Help involving Chi squared test

  • Context: Undergrad 
  • Thread starter Thread starter mmmboh
  • Start date Start date
  • Tags Tags
    Chi Test
Click For Summary

Discussion Overview

The discussion revolves around the challenges faced when performing a chi-squared test, specifically addressing the issue of having an expected value of zero in one of the rows. Participants explore potential solutions and implications of this situation within the context of statistical analysis.

Discussion Character

  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • One participant expresses concern about dividing by zero in the chi-squared formula when the expected value is zero and questions whether to exclude that row from the summation.
  • Another participant suggests combining cells with expected counts below 5 to ensure all cells meet the requirement, advising that degrees of freedom should be based on the combined cells.
  • A different participant questions the feasibility of combining rows, noting that their expected values are generally low and expresses uncertainty about alternative tests, stating that they have not been taught any.
  • One participant challenges the validity of having an expected value of zero, suggesting that if one row or column total is zero, the expected value cannot be zero and questions the calculation of expected values.

Areas of Agreement / Disagreement

Participants do not reach a consensus on how to handle the situation of having an expected value of zero. There are competing views on whether to combine rows or consider alternative tests, and some participants question the validity of the expected value calculations.

Contextual Notes

There are limitations regarding the assumptions made in the calculations of expected values, particularly concerning row and column totals. The discussion does not resolve whether the expected values were calculated correctly.

mmmboh
Messages
401
Reaction score
0
URGENT help involving Chi squared test!

Hello, I am doing a chi squared test for a project due soon and everything is fine, except for one thing...for one of my rows the expected value is zero!...and the equation is (O-E)^2/E...but I can't divide it by zero because that will give infinity...so what do I do if the expected value is zero? not include it in the summation?...

Help is greatly appreciated!
 
Physics news on Phys.org


mmmboh said:
Hello, I am doing a chi squared test for a project due soon and everything is fine, except for one thing...for one of my rows the expected value is zero!...and the equation is (O-E)^2/E...but I can't divide it by zero because that will give infinity...so what do I do if the expected value is zero? not include it in the summation?...

Help is greatly appreciated!

A common suggestion is to combine cells for which the expected count is below 5: pool cells until every cell has expected count >=5, then do the test. when you do this base the degrees of freedom on the number of cells after combination, not the original number.

you could also try a different test.
 


Hm I'm not quite sure I follow, I am just suppose to combine rows together? I don't think that would work in my case, and my expected values for any given row are low anyway...is there something else I can do?...I don't think there is an alternate test, at least not that we have been tought.
 


mmmboh said:
Hello, I am doing a chi squared test for a project due soon and everything is fine, except for one thing...for one of my rows the expected value is zero!...and the equation is (O-E)^2/E...but I can't divide it by zero because that will give infinity...so what do I do if the expected value is zero? not include it in the summation?...

Help is greatly appreciated!

unless one row total or column total is exactly 0, then the expected cannot be 0. is the expected calculated right?

expected=[row total]x[column total]/[table total]

if one of the totals indeed 0, then the category containing that is completely invalid!
 

Similar threads

  • · Replies 23 ·
Replies
23
Views
3K
  • · Replies 11 ·
Replies
11
Views
4K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 6 ·
Replies
6
Views
1K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 5 ·
Replies
5
Views
4K
  • · Replies 5 ·
Replies
5
Views
5K
  • · Replies 5 ·
Replies
5
Views
9K