Talk:Pseudoreplication

Update 2011[edit]

I restored the definition of 'pseudoreplication' by quoting Hurlbert's article, with attribution. I added a simple example (tanks) and a computationally correct definition (misformed F-ratio). I added detail to the types of 'pseudoreplication' as defined by Hurlbert. I revised several topics (Hypothesis testing, Notes). 174.116.84.237 (talk) 05:32, 5 December 2011 (UTC)David Schneider174.116.84.237 (talk) 05:32, 5 December 2011 (UTC)[reply]

Update[edit]

I think that this article could use a lot more information - it is a field that is majorly important in experimental design in the life sciences, and is a focus of study for a lot of people. I've tried to add a bit here and there, but it needs clarified and added to. Dackerts (talk) 23:38, 8 December 2010 (UTC)[reply]

Plagiarism[edit]

The first sentence of the original version of this article was a direct quote from the cited reference (Hurlbert), although it is not identified as such. Here is the offending material:

Pseudoreplication is defined as the use of inferential statistics to test for treatment effects with data from experiments where either treatments are not replicated (though samples may be) or replicates are not statistically independent.

As far as I can tell, the rest of the article was Krispi's original contribution, but I've edited that as well, for clarity. DavidCBryant 22:14, 28 April 2007 (UTC)[reply]

t-test assumptions[edit]

The following sentence was just removed, but I think it is correct, "This replication must also be appropriate: there is a requirement for independence at experimental unit level, as opposed to sample unit level. " The t- and F-tests require independence among all the observations to be correct. Am I missing something? 0¹⁸ (talk) 15:10, 10 December 2010 (UTC)[reply]

Basically I'm a stats n00b, and can't figure out the difference between an experimental unit and a sample unit, or if there is in fact a a difference - things I read seem to conflict on the matter, so I removed out of precaution. Any clarity appreciated! Dackerts (talk) 17:42, 11 December 2010 (UTC)[reply]

An experimental unit is an observation (i.e. a plot of land) a sampling unit is the smallest group that can be "in or out" of a treatment group (i.e. the land an individual farmer owns). I've never seen any confusion on this topic. Can you point to some? 0¹⁸ (talk) 19:11, 11 December 2010 (UTC)[reply]

A quick google brings up something that doesn't differentiate it - admittedly that's a pretty terrible site though. I can't remember where I saw it confused before, not much time to look just now [1] Dackerts (talk) 15:56, 12 December 2010 (UTC)[reply]

Incidentally, I noticed you said 'I don't get it isn't a good reason to delete' in your reversion explanation - I wrote the paragraph myself a few hours previously. Perhaps 'I don't get it' is a good reason not to write? :-) Dackerts (talk) 16:07, 12 December 2010 (UTC)[reply]

Sorry, didn't notice that. Yeah, probably. Thanks for helping out this this page. 0¹⁸ (talk) 20:41, 12 December 2010 (UTC)[reply]

Clarity[edit]

I apologize in advance to all who have put effort into this page, but I think it lacks clarity for the layman. I for one don't understand the first paragraph... in fact I tried reading the original Hulbert publication and it seemed rather more clear! (I'm not a specialist in the field)
My suggestions would be:

If indeed as I understood he coined the phrase, cite Hulbert as soon as the first paragraph, à la "Pseudoreplication was defined in a seminal paper and book by Hulbert as..."
define the term with a name, not a behavior. The first sentence is not "pseudoreplication is...", it is "pseudoreplication is used to test for...".
If, as I understand, pseudoreplication is an error, say so as soon as possible, so that the reader doesn't feel they have to understand a method only to discover it's erroneous. :-)

I really don't know enough about the subject to edit directly without consulting specialists, but I would add this first sentence:
Pseudoreplication is an methodology error in experimental science, first identified by Stuart Hulbert in a 1984 article.
...and then go on from there.
(Oh, and if the 1984 paper is one of the most cited papers in its field, perhaps that anecdotal fact may be brought up...) --Jrob kiwi (talk) 14:19, 19 August 2013 (UTC)[reply]

Agreed: I used to work in the field and I can't understand what the first paragraph means. 129.31.37.126 (talk) 14:16, 2 November 2015 (UTC)[reply]

Rework[edit]

Prof. Stuart Hurlbert himself commented on this article here. There are error which should be corrected: https://stats.stackexchange.com/questions/549711/how-does-pseudoreplication-differ-from-resampling Biggerj1 (talk) 07:52, 26 October 2021 (UTC)[reply]