0

this will be on my exam and i'm indignation to have such confusing/vague questions "validating" my results (explaining me why it is not "confusing/vague" and make me look like a fool would be appreciated, that I can forget in one day, not having a clear answer to the question at matter, hasn't let me sleep weell in 3 days already)

Question:

"A quality control specialist compares the output from a machine with a new lubricant to the out of machines with the old lubricant. Is that an experiment or an observational study?"

I mean... if he himself(the quality inspector) didn't brought the "old lubricant" or the "new lubricant" with-himself to, then start comparing; then he is just a person who walked into a bilding with already both lubricated machines (with the old lubricant and also the new lubricant) working already at the same time, right? and he is only ""comparing/observing"" the different outputs of both machines. Isn't that the nature of an "observational study" or i'm wrong?

Dayron
  • 3
  • 2
  • Has your teacher offered actual definitions of "experiment" and "observational study"? Presumably so: what are they? – whuber Mar 11 '22 at 21:44

1 Answers1

1

The absolute minimum requirement of an "experiment" as compared to an "observational study" is that the study conductor has control over which treatment condition is applied to each experimental unit.

The question you quote is more vague than questions of this type usually are, at least in my experience. If the quality control specialist is able to choose which machines use the new lubricant, then it is an experiment. If not, then it is an observational study. My guess is that the exam writer intended you to assume that the specialist is able to choose, but it is not entirely clear from the wording. If instead the lubricant assignment is pre-existing, for example if newer machines use the new lubricant and older machines use the old, then it is an observational study.

Moore & McCabe is a classic statistics textbook with many nice problems of this type comparing experiments to observational studies, all of which (unlike the current example) are quite unambiguous in my opinion. My experience is with the early editions, which I used to teach biostatistics classes --- I have not checked whether the most recent editions include the same material, but I assume they would.

Gordon Smyth
  • 12,807
  • thank you, this makes me feel much better. When I explained the same thing as I did here to my professor (instead of explaining things like you), he just told me "you are commiting the graitest mistake that people who manages data can commit wich is: not accepting facts, because of your feelings or gut instincs," and on my point of view i'm vere much not being "sentimenta" because at the question it is not written that the quality instructor "manipulated enething" and that is a fact not a "feeling" – Dayron Mar 12 '22 at 17:05