4.1 Psychometric Properties of the Measurement Instrument
We first assessed the analysis of the measurement model, which does not pose any issue regarding factor loading (all values are equal to 0.797 or higher (Appendix 3), reliability, and convergence validity. All Cronbach’s alpha, rho_A, and composite reliability values overcome the lower limit of 0.7 (see Table 2), therefore, all constructs are reliable. Regarding AVE, this criterion has been met as the values overcome the recommended lower limit (see Table 2).
Table 2. Construct Reliability and Convergent Validity
|
Mean
|
Standard
Deviation
|
Cronbach's Alpha
|
rho_A
|
Composite Reliability
|
Average Variance Extracted (AVE)
|
Perceived Vividness (PV)
|
5.4
|
1.509
|
0.766
|
0.882
|
0.891
|
0.803
|
Perceived Interactivity
(PI)
|
4.8
|
1.786
|
0.870
|
0.887
|
0.921
|
0.796
|
Spatial Presence
(SPAT)
|
4.5
|
1.826
|
0.966
|
0.968
|
0.971
|
0.806
|
Social Presence (SOC)
|
4.2
|
1.920
|
0.928
|
0.932
|
0.949
|
0.822
|
Inspired By (CI 1-5)
|
5.0
|
1.666
|
0.888
|
0.891
|
0.918
|
0.692
|
Inspired To (CI 1-10)
|
5.3
|
1.768
|
0.972
|
0.974
|
0.979
|
0.901
|
Donation Intention (DI)
|
4.0
|
1.795
|
0.941
|
0.942
|
0.962
|
0.895
|
To establish discriminant validity the Fornell-Larcker criterion was met (see Table 3), since the square roots values of AVE for each construct are all higher them the correlations with other constructs. We also verified that the HTMT values are all lower than the threshold of 0.9 (see Table 3). Finally, all the VIF (variance inflation factor) scores are below the threshold of 5 (Hair et al. 2014), showing that there are no inner collinearity issues (see Table 4). When comparing scores of path coefficients, t-values, and R2 of the original PLS model and the marker approach (Appendix 4), the differences cannot be considered significant; meaning that there is no potential impact of CMV on the study’s results (Chin et al. 2013).
Table 3. Discriminant Validity
Fornell-Larcker Criterion
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
1.Donation Intention
|
0.946
|
|
|
|
|
|
|
2.Inspired By
|
0.653
|
0.832
|
|
|
|
|
|
3.Inspired To
|
0.776
|
0.718
|
0.949
|
|
|
|
|
4.Perceived Interactivity
|
0.452
|
0.634
|
0.521
|
0.892
|
|
|
|
5.Perceived Vividness
|
0.352
|
0.457
|
0.389
|
0.635
|
0.896
|
|
|
6.Social Presence
|
0.532
|
0.669
|
0.580
|
0.748
|
0.476
|
0.907
|
|
7.Spatial Presence
|
0.566
|
0.663
|
0.580
|
0.865
|
0.641
|
0.856
|
0.898
|
Heterotrait-Monotrait Ratio (HTMT)
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
1.Donation Intention
|
|
|
|
|
|
|
|
2.Inspired By
|
0.714
|
|
|
|
|
|
|
3.Inspired To
|
0.811
|
0.769
|
|
|
|
|
|
4.Perceived Interactivity
|
0.498
|
0.719
|
0.570
|
|
|
|
|
5.Perceived Vividness
|
0.394
|
0.530
|
0.426
|
0.757
|
|
|
|
6.Social Presence
|
0.568
|
0.734
|
0.611
|
0.825
|
0.534
|
|
|
7.Spatial Presence
|
0.592
|
0.711
|
0.597
|
0.840
|
0.716
|
0.800
|
|
Note: In the Fornell-Larcker’s table the diagonal values in bold indicate the square root of average variances extracted (AVE). The values in the lower diagonal indicate inter-construct correlations
Table 4. Collinearity Assessment for Structural Model.
