the cross-level full model (Model 2 of Table 3), respectively. For the individual-level model’s effect size, we calculated the portion of variance (r²) in loyal use explained by the model in relation to the within-group variance of the null model. Using the formula, $({\sigma }_{\mathrm{n}\mathrm{u}\mathrm{l}\mathrm{l}}^2-{\sigma }_{\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{o}\mathrm{m}}^2)/{\sigma }_{\mathrm{n}\mathrm{u}\mathrm{l}\mathrm{l}}^2$, we obtained a value of .61; that is (.193 – .075)/.193 = .61. We also analyzed the effect size of the full model by calculating the portion of variance (r²) in loyal use explained by the full model in relation to the between-group variance of the null model. By applying the formula, $\left({\tau }_{\mathrm{n}\mathrm{u}\mathrm{l}\mathrm{l}}^2-{\tau }_{\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}}^2\right)/{\tau }_{\mathrm{n}\mathrm{u}\mathrm{l}\mathrm{l}}^2$, we noted that the effect size of the full model was .79; that is, (.028 – .006)/.028 = .79. Regarding the power of testing Model 4 in Table 2 and Model 2 in Table 3, our results showed that both values exceeded .90, with a sample size of 485. Together, these results suggest acceptable validity of the individual-level model (Model 4 in Table 2) and the full model (Model 2 in Table 3).

Additional information

Notes on contributors

HsiuJu Rebecca Yen

HsiuJu Rebecca Yen is a professor and director of the Institute of Service Science at the National Tsing Hua University, Taiwan. She received her Ph.D. in psychology from Rutgers, the State University of New Jersey. Her current research interests include information technology infusion in service management, service innovation, and service-oriented organizational citizenship behaviors. She has published in Research Policy, Journal of the Academy of Marketing Science, Decision Support Systems, Information and Management, International Journal of Electronic Commerce, IEEE Transactions on Engineering Management, and other venues.

Paul Jen-Hwa Hu

Paul Jen-Hwa Hu is David Eccles Chair Professor at the David Eccles School of Business, the University of Utah. He received his Ph.D. in management information systems from the University of Arizona. His current research interests include the use of information technology in health care, electronic commerce, digital government, knowledge management, and technology-mediated learning. He has published in Journal of Management Information Systems, Information Systems Research, Journal of the Association for Information Systems, Decision Sciences, Journal of Information Systems, Decision Support Systems, and various IEEE and ACM journals and transactions, among other venues.

Sheila Hsuan-Yu Hsu

Sheila Hsuan-Yu Hsu is an assistant professor in the Department of Information Management at the Tatung University, Taiwan. She holds a Ph.D. from the National Central University, Taiwan. Her research interests include virtual communities, Internet marketing, and electronic commerce. She has published in International Journal of Electronic Commerce, Decision Support Systems, and International Journal of Information Management.

Eldon Y. Li

Eldon Y. Lı (corresponding: [email protected]) is University Chair Professor and chairperson of the Department of Management Information Systems at the National Chengchi University, Taiwan. He was previously dean of the College of Informatics at Yuan Ze University and director of the Graduate Institute of Information Management at National Chung Cheng University in Taiwan, as well as professor and coordinator of the MIS Program at the College of Business, California Polytechnic State University, San Luis Obispo. He received his Ph.D. from Texas Tech University. He has published over 200 papers on various topics related to innovation and technology management, human factors in information technology (IT), strategic IT planning, software quality management, and information systems management.