DE COMUNICACION AUDIOVISUAL CAPÍTULO
ARTÍCULO 38 Multiplicidad de licencias.
After examining the measurement properties of our instrument, we tested hypotheses regarding user choices of strategies of adaptation.
4.7.1. Direct effects
We first included age and gender as control variables in the model. Since these control variables were not found significant for either strategies of adaptation, we did not include them in further analyses.
OPPORTUNITY CONDITION. The results show that control is positively associated with choice
of a BM strategy (
β
= .34, p<.000) rather than of a BS strategy. Next, we found EUA is directly and significantly related to greater preference of a BM strategy over a BS strategy (β
= .32, p<.000). H1a is thus validated.
THREAT CONDITION. As expected, the results for the threat condition indicate that control is
significantly and positively associated with the choice of a DH strategy of adaptation over a SP strategy (
β
= .22, p<.005). We also found a significant influence of EUA on ThChoice (β
= .20, p<.005). This means that the more users espouse uncertainty avoidant values, the more they adapt to the situation with a DH coping strategy. H1b is thus validated. The results of the
Table 4.9. Path Coefficients (Full Model)
Dependent
Variable Independent Variable Original Sample Sample Mean Deviation Standard Standard Error Statistics T- OpChoice Control .34 .34 .06 .06 6.13*** EIC x Control -.17 -.16 .06 .06 2.73** ECI .05 .08 .05 .05 .90 EUA .32 .31 .07 .07 4.28*** ThChoice Control .22 .22 .07 .07 3.21*** EIC x Control -.08 -.11 .05 .05 1.56 EIC -.13 -.15 .06 .06 2.11* EUA .20 .20 .07 .07 2.70**
EIC = Espoused Individualism-Collectivism; EUA = Espoused Uncertainty Avoidance *p<.05, **p<.01, *** p<.001
4.7.2. Moderation Effects
Direct effects posited in the models where thus found to be significant. The assessment of moderation effects requires further calculation. Given a dependent variable X and a dependant variable Y, according to Carte and Russell (2003, p. 3), Z is a moderator if “the nature of the X
→
Y relationship varies as a function of Z”. We posited that IC had a moderating influence on the relationship between control (CO) and the choice of a coping strategy. In order to identify the moderating effects of this variable, we followed the guidelines proposed respectively by Chin et al. (2003) and Carte and Russell (2003) which differ each other. First, consistent with the procedure explained in Chin et al. (2003) and with Goodhue et al. (2007) we created the moderating construct by multiplying the items of the direct effect variable, Control (CO), by the items of the moderating variable, IC. Since IC is reflective, we standardized indicators before multiplication. In fact, standardization as well as the items cross multiplication were directly computed by SmartPLS (Ringle et al. 2005). We then followed the hierarchical process described by Chin et al. (2003), comparing a model with the interaction construct with another model called the main effects model without the interaction construct. A pseudo F-test was then computed. According to Cohen (1988), a pseudo-F with a value of .02 the effect is small, with .10 it is medium, and with .15 it is strong. The results ofTable 4.10. Effect Size Path Beta T R2 Main Effect Model R2 Interaction Model Δ f 2 Interpretation
[EIC x CO] → [CO → OpChoice] .17*** - 2.73 .21 .23 .03 .04 Small
[EIC x CO] → [CO → ThChoice] -.08 1.56 .08 .08 .01 .01 N.S.
EIC = Espoused Individualism-Collectivism; CO = Control p< .05, ** p<.01, *** p<.001
N.S. = Non-Significant; CO= Control
OPPORTUNITY/CONTROL CONDITION. The results given in Table 3-9 and Table 3-10 are a
standardized beta of .34 from CO to OpChoice, of .05 from EIC to OpChoice, and of -.17 for the interaction term CO x EIC with a total R2 of .25. According the Chin et al. (2003)
procedure, one standard deviation increase in IC will have an insignificant impact of .05 on OpChoice, but it decreases the impact of CO on OpChoice from .34 to .17. Further, Table 3- 10 shows that the squared multiple correlation (R2) of the main effect model (R2 = .21) is smaller than the one of the interaction model (R2 =.23). The effect size of the interaction construct is thus f2 = .04, which is between small and medium (Cohen 1988). From our analyses, we conclude that high collectivistic subjects are less likely to adopt a BM strategy for a given level of control, while high individualist subjects tend, conversely, to adopt a BM strategy. Therefore, H2a is validated.