|
Donor Inspiration
|
Donation Intention
|
Perceived Interactivity
|
Perceived Vividness
|
Social Presence
|
Spacial Presence
|
Donor Inspiration
|
|
1.912
|
|
|
|
|
Donation Intention
|
|
|
|
|
|
|
Perceived Interactivity
|
|
|
|
|
1.677
|
1.677
|
Perceived Vividness
|
|
|
|
|
1.677
|
1.677
|
Social Presence
|
3.735
|
3.991
|
|
|
|
|
Spacial Presence
|
3.735
|
3.970
|
|
|
|
|
Note: Variance Inflation Factor (VIF) less than 5 (Hair et al., 2014).
4.2 Analysis of the Theoretical Model
The non-parametric bootstrapping procedure was conducted to evaluate the structural model relationships (5,000 sub-samples). Except for H1a, all hypotheses are supported by the results (Table 5). Regarding hypotheses H2a and H2b, the results show β = 0.748, t = 11.315, p < 0.001; and β = 0.767, t = 15.076, p < 0.001, respectively. Hence, the more the user perceives the immersive video as interactive, the more will the user experience both social and spatial presence. H1a is not supported by the results (β = 0.000, t = 0.004, p = 0.997), meaning that perceived vividness is not a good predictor of social presence. Contrariwise, H1b is supported (β = 0.153, t = 2.524, p < 0.05). Thus, the more the user perceives the immersive video as vivid, the more strongly will the user experience spatial presence. Further, H3 theorizes that social presence positively influences donor inspiration, and H4 proposes that spatial presence positively influences donor inspiration. Both hypotheses are supported by the results (β = 0.366, t = 2.953, p < 0.01; and β = 0.351, t = 2.881, p < 0.01, respectively), and thus, when the user experiences higher levels of social presence as well as higher levels of spatial presence, the level of donor inspiration increase. H5 posits that donor inspiration positively influences donation intention. This hypothesis is supported by the results (β = 0.739, t = 8.985, p < 0.001), meaning that donor inspiration is a good predictor of donation intention. All the direct effects can be found in Table 5.
When analyzing the direct effects and the specific indirect effects, it is possible to observe that the effect of social presence on donation intention is not significant (β = -0.093, t = 0.759, p = 0.448) and the same is verified in the effect of spatial presence on donation intention (β = 0.155, t = 1.185, p = 0.237). Yet, the specific indirect effects of donor inspiration are significant. Thus, donor inspiration act as a mediator in the relationships of Social presence → Donation intention, and Spatial presence → Donation intention.
Considering the Stone-Geiser’s Chi-Square (Q²), there is proof of relevance when Q² scores are above 0 for a specific reflective endogenous variable (Hair et al. 2011). The results indicate that the model has predictive relevance and good predictive power (Hair et al. 2011; Hair et al., 2017). Lastly, the model fit, measured through the standardized root mean residual (SRMR), should be applied to identify model misspecifications. The model presents a good fit as the SRMR equals 0.078 (Table 5).
