Suppose now that instead of herders buying the UThere phones, herders in the region were randomly given either a new UThere phone or soap of equivalent value. The impact evaluation team then randomly...

Suppose now that instead of herders buying theUThere phones, herders in the region were randomly given either a newUThere phone or soap of equivalent value. The impact evaluation team then randomly surveyed 250 herders who receive the phone and 250 who received the soap. When you directly compare average income for these two groups, the difference gives you:

• Naïve ATE due to selection bias, since those who randomly got phones are different from those who got soap.


• True ATE due to the randomization which implies that those who got soap are otherwise comparable to those who got phones.


• different way of measuring selection bias due to the randomization, since getting soap might increase the average income but not as much asUThere phones.


• the same value of Naïve ATE that you might expect to get without a random distribution of phones and soap.