THREAT/CONTROL CONDITION. Following the procedure of Chin et al.(2003) (2003), we failed
to find a moderating effect of IC on ThChoice when the situation has many threats. Following this procedure, therefore, H2b is not validated. However, while not hypothesized, we found a
significant direct effect of IC (
β
= -.13, p<.005) on ThChoice. Irrespective of control, it means high collectivistic individuals have a greater preference for a SP coping strategy than low collectivistic ones.To test for moderation effect, Carte and Russell (2003) suggest a different procedure from that of Chin et al. (2003). Because moderation effects calculations can lead to major errors of
interpretation, these researchers suggest a set of guidelines to further ensure the existence of interaction effects. In this analysis, we have been cautious in avoiding the nine errors they highlighted, as shown in Appendix 1. To assess a moderating effect, Carte and Russell (2003) suggest the calculation of a F-statistic that has to lead to the rejection of H0 = ΔR2 = 0. As
opposed to Chin et al.(2003), who took into consideration the estimated
β
of the interaction term, Carte and Russell (2003) believe that only the ΔR2 effect size is important to estimate interaction effects. They argue that, in particular, when interval scales are used, measurement units are arbitrary (Carte et al. 2003, p. 6). Therefore, they suggest the calculation an F- Statistic given by the equation (E) below:(E):
(
)
(
)
) 1 /( ) 1 ( 1 , 22 − − − − Δ = − − − mult mult add mult mult add mult df N R df df R df N df df FFollowing this formula, Carte and Russell (2003, p. 3) argue that an F significantly higher than 1.00 leads to the rejection of H0 = ΔR2 = 0. However, Carte and Russell (2003) do not
provide any guideline about how much greater than 1.00 should the F statistic be. The results of calculating the equation above are given in Table 4-11 below:
Table 4.11. F Test
Moderator N R2
add R2 mult ΔR2 F Interpretation
[IC x CO] → [CO x OpChoice] 207 0.21 0.23 0.03 7.46 H0 = ΔR2 = 0 rejected
[IC x CO] → [CO x ThChoice] 207 0.08 0.08 0.01 1.33 H0 = ΔR2 = 0 not rejected
EIC = Espoused Individualism-Collectivism, CO = Control
The results show that with F=7.46, the F statistic is much greater than 1 for the opportunity condition, leading to the rejection of H0 = ΔR2 = 0 and to the conclusion that EIC is a
significant moderator of the relation between control and coping strategies. When the situation is threatening, we obtain F=1.33, which is, however, marginally greater than their 1.0 threshold. We thus cannot clearly reject the null hypothesis. Following the formulation of
Carte and Russell (2003), EIC is thus not a significant moderator to the relation between CO and ThChoice, which did not validate H2b.
Therefore, according to both the procedure suggested by Carte and Russell (2003) and Chin et al.(2003), IC has a significant moderating effect on user coping strategies of adaptation in opportunity condition but not in threat condition. Overall, our model is thus strongly supported since all but one hypothesis are strongly supported, the remaining one being partly supported. Hypotheses are summarized in Table 4-12 below.
Table 4.12. Summary of Hypotheses
Hypotheses Validated
(Yes/no) H1a: In the case in which the technology offers many opportunities, individuals who espouse high UA national
cultural values will tend to adopt a BM coping strategy of adaptation rather than a BS strategy of adaptation. Yes H1b: In the case in which the technology offers many threats, users who espouse high UA national cultural values will tend to adopt a DH coping strategy of adaptation rather than a SP coping strategy of adaptation. Yes H2a: In the case in which the technology holds many opportunities, the higher the control they have over the situation, the less users who espouse more collectivistic national cultural values are likely to adopt a BM coping strategy of adaptation rather than a BS coping strategy of adaptation.
Yes1
H2b: In the case in which the technology holds many threats, the higher the control over the situation, the less users who espouse more collectivistic national cultural values are likely to adopt a DH coping strategy of adaptation rather than a SP coping strategy of adaptation.
No1 1The hypotheses were tested following both the methodology for moderation identification of Carte and Russell (2003) and that
of Chin et al. (2003). A significant direct effect of IC on ThChoice has been found.