Table 5. Results of the structural model
Relationship
|
β
|
Standard Deviation (STDEV)
|
T Statistics (|O/STDEV|)
|
p-values
|
f 2
|
Bias Corrected Confidence Interval
|
Hypothesis
|
|
|
|
|
|
|
Lower Bound
|
Upper Bound
|
|
Direct Effect
|
|
|
|
|
|
|
|
|
Perceived vividness → Social presence
|
0.000 ns
|
0.076
|
0.004
|
0.997
|
0.000
|
-0.128
|
0.148
|
H1a not supported
|
Perceived vividness → Spatial presence
|
0.153**
|
0.061
|
2.524
|
0.012
|
0.059
|
0.029
|
0.270
|
H1b supported
|
Perceived interactivity → Social presence
|
0.748***
|
0.066
|
11.315
|
0.000
|
0.759
|
0.590
|
0.869
|
H2a supported
|
Perceived interactivity → Spatial presence
|
0.767***
|
0.051
|
15.076
|
0.000
|
1.473
|
0.659
|
0.858
|
H2b supported
|
Social presence → Donor inspiration
|
0.366**
|
0.124
|
2.953
|
0.003
|
0.069
|
0.110
|
0.617
|
H3 supported
|
Social presence → Donation intention
|
-0.093 ns
|
0.123
|
0.759
|
0.448
|
0.006
|
-0.319
|
0.150
|
|
Spatial presence → Donor inspiration
|
0.351**
|
0.122
|
2.881
|
0.004
|
0.063
|
0.120
|
0.595
|
H4 supported
|
Spatial presence → Donation intention
|
0.155 ns
|
0.130
|
1.185
|
0.237
|
0.016
|
-0.091
|
0.420
|
|
Donor inspiration → Donation intention
|
0.739***
|
0.082
|
8.985
|
0.000
|
0.740
|
0.559
|
0.877
|
H5 supported
|
Second Order Reflective
|
|
|
|
|
|
|
|
|
Donor inspiration → Inspired By
|
0.904***
|
0.017
|
53.249
|
0.000
|
|
0.866
|
0.933
|
|
Donor inspiration → Inspired To
|
0.947***
|
0.007
|
126.573
|
0.000
|
|
0.931
|
0.960
|
|
Specific Indirect Effect
|
|
|
|
|
|
|
|
|
Spatial presence → Donor inspiration → donation intention
|
0.259 **
|
0.259
|
0.094
|
2.758
|
|
0.006
|
0.080
|
|
Social presence → Donor inspiration → Donation intention
|
0.271 **
|
0.265
|
0.094
|
2.874
|
|
0.004
|
0.082
|
|
|
|
|
|
Model Fit
|
|
|
|
|
Donor Inspiration
|
0.477
|
0.316
|
SRMR
|
0.078
|
|
|
|
|
Donation Intention
|
0.614
|
0.538
|
|
|
|
|
|
|
Spatial Presence
|
0.762
|
0.607
|
|
|
|
|
|
|
Social Presence
|
0.560
|
0.451
|
|
|
|
|
|
|
Note: f 2 effect size; ns-not significant; **p<0.01; ***p<0.001
4.3 Multigroup analyses: testing the moderation hypotheses
To analyze the moderating effect of video immersiveness on the relationship between donor inspiration and donation intention, the immersion level is considered a moderating factor. Before testing the moderating effect, quantification of measurement invariance is crucial since measurement invariance is a major issue when conducting PLS-SEM multigroup analysis. To assess it, a measurement invariance of composite models (MICOM) procedure (Henseler et al. 2016) was developed.
Next, a comparative two-level of immersion approach was formulated, resulting in three comparisons: High Immersion (HI) -Moderate Immersion (MI), High Immersion (HI) -Low Immersion (LI), and Moderate Immersion (MI) -Low Immersion (LI). As noted in Table 6, the relationship between social presence and donor inspiration is significant for both HI (p=0.023) and MI (p=0.008) conditions and is not for the LI (p=0.872) condition.
The relationship between spatial presence and donor inspiration is significant for MI (p=0.000) condition and is not for both HI (p=0.105) and LI (p=0.961) conditions. Further, the relationship between donor inspiration and donation intention does not indicate a major difference between all immersion levels, i.e., HI, MI, and LI conditions. Concerning the differences between groups, the relationship between social presence and donor inspiration, as well as the relationship between spatial presence and donor inspiration, suggests a significant difference between MI and LI conditions (p-values of 0.094 and 0.019, respectively). Likewise, the relationship between donor inspiration and donation intention reveals a significant difference between MI and LI conditions, with a permutation p-value of 0.075 (Table 7). No significant differences were found between HI and LI conditions, or between HI and MI conditions. Consequently, H6, H7, and H8 are only partially supported.
