4
$\begingroup$

I am reading Yudi Pawitan's In All Likelihood chapter 7 section 5. He mentioned the following in sequential design using likelihood approach with justification of stopping rule not affecting likelihood inference.

'A strict adherence to the strong likelihood principle in this case implies that the evidence from such an experiment does not depend on the stopping rule. That is, if we want to find out what the data say, we can ignore the stopping rule in the analysis. This is convenient if we stop a study for some unrelated reason (e.g. a power cut), but it is also the case even if we decide to stop because ‘we are ahead’ or the ‘data look good’.'

Q1: What is power cut here? Is it just lowering originally planned 80% power to 50% by early stopping?

Q2: When do we use power cut? Most of studies will not be overpowered as it is a waste of resources to overpower.

$\endgroup$
9
  • 4
    $\begingroup$ Doesn't it mean an outage in the power grid? $\endgroup$
    – dipetkov
    Jul 25 at 16:21
  • $\begingroup$ @dipetkov I am not sure whether it means power outage here. I thought power means 1-P(type II error). Or is cutting statistical power unrelated to the study? I think it is. $\endgroup$
    – user45765
    Jul 25 at 16:23
  • $\begingroup$ To most of the world "power" means something else. This is a popular science book. $\endgroup$
    – dipetkov
    Jul 25 at 16:24
  • 3
    $\begingroup$ @dipetkov (+1) I suggest you post that as an answer. I've noticed you've provided useful and succinct answers in comments before; the consensus in SE is that even very short answers are worth posting as answers and not as comments. I guess this one is a somewhat grey area since the answer is not actually about stats, but I think it's still worth it. $\endgroup$
    – mkt
    Jul 25 at 17:07
  • 1
    $\begingroup$ @dipetkov: In all likelihood is not a popular science book ... $\endgroup$ Jul 25 at 17:38

1 Answer 1

11
$\begingroup$

This is an interesting observation: the paragraph talks about experiments and how to analyze them if stopped early. So it's not a stretch to connect "power cut" with "statistical power", a concept relevant to planning scientific experiments. However, In All Likelihood takes an informal approach*, so the odds are (pun intended) that "power" refers to electricity and "power cut" means a power outage.

٭ From the description on Goodreads: "The book generally takes an informal approach, where most important results are established using heuristic arguments and motivated with realistic examples."

$\endgroup$
2
  • 3
    $\begingroup$ +1. In the US a native English speaker would write "power outage." $\endgroup$
    – whuber
    Jul 25 at 17:51
  • 8
    $\begingroup$ +1 'Power cut' is standard usage for a failure in electricity supply in the UK, India and perhaps South Asia in general (I think the author is from Indonesia). I think the US might be the exception in using 'power outage'. $\endgroup$
    – mkt
    Jul 25 at 17:55

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.