The two assumptions of the Chi-square test are: 1) Independent observations ; 2)For a 2 by 2 table, all expected frequencies > 5.* For a larger table, no more than 20% of all cells may have an expected frequency < 5 and all expected frequencies > 1 . My question is how to interpret the results if the second assumption is violated?
Haha this is from 3 years ago but well... My reference for this is Discovering Statistics using SPSS (3e Field). A. For a 2X2 table with at least 1 2X2 with >20% of cells having
@@jayraldbasan5354 why do we use the p-value under Likelihood ratio in this case? I need to use it in research for bachelor thesis but i cannot find an argument why it is a substitute for chi-squared.
i actually searched in the books you used, but did not find anything that says, that the likelihood ratio can be used that way. could you give your exact source for this assuption? :/ in fact andy field says, that the fisher's exact test can be used for tables bigger than 2x2 as well, which you can do in spss by forcing an exact test. the only case you should not use it is, when your sample is too large. but most of the time your sample is small if the expected values are less than 5.
HI WANNA ASK IF OUR SAMPLE SIZE IS LESS,, IT WILL BE VIOLATED AND THAT MEANS OUR CHI-SQUARE WILL BE NOT SIGNIFICANT ITSNT? AND IT WILL NOT BE USELESS IF WANT TO PROCEED WITH RELATIVE RISK?
Hello, Thank you very much for this video. what is the difference between likelihood ratio and Fisher's exact test. I think both are used for assumptions violated. Thank you again
Oh! I see so when the Chi-Square assumption is violated (25% with p-value = 0.093) we use the Likelihood Ratio significance value (0.023) to determine if there is or is not a relationship between the 2 variables? BUT THEN SHOULD I STILL REPORT THE EFFECT SIZE MEASURE E.G. CRAMER'S V? CRAMER'S V IN THE SPSS OUTPUT STILL SHOWS THE Chi-Square p-value of 0.093???
Hi. In the first example (2 by 2 table with assumptions violated) you say that we can use Fisher's exact test value, but likelihood ratio is also an option, but you preferred Fisher's exact test value. But, I see that there is no value written immediately adjacent to the row labeled "Fisher's exact test", rather, I see the p values 0.743 and 0.426, So, Why SPSS didn't show Fisher's exact test value?
Thank you for this helpful video! In case I have more than 2 groups, and I get a significant results is there some post-hoc test that could indicate which comparison drives the significance?
I'm just wondering if the degree of freedom is gonna influence the chi-square results? The table is a 10*3, more than 50% cells counted less than five, as a result, I used the likelihood ratio to interpret the sig. of the relationship and Crama's V to interpret the effect size, both are statistically significant(p
Dear, How can we report the likelihood ratio. E.g. the Pearson chi square test of independence has to be reported as: χ²(1, N = 167) = 22.01, p < 0.001.
It has been a while since I completed my thesis but here is an example of how I have reported it: The Likelihood-ratio test could accept the null hypothesis with a 95% confidence level (LR (4, n=386) = 4.902, P= 0.297). Hope it helps!
@@bobbymotion hi, if possible, could you please explain what each value means (when taken from SPSS, rather than calculated)? Would be very helpful, thank you.
Hi Brune Thank you for your videos. I have a problem I run an chi-square and checked exactly like this exampel but I do not get Fisher's Exact test. the expected count are so near the counted so the results are violating 10 cells (40.0%) have expected count less than 5 . The minimum expected count is 1.03. I suppose I have to accept the null H but how to write the results? I appreciate your help, thank you
How about my result is (for 2 x 2) "1 cells (25%) have expected count less than 5" and Asymp. Sig. for Pearson is .015? This is significant but the assumption is violated?
Thank you for the nice video. If the assumptions have been violated, can I still use the linear to linear value? If not, then what other statistical method should I use? I have an "Age group - proportion" data which has 0% value in some age groups.
One Question: When u analysed the "Modus", we have to write down as interpretation: Most of the people are... (N=...; ...%) But I DO NOT KNOW WHERE I CAN FIND THE "N".. :/ Could s.o. tell me where I can find that? thanks! :)
So for the second dataset, what if you have significance for Chi-square and don't have significance for Likelihood ratio? Because that's what I have for my data analysis and 70% expected count less than 5
Hi, is anyone able to explain how you would report these findings differently to if the test had not violated the assumptions? using the second example given in the video
Hello! First of all, thank you for the video! Quick question: if in the second example the results had been significant and you had rejected the null hypothesis (if a significant association had been found), how would interpret the effect size?
If I have the same variable in the same population (qualitative, like smoking habit: 1:yes 2:no 3: ex) and I want to see how it evolves on time (baseline-12 months-24 months) can I use Chi-Square?
