Say I have the problem: I roll a die three times and I am interested in the probability of ending up with two 1's.

My impression is that a single roll is called a trial.

What is the full 3-roll action called? I thought it was called an experiment, but I've seen experiment used synonymous with trial.

Using the binomial distribution, we call getting a 1 here a "success" and getting anything else a "failure." This might confuse students who think it's a "failure" to get three 1's. Is there an alternative/better way to say this acquiring of a 1 vs. not a 1?

  • 1
    $\begingroup$ "ending up with two 1's" Do you mean exactly two or at least two? $\endgroup$ Apr 12, 2021 at 0:45
  • $\begingroup$ I mean exactly two 1's. $\endgroup$
    – Nights
    Apr 12, 2021 at 8:22
  • $\begingroup$ You should edit your post to reflect that. $\endgroup$ Apr 12, 2021 at 8:51

1 Answer 1

  • You might point out that the $3$-trial experiment has $8$ possible outcomes and $256$ possible events, which are subsets of the sample space; whereas each trial has $2$ possible outcomes, which can be contextually characterised as a success and a failure.

    To distinguish between trial outcomes and experiment outcomes (or to remind about their distinction), I sometimes call the former “sub-outcomes” and the latter “elementary events”. When introducing the terminology, I indicate that “sub-outcome” is a nonstandard word (actually, I coined it).

  • To directly address your question:

    I would point out that the assignments ‘success’ & ‘failure’ naturally arise out of Bernoulli trials having exactly two outcomes of fixed probabilities.

    In contrast, it is not so useful to call the event $\{n11,1n1,11n\}$ or its outcomes “successes”; moreover, we might be interested in multiple events in the experiment, in which case the notion of a “successful” event (or elementary event) will have to keep varying according to the event being considered.

  • $\begingroup$ So you agree that experiment is the word for the full 3-roll activity, and is not synonymous with trial? $\endgroup$
    – Nights
    Apr 12, 2021 at 8:25
  • $\begingroup$ Apparently my stats are rusty because I don't know where you're getting 256. Do you mean 216 (6*6*6)? $\endgroup$
    – Nights
    Apr 12, 2021 at 8:31
  • $\begingroup$ @Nights 1. Yes. And given enough discussion time, I might also point out that while any two trials of this experiment are independent of each other, two events with reference to this experiment could be either dependent or independent. $\quad$2. The sample space has $2^3=8$ members/outcomes and thus $2^8=256$ subsets/events. $\endgroup$
    – ryang
    Apr 12, 2021 at 8:39
  • $\begingroup$ There are six outcomes in a trial: 1 - 6. When you write out every possibility you have 216 possible 3-die rolls. I thought that THIS constitutes the sample space (I haven't opened the class textbook yet, but this is what I remember.) The probability is calculated by hand by finding the number of rolls which give two 1's, where a roll (1,1,2) is different from a roll of (1,1,3), for example, and dividing this number by 216. Where are you getting this other use of the term "sample space" in terms of success/failure only? And what's the significance of your power set calculation? $\endgroup$
    – Nights
    Apr 12, 2021 at 9:16
  • $\begingroup$ 1. In your binomial model, each trial has exactly two possible outcomes: ‘1’ (success) and ‘n’ (failure, whose likelihood here is five times that of success). The game can of course be reframed as a different probability experiment in which each trial has six equally-likely outcomes; in this model, the experiment indeed has $6^3=216$ outcomes. $\endgroup$
    – ryang
    Apr 12, 2021 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.