I work in market research.
For years, I’ve been trying to wheedle my vast statistical knowledge into my workplace.
From my tone, you can probably guess how that’s gone. My bosses really haven’t cared one whit about statistical significance.
Now, however, I have a manager who has some appreciation for the statistical art. In the meantime, I am now 18 years separated from formal instruction therein. Therefore, I submit my sorry state to the learned fellows (in the gender-neutral term) of the Dope.
Say you do a sales mailing to 1,000,000 people. From this, say you get 1,000 sales. What is the statistical test to determine (within two standard deviations, say) what a predictable outcome for the next mailing would be? Further, how may one calculate the relationship between volume (either number of sent pieces or responses) and the standard deviation? What my boss REALLY wants to know is how many sales he needs to get before the test is really predictive, but I realize this is predicated on other underlying statistics.
Thank you, mathematical sachems, in advance. Please note that I am not asking you to do my work for me. Rather, I am prevailing on your notorious generousity to give me a nudge in the right direction.