Re: Computing overall rank
- From: "Stratocaster" <stotz1@xxxxxxxxxxx>
- Date: Sat, 16 Jun 2007 23:58:38 GMT
"Swati" <swati.smita@xxxxxxxxx> wrote in message
news:1181928060.436219.107170@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Hi!
I have a data set which has responses in rank order. That is,
respondents have ranked 14 items in a list (say leadership qualities)
according to their imporantance to a certain function. I have two
quaries:
1. How do I get overall rank for a particular item for the sample?
e.g.
(A) Take sum of all ranks and then arrange the items in decending
order and rank them again.
(B) Take sum of all ranks and then devide it with total N, arrange the
items in decending order and rank them again.
(B) (Rank1*freqency of Rank1 + Rank2*frequency of Rank2......)/N
I think if you perform all three, you will find no difference in the ranks
they determine.
Lets say item A's ranks are summed up and equal a. Assume B's rank sums to
b, etc.
Whether or not there exists a differnce between 'a' and 'b' will not be
altered by division by identical denominators.
i.e. If a>b, Then (a/n)>(b/n), whenever 'n' is a positive number (and this
will always be the case for sample sizes).
Likewise with (rank1*frequency+rank2*frequency...) this is the same as
summing all the ranks provided. Long story short, to determine rank, it
doesn't really matter, any of the above methods will assign the same ranks.
.
- References:
- Computing overall rank
- From: Swati
- Computing overall rank
- Prev by Date: How to get confidence intervals by simulation
- Next by Date: Re: How to get confidence intervals by simulation
- Previous by thread: Computing overall rank
- Next by thread: Math Ebooks
- Index(es):
Relevant Pages
|
