The authors collect 43 indicators for 149 countries and compute a well-being index based on these numbers. Unfortunately they don't explain how they compute the index. But since the indicators comprise a wide range starting from GDP per capita, quality of roads, Gini coefficient, press freedom etc. my first idea would be to combine these indicators to an overall index using a geometric mean.
However the authors mention in the appendix, that each indicator is normalized into a scale of 0-100. Since the scale starts at zero, I believe they cannot use a geometric mean, so my guess is they use an arithmetic mean to compute the index in the end.
So my question is: In such a case should one use an arithmetic mean of normalized scores or better a geometric mean of the indicators without normalization?