Hi, I just wanted to ask that when I was running the chi square test it was written that 2 cells have expected count less than 5. The MInimum expected count is 4.00. So, What does that mean please explain!
i am doing quit similar analysis with yours, but the results show that there are 25% expected count less than 5, the likelihood value is 4.528, with df=1, p=0.033. So what can i do with the outcome?
BrunelASK So you mean it doesn't matter whether the outcome ( percentage of expected count) is violet the assumption or not, just look at the P value. Am I right?
by far the best and easiest illustrations I came across online on how to run tests on SPSS, well done. I am doing my desertaion as well.
Did you graduate?
YOU ARE Amazing! Thank you 1 hr lectures in 10 minutes, life saver
So basically, if the 20% rule is violated, we take likelihood instead of pearson as the p-value?
The two assumptions of the Chi-square test are:
1) Independent observations ;
2)For a 2 by 2 table, all expected frequencies > 5.* For a larger table, no more than 20% of all cells may have an expected frequency < 5 and all expected frequencies > 1 .
My question is how to interpret the results if the second assumption is violated?
Haha this is from 3 years ago but well...
My reference for this is Discovering Statistics using SPSS (3e Field).
A. For a 2X2 table with at least 1 2X2 with >20% of cells having
@@jayraldbasan5354 thank you, I'm glad you answered the question
@@jayraldbasan5354 why do we use the p-value under Likelihood ratio in this case? I need to use it in research for bachelor thesis but i cannot find an argument why it is a substitute for chi-squared.
Thank you SO much! This series has really helped me as I interpret the results of my research. Many many many thanks!!!!
i literally love you. helped my market research so much
I love your videos, they are so easy to understand and brief too.
Thank you so much! This really helped with my dissertation data analysis.
Me tooo
What happens if I achieve statistically significant results (p
The last 10 second were like Whaaaaaaaaaaaaaaaaaaat ???
ABSOLUTELY a turn around
I think its wrong interpretation
Thank you. This was very useful. Could you provide a citeable reference on using likelihood ratio when the chi-square assumption is violated?
Diana Nazaryan the best reference i could find so far:
www.ncbi.nlm.nih.gov/pmc/articles/PMC3900058/
thankyou so much for this!
A big Thanks from Sri Lanka really helped
Loved it ! Video was immensely useful for the data analysis for my thesis.
When I run SPSS I don't see a Fisher
Your method and problem picking analytical thing is best
Thanks for this video it is very helpful however what is interpretation in case the 20% and below rule is violated but the p-value is less than 0.05?
same case here and don't know why
When I run my student SPSS, I don't see a "Fisher"
I dont see it ony my SPSS as well, using SPSS version 26 btw
@@Jeffri178 Select 'Exact"
@@adrienjoshua6710 I did stilll dont...
@@Jeffri178 it will be seen only when we interpret for 2*2 table. if its more than that fisher test wont be seen
i actually searched in the books you used, but did not find anything that says, that the likelihood ratio can be used that way. could you give your exact source for this assuption? :/
in fact andy field says, that the fisher's exact test can be used for tables bigger than 2x2 as well, which you can do in spss by forcing an exact test. the only case you should not use it is, when your sample is too large. but most of the time your sample is small if the expected values are less than 5.
HI WANNA ASK IF OUR SAMPLE SIZE IS LESS,, IT WILL BE VIOLATED AND THAT MEANS OUR CHI-SQUARE WILL BE NOT SIGNIFICANT ITSNT? AND IT WILL NOT BE USELESS IF WANT TO PROCEED WITH RELATIVE RISK?
This was so clearly explained. Thank you so much!!
thank u so much.. u cleared my confusion about when to use Fisher's Exact test and when to use Likelihood Ratio.
Hello, Thank you very much for this video. what is the difference between likelihood ratio and Fisher's exact test. I think both are used for assumptions violated.
Thank you again
Thank you!! Your video this process so easy to understand.
you've made my day brighter :X
how about if there is association for chi-square bugger than 2x2? how to run a post-hoc for that?
This helped so much! Thank you.
Very good video!! Thanks a lot. U really helped me out
Fisher does not come out for my table...i have already followed all the steps
same
Oh! I see so when the Chi-Square assumption is violated (25% with p-value = 0.093) we use the Likelihood Ratio significance value (0.023) to determine if there is or is not a relationship between the 2 variables? BUT THEN SHOULD I STILL REPORT THE EFFECT SIZE MEASURE E.G. CRAMER'S V? CRAMER'S V IN THE SPSS OUTPUT STILL SHOWS THE Chi-Square p-value of 0.093???
This is amazing!! thank you
Can we use the chi-square test between an ordinal data type and a nominal data type?
How about when it is exactly 20%? My Pearson's is not statistically significant while my Linear-by-Linear is.