Table 6. Bootstrapping results for HI. MI. and LI separately
Relationship
|
|
Original sample
|
Sample mean
|
Std. Deviation
|
t-value
|
p-value
|
Social Presence DOI
|
HI
|
0.463
|
0.420
|
0.204
|
2.273
|
0.023
|
MI
|
0.579
|
0.613
|
0.216
|
2.674
|
0.008
|
LI
|
0.037
|
0.056
|
0.231
|
0.161
|
0.872
|
Spatial Presence DOI
|
HI
|
0.314
|
0.362
|
0.194
|
1.621
|
0.105
|
MI
|
0.755
|
0.736
|
0.216
|
3.497
|
0.000
|
LI
|
0.011
|
-0.026
|
0.218
|
0.049
|
0.961
|
DOI DI
|
HI
|
0.792
|
0.791
|
0.052
|
15.320
|
0.000
|
MI
|
0.858
|
0.856
|
0.040
|
21.463
|
0.000
|
LI
|
0.706
|
0.709
|
0.076
|
9.342
|
0.000
|
Abbreviation: DOI, Donor Inspiration; DI, Donation Intention; HI, High Immersion; MI, Moderate Immersion; LI, Low Immersion.
Table 7. Permutation test path coefficient results
Relationship
|
Comparison
|
|diff|
|
2.5%
|
97.5%
|
p-value Permutation
|
Social Presence DOI
|
HI vs. MI
|
0.423
|
-0.697
|
0.674
|
0.255
|
HI vs. LI
|
-0.120
|
-0.557
|
0.546
|
0.733
|
MI vs. LI
|
0.543
|
-0.651
|
0.624
|
0.094
|
Spatial Presence DOI
|
HI vs. MI
|
-0.440
|
-0.644
|
0.659
|
0.208
|
HI vs. LI
|
0.307
|
-0.564
|
0.546
|
0.310
|
MI vs. LI
|
0.747
|
-0.632
|
0.627
|
0.019
|
DOI DI
|
HI vs. MI
|
-0.066
|
-0.126
|
0.123
|
0.306
|
HI vs. LI
|
0.087
|
-0.180
|
0.185
|
0.349
|
MI vs. LI
|
0.153
|
-0.162
|
0.168
|
0.075
|
Abbreviation: DOI, Donor Inspiration; DI, Donation Intention.
4.4 Control variables
When analyzing the control variables considering the immersion levels (see Table 8), it is possible to understand that gender and previous usage of VR do not influence donor inspiration or donation intention.
Table 8. Control Variables Estimation Model.
|
|
Original sample
|
Sample mean
|
Std. Deviation
|
t-value
|
p-value
|
Gender DOI
|
HI
|
0.086
|
0.079
|
0.107
|
0.804
|
0.422
|
MI
|
0.198
|
0.198
|
0.102
|
1.940
|
0.053
|
LI
|
0.034
|
0.012
|
0.141
|
0.238
|
0.812
|
Gender DI
|
HI
|
0.229
|
0.230
|
0.110
|
2.083
|
0.051
|
MI
|
0.121
|
0.124
|
0.091
|
1.327
|
0.185
|
LI
|
0.085
|
0.078
|
0.111
|
0.767
|
0.443
|
VR usage DOI
|
HI
|
0.042
|
0.054
|
0.113
|
0.368
|
0.713
|
MI
|
-0.115
|
-0.108
|
0.097
|
1.189
|
0.235
|
LI
|
-0.190
|
-0.188
|
0.125
|
1.518
|
0.129
|
VR usage DI
|
HI
|
-0.044
|
-0.050
|
0.093
|
0.468
|
0.640
|
MI
|
-0.073
|
-0.077
|
0.080
|
0.911
|
0.362
|
LI
|
0.058
|
0.056
|
0.119
|
0.483
|
0.629
|
Abbreviation: DOI, Donor Inspiration; DI, Donation Intention.