Group-level effects of facilitating conditions on individual acceptance of information systems

Abstract

Much of the research effort in the area of technology acceptance has been directed to investigating the effects of various variables operating at the individual-level without considering the conjoint effects of group-level variables on individual acceptance. The present research addresses this issue by proposing a group-level variable, organizational facilitating conditions, and examining its effects on the unified theory of acceptance and use of technology model, a widely used individual user acceptance model. Two field studies were conducted to explore the multilevel nature of technology acceptance. In the first study, we refined the construct of facilitating conditions and developed a new measure of facilitating conditions to explicitly add the organizational facilitating conditions dimension as well as to augment the existing measure. Subsequent testing of the measure confirmed the multilevel nature of the construct. In the second study, we examined the effects of the organizational facilitating conditions on individual acceptance behaviors by utilizing the hierarchical linear modeling approach. The results indicate that the two constructs, individual facilitating conditions and organizational facilitating conditions, are distinct and that, compared to individual facilitating conditions, the organizational facilitating conditions as a group-level variable explain a larger amount of variance in individual acceptance behavior. The resulting model offers a multilevel perspective to the technology acceptance research area while the study results provide an augmented way to evaluate facilitating conditions with a prescriptive guidance to managers.

Appendix

We provide additional analyses to further establish that organizational facilitating conditions (OFC) and individual facilitating conditions (IFC) should be treated as separate constructs. We built five models for a model comparison. The first model has all final items of OFC and IFC as one comprehensive facilitating conditions construct in UTAUT. The second model has the three final items of OFC, while the third model has the three final items of IFC for representing the FC construct in UTAUT. The fourth model has the four original items of the facilitating conditions construct in UTAUT. The fifth model has all final items of OFC and IFC and two separate constructs (OFC and IFC) for representing the FC construct in UTAUT. The five models are based upon the original UTAUT with various combinations of facilitating conditions (combined OFC and IFC as one construct, OFC only, IFC only, the original UTAUT facilitating condition, OFC and IFC as two separated constructs) and were tested using LISREL 8.8. It should be noted that all other UTAUT constructs except facilitating conditions were measured with the original items from UTAUT. Model-data fit statistics for the five models are shown in Table 17. In addition, Table 17 reports total explained variances for the five alternative models.

Table 17 Model fit and explained variances comparisons

The model fit statistics for the five models were reasonable. Although the RMSEA values were slightly outside the recommended cut-off value, these values are acceptable. According to the three fit indicators, Model 2 and Model 5 showed better model fit than others. However, the differences in model fit among the five models were found very small (e.g., biggest difference was only at the level of 0.02). So, although it can be argued that Model 2 and 5 are a bit better representations of the data, all five models are very similar. In terms of total explained variances of two dependent variables (behavioral intention and use behavior), Model 5 showed the highest value while increasing the total explained variances by 5% from Model 4 (the original UTAUT model). That is, Model 5 with OFC and IFC as two separated constructs is the best when evaluated by explanatory and predictive power.

Additionally, we conduct more simplified model comparisons, where we have facilitating conditions as only predictor (without other independent variables) and behavioral intention as the only dependent variable. In terms of facilitating conditions, we used the same five variations: Model 1—combined OFC and IFC items as one facilitating conditions construct; Model 2—OFC only; Model 3—IFC only; Model 4—the original UTAUT facilitating condition; and Model 5—OFC and IFC as two separated constructs. The five simplified models were tested using LISREL 8.8, and the results are shown in Table 18. Model 3 showed the best model fit, but Model 5 showed highest explained variances while OFC and IFC factors were all significant and explained 38% of behavioral intentional (BI) variance. Standardized parameter estimates for the OFC-BI and IFC-BI links were 0.24 and 0.40 in Model 5.

Table 18 Simplified model comparison with behavioral intention as dependent variable

Table 19 displays the results from the additional five simplified model tests with satisfaction as dependent variable. Satisfaction is often used as an important dependent variable in IT acceptance studies [6], and we believe it is worthy to see how the variations of facilitating conditions work with satisfaction. As the results indicate, Model 2 showed the best model fit, while Model 1 and 5 showed highest explained variances of satisfaction. Standardized parameter estimates for the OFC-BI and IFC-BI links were 0.27 and 0.49 in Model 5.

Table 19 Simplified model comparison with satisfaction as dependent variable

Overall, the results from the three rounds of model comparisons indicate that Model 5 with OFC and IFC as two separated constructs showed better fit indices and also explained variances more than other models. This suggests that conceptualizing facilitation conditions as two separate constructs (OFC and IFC) would provide better representations of the reality, as well as enhance our explanatory and predictive power in the IT acceptance context.