Hi. In the first example (2 by 2 table with assumptions violated) you say that we can use Fisher's exact test value, but likelihood ratio is also an option, but you preferred Fisher's exact test value. But, I see that there is no value written immediately adjacent to the row labeled "Fisher's exact test", rather, I see the p values 0.743 and 0.426, So, Why SPSS didn't show Fisher's exact test value?
Thanks that was really clear.
Thank you for this helpful video! In case I have more than 2 groups, and I get a significant results is there some post-hoc test that could indicate which comparison drives the significance?
thank you so much, I really appreciate your effot
Very helpful. Thank you
"No more than 20% of the expected counts are less than 5 and all individual expected counts are 1 or greater" (Yates, Moore & McCabe, 1999, p. 734)
Nice video! Thanks!
The thing is, what if its super significant (less than 1% p value) but it violates because its more than 20% in that footnote? 😬
Hello!
Can I use the same approach to a (5 x 5) table?
Thanks!
Now i have the same issue, its the same for a 5x5 table?
I'm just wondering if the degree of freedom is gonna influence the chi-square results? The table is a 10*3, more than 50% cells counted less than five, as a result, I used the likelihood ratio to interpret the sig. of the relationship and Crama's V to interpret the effect size, both are statistically significant(p
Dear,
How can we report the likelihood ratio. E.g. the Pearson chi square test of independence has to be reported as: χ²(1, N = 167) = 22.01, p < 0.001.
did you find it?
It has been a while since I completed my thesis but here is an example of how I have reported it:
The Likelihood-ratio test could accept the null hypothesis with a 95% confidence level (LR (4, n=386) = 4.902, P= 0.297).
Hope it helps!
Thanks a lot Bob
@@bobbymotion hi, if possible, could you please explain what each value means (when taken from SPSS, rather than calculated)? Would be very helpful, thank you.
Hi Brune
Thank you for your videos. I have a problem I run an chi-square and checked exactly like this exampel but I do not get Fisher's Exact test. the expected count are so near the counted so the results are violating 10 cells (40.0%) have expected count less than 5 . The minimum expected count is 1.03. I suppose I have to accept the null H but how to write the results? I appreciate your help, thank you
can you please explain the violation of assumptions? I don't understand what it means
How about my result is (for 2 x 2) "1 cells (25%) have expected count less than 5" and Asymp. Sig. for Pearson is .015? This is significant but the assumption is violated?
this is great thank you
What if my pearson vaule is ,094 but the expected count is 40%? What should I do then?
Can it be used for more than 500 respondents and whereby the assumptions has been violated?
Thank you for the nice video. If the assumptions have been violated, can I still use the linear to linear value? If not, then what other statistical method should I use? I have an "Age group - proportion" data which has 0% value in some age groups.
what the "linear by linear association" means? when should we use this?
what does it mean if the assumptions been violated?
One Question:
When u analysed the "Modus", we have to write down as interpretation:
Most of the people are... (N=...; ...%)
But I DO NOT KNOW WHERE I CAN FIND THE "N".. :/
Could s.o. tell me where I can find that? thanks! :)
Thank you for this video. I have run the chi square test on 2x2, one of the results of Asymp. Sig. is 1.000 how this will be interpreted?
So for the second dataset, what if you have significance for Chi-square and don't have significance for Likelihood ratio? Because that's what I have for my data analysis and 70% expected count less than 5
how to do impact analysis suggest statistical tools for the same
Hi, is anyone able to explain how you would report these findings differently to if the test had not violated the assumptions? using the second example given in the video
Hello! First of all, thank you for the video!
Quick question: if in the second example the results had been significant and you had rejected the null hypothesis (if a significant association had been found), how would interpret the effect size?
(sorry but how did you calculate the assumption value (10
i have the same question
If I have the same variable in the same population (qualitative, like smoking habit: 1:yes 2:no 3: ex) and I want to see how it evolves on time (baseline-12 months-24 months) can I use Chi-Square?
Hi +BrunelASK How do i know whats the level of significance i need to compare the significance value to?
wow hang on wait a second you committed a stats taboo! NEVER accept the null hypothesis!
Could we not just look at Cramer's V instead of PHI?
hai. what if on the chi square test table, it says ''3 cells (50%) have expected count less than 5''
what should i do next?
Hi, I just wanted to ask that when I was running the chi square test it was written that 2 cells have expected count less than 5. The MInimum expected count is 4.00. So, What does that mean please explain!
Thank You
Nice
such a dolllll
i am doing quit similar analysis with yours, but the results show that there are 25% expected count less than 5, the likelihood value is 4.528, with df=1, p=0.033. So what can i do with the outcome?
Hi Yunshuo, you interpret it in the same way as shown in the video. If p < 0.05 then the association is significant. Does that answer your question?
BrunelASK So you mean it doesn't matter whether the outcome ( percentage of expected count) is violet the assumption or not, just look at the P value. Am I right?
thanks for violating the assumption. Found no such videos
this video gave me some confusion