Further religious experiments (cont‘d)

• Solution:

  • Use only one „wealth“ variable in the model
  • Avoid pairs of variables with strong covariance
  • Drop „insignificant“ religions

• Experiments 5, 6, 7, 9

Notes: