- Ground-motion model is (same functional form as Bindi et al. (2011) to which study is similar):
_{1}= 3.72318, b_{1}= -0.06573, b_{2}= -0.05886, b_{3}= 0, c_{1}= -1.81275, c_{2}= 0.31914, h = 8.61357, c_{3}= -0.00008, f_{1}= -0.02604, f_{2}= 0.13674, f_{3}= 0.02803, f_{4}= 0, s_{1}= 0, s_{2}= 0.16032, s_{3}= 0.18900, s_{4}= 0.17194, s_{5}= 0.59823, τ = 0.21738 (inter-event), ϕ = 0.28063 (intra-event) and σ = 0.35498 (total) for BIea dataset; e_{1}= 3.80924, b_{1}= 0.19680, b_{2}= -0.11244, b_{3}= 0.00001, c_{1}= -1.50507, c_{2}= 0.02642, h = 5.52095, c_{3}= 0.00004, f_{1}= 0.09705, f_{2}= 0.15338, f_{3}= 0.14478, f_{4}= 0, s_{1}= 0, s_{2}= 0.12467, s_{3}= 0.21216, s_{4}= -0.00558, s_{5}= 0.55732, τ = 0.27238 (inter-event), ϕ = 0.28635 (intra-event) and σ = 0.39520 (total) for BIea2 dataset; e_{1}= 3.49977, b_{1}= -0.00084, b_{2}= -0.11122, b_{3}= 0, c_{1}= -1.59946, c_{2}= 0.21243, h = 5.63087, c_{3}= -0.00097, s_{1}= 0, s_{2}= 0.16701, s_{3}= 0.09431, s_{4}= 0.04135, s_{5}= 0, τ = 0.13690 (inter-event), ϕ = 0.25777 (intra-event) and σ = 0.29187 (total) for ABR dataset (mechanism terms removed). - Use 5 site Eurocode 8 (EC8) classes (150 stations in total):
- A
- V
_{s,30}> 800m∕s. C_{A}= 1 and other C_{i}s are zero. - B
- 360 < V
_{s,30}≤ 800m∕s. C_{B}= 1 and other C_{i}s are zero. - C
- 180 < V
_{s,30}≤ 360m∕s. C_{C}= 1 and other C_{i}s are zero. - D
- V
_{s,30}≤ 180m∕s. C_{D}= 1 and other C_{i}s are zero. - E
- 5–20m of C- or D-type alluvium underlain by stiffer material with V
_{s,30}> 800m∕s. C_{E}= 1 and other C_{i}s are zero.

About 130 stations are classified based on shear-wave velocity profiles and rest from geological and geophysical data.

- Use 4 faulting mechanism classes using classification of Zoback (1992):
- Normal
- E
_{1}= 1 and other E_{i}s are zero. - Reverse
- E
_{2}= 1 and other E_{i}s are zero. - Strike-slip
- E
_{3}= 1 and other E_{i}s are zero. - Unknown
- E
_{4}= 1 and other E_{i}s are zero.

- Derive models using 3 datasets:
- BIea
- Dataset of Bindi et al. (2011) but retaining singly-recorded events to increase number of stations with > 2 records. Only 25 stations recorded > 9 events. 117 stations.
- BIea2
- Extend BIea to include all records from 4.0 ≤ M
_{w}6.9 including those for which magnitude conversion required. 254 stations. - ABR
- Data from 2009 L’Aquila sequence (42.4-42.8N, 13.2-13.6E). 38 stations. All events are normal-faulting (hence mechanism terms removed) and have focal depths < 10km.

- Records baseline corrected and filtered using 2-order acausal Butterworth filter after cosine tapering. Select cut-offs based on Fourier amplitude spectra. Double integrate to get displacements. Linearly detrend displacements. Double differentiate to get correct accelerations.
- Develop model to examine single-station σ (computed using approach of Rodriguez-Marek et al. (2011)) and influence of datasets on its value.
- Examine inter-event residuals w.r.t. M
_{w}and find no trends. - Examine intra-event residuals w.r.t. station number.
- Examine intra-event residuals w.r.t. r and find large variability in range 80–100km, which relate to Moho bounce.
- Examine histograms of ϕ
_{SS,S}, distribution of ϕ_{SS,S}w.r.t. site class, and ϕ_{SS,S}w.r.t. station number. Identify stations with large ϕ_{SS,S}and discuss reasons for large values. - Examine single-station σ w.r.t. magnitude-distance bins. Find that it is generally higher for 0–40km.
- Examine influence of number of records per station on mean ϕ
_{SS}and find results are quite stable. - Find σ is higher for BIea2 compared to BIea, which relate to use of converted M
_{w}. - Find σ for ABR is lower compared to other datasets, which relate to restriction of events from small geographical area.