Случайное всегда должно контроллироваться экспериментатором =(

Правильная рандомизация:

1) ) cycle through all items in a list before repeating them, and repeat in a different random order the next time through (we called this «mini-blocking» in grad school, not sure how widely that term is used),

2) constrain the number of possible consecutive items in a row (e.g., no more than 3 otherwise people start to wonder if the next item will also be the same),

3) fix the first 3-5 items so that all subjects in a study will experience the same thing initially, avoiding potential differences in anchoring (especially if some items are low frequency but by chance are over represented early in a random sequence).

отсюда