is there any other statistical test which can be applied n frequency set of data for figuring out statistical significance...
Hi,
We can use chi-square test to test the goodness of fit.
http://stattrek.com/chi-square-test/goodness-of-fit.aspx
http://www.itl.nist.gov/div898/handbook/eda/section3/eda35f.htm
Hi, Good Morning
Your question: Alternative test for chi square on frequency data ?
For alternative test for chi-square is Fisher's exact test.
Note:
1. All value in 2 x 2 table greater then or equal to 5, then you can use the chi-square test.
2. Any one having in 2 x 2 table less than 5 then you will go to Fisher exact test.
3. If any one cell having zero, then you have to do the Yates' Chi-Square test.
Ok..............
Thanks and all the best for you.
Fisher's exact test is available for r x c tables, too (with r>2 and/or c>2). Only the odds ratio statistic cannot be calculated in non-2x2-tables.
Mehta, C. R. and Patel, N. R. (1986) Algorithm 643. FEXACT: A Fortran subroutine for Fisher's exact test on unordered r*c contingency tables. ACM Transactions on Mathematical Software, 12, 154–161.
Clarkson, D. B., Fan, Y. and Joe, H. (1993) A Remark on Algorithm 643: FEXACT: An Algorithm for Performing Fisher's Exact Test in r x c Contingency Tables. ACM Transactions on Mathematical Software, 19, 484–488.
Patefield, W. M. (1981) Algorithm AS159. An efficient method of generating r x c tables with given row and column totals. Applied Statistics 30, 91–97.
Dear Manu,
there are dozens of alternatives whose suitability and efficacy depend on the features of your data. Are they categorical, ordinal or interval variables? Are they continuous or discrete? Do they shows (un)known distributive profiles to be compared?
For example, if you have continuous, one-dimensional probability distributions KS test can be used to compare a sample with a reference probability distribution (one-sample KS test), or to compare two samples (two-sample KS test) - http://en.wikipedia.org/wiki/Kolmogorov–Smirnov_test
Similarly, Anderson-Darling Tests may help you without any hypothesis about continuity of the underlying distribution - http://www.jstor.org/stable/2288805?seq=1#page_scan_tab_contents
Conversely, if you have a-priori knowledge about the distribution form, other parametric test might suit better your case. For categorical data, again, it's another story.
If you could characterize better your problem I'll be happy to help you.
All the best
Stefano
If any cell value is less than 5 in 2*2 contingency table, Fisher's exact test may be used in place of Chi-square test.
There are many options to Chi-square tests for contingency table.
Some caution is needed, tough:
1) Chi-square test assumes random marginal for columns and rows; when there are few observation in one or more cells (tipically,
Hi Manu,
As i know, there exist discrete form of popular goodness of fit tests which they are
useful for frequency data. To know more about them checkout below links:
http://arxiv.org/ftp/arxiv/papers/1007/1007.5109.pdf
http://works.bepress.com/cgi/viewcontent.cgi?article=1030&context=mike_steele
http://journal.r-project.org/archive/2011-2/RJournal_2011-2_Arnold+Emerson.pdf
good luck.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics