[dingfeng_quek]

The study measured primary home device use based on click-stream data from comscore. This requires interpretation with the following considerations:

- The dataset they used is from a commercial entity with its own vested interests

- Secondary home device and non-home device use is not included; An increase in substitutes to the primary home-device should lead to lower primary home-device use

- Only click-stream data that can be gathered by comscore can be included

- Sampling bias of comscore (paper does not describe correction for this)

[matco11]

The higher the income, the higher the impact of non-primary devices is (simply, because with higher income you have more secondary devices)