Quote:
Originally Posted by ApK
I'm likely to try the experiment myself, if I can think of some good control topics, that I haven't seen ads for before and haven't explored online.
|
I have some suggestions for anyone attempting this sort of experiment to help eliminate confounding factors that could influence the results.
Decide in advance how many days you will be collecting data. If you leave it open ended the chances increase of a spurious result.
Select the subject topic ahead of time, but do not tell anyone else what it is or post about it online.
Before mentioning the topic in front of your device monitor your ads for that subject for the number of days that you have decided on. That will be your control sample. You may be surprised to learn that you have been getting ads for all kinds of things that you never noticed because they are on no interest to you. If your test subject comes up then count that as a miss.
Discuss the topic in front of your device. Take special care not to mention the trigger word or anything close to it.
Now monitor ads again for the predetermined number of days and see what comes up.
Realize that the occurrence of a hit could be coincidence. You will need to repeat the test a number of times counting hits and misses before you can draw any real conclusions.