Mann-Whitney test pivotal quantity & randomized block design

  • Context: Graduate 
  • Thread starter Thread starter Sander1337
  • Start date Start date
  • Tags Tags
    Block Design Test
Click For Summary
SUMMARY

The discussion revolves around proving pivotal quantities in the context of the Mann-Whitney test and understanding the components of a randomized block design. Specifically, the user seeks to demonstrate that W1(δ) and U1(δ) are pivotal quantities, where W1(δ) is defined as the sum of ranks assigned to Y11-δ,...,Y1n-δ, and U1(δ) is derived from W1(δ) by subtracting n(n+1)/2. Additionally, the user is tasked with proving the equation SS = SSE + SST + SSB, where SSE measures variability within populations, SST measures variability due to differences between populations, and SSB measures variability between blocks.

PREREQUISITES
  • Understanding of pivotal quantities in statistics
  • Familiarity with the Mann-Whitney test
  • Knowledge of randomized block design concepts
  • Ability to manipulate statistical formulas and summations
NEXT STEPS
  • Study the properties of pivotal quantities in statistical inference
  • Learn about the derivation and application of the Mann-Whitney test
  • Research the principles of randomized block design and its statistical significance
  • Explore techniques for simplifying complex statistical formulas
USEFUL FOR

Students in statistics, researchers conducting non-parametric tests, and anyone involved in experimental design and analysis of variance.

Sander1337
Messages
10
Reaction score
0
Hi there,

I have a question about an assignment I got from school.

They were initially 12 assignments, I've finished 10 so far but I still can't figure how my last 2 assignments work.

First there is the question;

Prove that W1(δ) and U1(δ) are pivotal quantities, where W1(δ) = sum of ranks assigned to Y11-δ,...,Y1n-δ and U1(δ) = W1(δ) - n(n+1)/2.

I really don't understand how to prove that. I can imagine that proving that W is a pivotal quantity, will automatically result in U being a pivotal quantity since W is independent from parameter δ (is that the right parameter?).

Second question is about randomized block design.

The given question/assignment was; prove that SS=SSE+SST+SSB where;
SSE=measures variability in populations
SST=measures variability due to differences in populations
SSB=measures variability between blocks (?)
SS= measures total variability in data

http://img4.imageshack.us/img4/2817/knipseli.jpg

I decided that it would be a lot more convinient to prove SS-SST-SSB=SSE since their formula's aren't so complex as the one from SSE is

Nevertheless it doesn't add up... When I simplefy them all (and with some help of reverse engineering -> simplefy SSE as well) I eventually end up with \Sigma\Sigma(YijYi\bullet-YijY\bulletj which should be equal to 2... (the bullets are supposed to be before respectively behind the j and the i in subscript)

I worked it all out, if any of you would like to see scans/images of what I've written on paper to see what I've done, just ask. I think I've made a mistake in simplefying the initial errors before I all summed them up. Nevertheless, on request, I will post my complete 3-page (bad handwriting) simplification/solution so far...

Please help me out on this, I'm going to get beserk in a matter of minutes cause the first 10 assignments already took me about 3 days to finish, but these 2 already took me a day and I still can't get how to prove them both...

Kind regards,

Sander
 
Last edited by a moderator:
Physics news on Phys.org
Sorry for posting this in the wrong forum. I didn't knew wether this was a homework question or not, since I have a question about a method instead of about homework..

Please move post to right section if moderator thinks otherwise!
 

Similar threads

Replies
24
Views
8K