Great explanation. Only one thing still unclear. Shouldn't the p-values be multiplied by 6 instead of 3 since there are 6 separate analyses in this crosstabulation?
Thank you very much for the great presentation. I made a cross tabulation for two variables (one of them with 3 categories and the other with 5 categories). Now, I need to estimate the adjusted Z-value. The question here, should I do that for every category of the independent var."the one with 3 categories"? And If I do so, What will be the used degree of freedom for estimating Chi-square?
First thank you a lot for your informative videos ,,plz may I ask 1-if after I multiply the p value with number of analyses [ it was 6 in my data] I got an adjusted p value of 5.03 ,,,is this value of 5.03 logic for a p value to be written in a scientific paper ,,I mean I know that p values must be between 0 to 1 or this is not right idea , 2-also how can I know the degree of freedom to use it to calculate the p value from the chisquared value ,, I assumed it as 1 as u did so is it usually 1 or I should use another number and how to know it by SPSS ,,,many thanks
Hello. I successfully calculate the adjusted p value, thanks to your video. However, I have a contradicting Bonferroni adjustment from SPSS chi square output with the adjusted p-value i have calculated. In the SPSS output it is stated that there are 2 different subscript across the column which indicating significant difference..but when I calculate the adjusted p-value, it is concluded that there are no significant difference. Is this possible? How can I report the result?
Have you checked chapter 4 of the (free) textbook? That's where I discuss this analysis: www.how2statsbook.com. There are also references in the chapter.
Is there a way that you can see if the 3 ad types differ from each other. For example, is search significant different from purchase and is search significantly different from trending? Is that a pairwise comparison? How would that be done? Hope that makes sense. Thank you!
Not that I'm aware of (but check the chapter, as I might be forgetting something). Many people would follow-up with a series of 2x2 crosstab analyses to isolate the proportion comparisons. I do talk about that in the textbook.
There's not easy way that I know. I prefer to simply multiply the p-values by the number of analyses and still use alpha of .05 to demarcate a significant effect.
Hi, great video, I´m reading a paper where you can find a contingency table with a RR witch value is different than the one I calculate by hand (that is the same that epi info gives me), the authors mention they used the Bonferroni correction. Could this difference be explained by this correction?
The actual relative risk value should not change, irrespective of whether a Bonferroni correction was applied, as this sort of correction impacts only the p-value.
@@how2statsbook477 thank you!. last question I promise, all the ratios presented by this trial are smaller than usual, the statistical analysis section describes GLM, nonparametric methods like Mann Whitney U, also mention the Poisson distribution and negative binomial distribution. Do you have any idea about how did they calculate those ratios?. I´m reading you book , congrats!!!, I liked it very much.
Fantastic! So many thanks!
Thank you for a great lecture on adjusted residuals!
Thank you so much, it was helpful to use my question!
Great explanation. Only one thing still unclear. Shouldn't the p-values be multiplied by 6 instead of 3 since there are 6 separate analyses in this crosstabulation?
Thank you very much for the great presentation. I made a cross tabulation for two variables (one of them with 3 categories and the other with 5 categories). Now, I need to estimate the adjusted Z-value. The question here, should I do that for every category of the independent var."the one with 3 categories"? And If I do so, What will be the used degree of freedom for estimating Chi-square?
I think the last step, we should divide p-value by 3 to get adjusted p-value, not multiplied by 3. Please correct me if I'm not correct.
First thank you a lot for your informative videos ,,plz may I ask
1-if after I multiply the p value with number of analyses [ it was 6 in my data] I got an adjusted p value of 5.03 ,,,is this value of 5.03 logic for a p value to be written in a scientific paper ,,I mean I know that p values must be between 0 to 1 or this is not right idea ,
2-also how can I know the degree of freedom to use it to calculate the p value from the chisquared value ,, I assumed it as 1 as u did so is it usually 1 or I should use another number and how to know it by SPSS ,,,many thanks
1: report the largest p-values as .999
2: the df is always 1 for this analysis.
@@how2statsbook477 thank you a lot
Hello. I successfully calculate the adjusted p value, thanks to your video. However, I have a contradicting Bonferroni adjustment from SPSS chi square output with the adjusted p-value i have calculated. In the SPSS output it is stated that there are 2 different subscript across the column which indicating significant difference..but when I calculate the adjusted p-value, it is concluded that there are no significant difference. Is this possible? How can I report the result?
I am not sure why we use degree of freedom =1... Also where can I find explanation regarding Chi sq = Z^2?
Have you checked chapter 4 of the (free) textbook? That's where I discuss this analysis: www.how2statsbook.com. There are also references in the chapter.
Is there a way that you can see if the 3 ad types differ from each other. For example, is search significant different from purchase and is search significantly different from trending? Is that a pairwise comparison? How would that be done? Hope that makes sense. Thank you!
Not that I'm aware of (but check the chapter, as I might be forgetting something). Many people would follow-up with a series of 2x2 crosstab analyses to isolate the proportion comparisons. I do talk about that in the textbook.
Can you give some advice on how to adjust alpha values at 0.005 in SPSS before I perform a Chi-aquare analysis?
There's not easy way that I know. I prefer to simply multiply the p-values by the number of analyses and still use alpha of .05 to demarcate a significant effect.
Hi, great video, I´m reading a paper where you can find a contingency table with a RR witch value is different than the one I calculate by hand (that is the same that epi info gives me), the authors mention they used the Bonferroni correction. Could this difference be explained by this correction?
The actual relative risk value should not change, irrespective of whether a Bonferroni correction was applied, as this sort of correction impacts only the p-value.
@@how2statsbook477 thank you!. last question I promise, all the ratios presented by this trial are smaller than usual, the statistical analysis section describes GLM, nonparametric methods like Mann Whitney U, also mention the Poisson distribution and negative binomial distribution. Do you have any idea about how did they calculate those ratios?. I´m reading you book , congrats!!!, I liked it very much.