Table 3 Correlations between constructs...

Table 3

Correlations between constructs and the square root of AVE

	1	2	3	4	5	6	7	8	9	10	11
1. Autonomy	0.804
2. Belongingness	0.501	0.863
3. Competence	0.406	0.499	0.807
4. Intrinsic Motivation	0.433	0.452	0.501	0.922
5. Invasion	0.031	0.299	0.203	0.177	0.826
6. LinkedIn Exhaustion	−0.200	0.071	−0.095	−0.239	0.303	0.924
7. Perceived Ease of Use	0.275	0.183	0.451	0.370	0.001	−0.199	0.878
8. Perceived Usefulness	0.321	0.344	0.505	0.527	0.215	−0.121	0.461	0.783
9. Privacy Concerns	−0.151	0.064	−0.050	−0.053	0.261	0.324	−0.168	−0.067	0.927
10. Privacy Risks	−0.079	0.047	0.060	0.018	0.279	0.218	−0.091	−0.015	0.603a	0.878
11. Social Overload	0.178	0.459	0.209	0.293	0.574	0.329	0.019	0.221	0.239	0.222	0.858

	1	2	3	4	5	6	7	8	9	10	11
1. Autonomy	0.804
2. Belongingness	0.501	0.863
3. Competence	0.406	0.499	0.807
4. Intrinsic Motivation	0.433	0.452	0.501	0.922
5. Invasion	0.031	0.299	0.203	0.177	0.826
6. LinkedIn Exhaustion	−0.200	0.071	−0.095	−0.239	0.303	0.924
7. Perceived Ease of Use	0.275	0.183	0.451	0.370	0.001	−0.199	0.878
8. Perceived Usefulness	0.321	0.344	0.505	0.527	0.215	−0.121	0.461	0.783
9. Privacy Concerns	−0.151	0.064	−0.050	−0.053	0.261	0.324	−0.168	−0.067	0.927
10. Privacy Risks	−0.079	0.047	0.060	0.018	0.279	0.218	−0.091	−0.015	0.603a	0.878
11. Social Overload	0.178	0.459	0.209	0.293	0.574	0.329	0.019	0.221	0.239	0.222	0.858

Note(s): The square root of average variance extracted (AVE) is shown in the main diagonal

We acknowledge that the correlation between privacy concerns and privacy risks (0.603) may appear high. However, it remains below the square root of the average variance extracted (AVE) for each construct, indicating acceptable discriminant validity. As a reminder, these two constructs belong to the second-order reflective-reflective construct of privacy threats. As emphasized by methodological experts (e.g. Hair et al., 2024), in reflective-reflective higher-order constructs, the lower-order constructs (in this case, privacy concerns and privacy risks) are expected to be highly correlated, as the higher-order construct accounts for and explains these correlations. Consequently, this level of correlation is not problematic but rather necessary to support the validity of the measurement modelSource(s): Authors’ own work

[ViewLarge]