- Ground-motion model is:
where gal, T is predominant period of site, α = 0.509, β = 2.32, γ = 0.039 and δ = 2.33 (σ not given).

is in - Correlation coefficient between magnitude and distance is 0.84, so magnitude and distance cannot be considered independent, so attenuation rate, β, is difficult to find.
- First step fit log = -β
_{i}log(Δ + 10) + δ_{i}to each earthquake. Define reliability parameter, ψ_{i}= N_{i}R_{i}^{2}, where N_{i}is degrees of freedom for i earthquake and R_{i}is correlation coefficient. Plot ψ_{i}against β_{i}and find attenuation rate scattered, between -6 and 9, for ψ_{i}< 1 (Group B) and for ψ_{1}> 1 attenuation rate converges (Group U). - Group B includes earthquakes with focal depths > 388km, earthquakes with small magnitudes and records from distances ≈ 100km, earthquakes with records from great distances where spread of distances is small, earthquakes recorded by only 3 stations and earthquakes with abnormal records. Exclude these records.
- Apply multiple regression on Group U to find α, β, γ and δ simultaneously. Also fix β = ∑
ψ
_{i}β_{i}∕∑ ψ_{i}and find α, γ and δ. Find different coefficients but similar correlation coefficient. Conclude due to strong correlation between M and Δ so many regression planes exist with same correlation coefficient. - Perform Principal Component Analysis (PCA) on log A, M, log(Δ + 10), T and log ∕A and find that equation found by fixing β is not affected by ill-effect of correlation between M and Δ.
- Omit T from regression and find little effect in estimation.