Hello,
I have read a paper about how to carry out meta-analysis in 5 datasets of microarray. The method they used is combing effect sizes. After they calculate the pooled effect size, they compare it with the standard error ( in the paper, they said "study-specific effect sizes were combined to obtain the pooled effect size and its stand error using the random effects inverse-variance technique. The z-statistic was computed as a ration of the pooled effect size to its standard error for each gene, and the result was compare with a standard normal distribution to obtain a nominal P value".
My question is: why do they need to compare effects size and standard error (z-statistics). How to calclulate the standard normal distribution to obtain a nominal P value? Why we have to compare beween that Z-statistic and the nominal P value? (If it is bigger or smaller, what does it mean?)
Please, help me to understand this. Thank you so much
Hello
I`m waiting the answer from an expert.
But regarding Z test , the following :-
https://www.researchgate.net/post/Mean_or_Median_which_is_better_for_calculating_statistical_significance_through_ANOVA_in_excel
By an answer from professor Dhritikesh Chakrabarty
Z denotes the standard normal variate. Z test is nothing but normality test. It is based on the area property of the standard normal distribution.
From the theory of normality approximation, it is known that if a data set is large (i.e. data size is more than 30) then the probability distribution of the data set can be approximated by normal probability distribution.
Again the distributions of chi-square statistic, t statistic etc. approach normal distribution as the respective degrees of freedom ( equivalently the respective data sizes) become large.
Hence, z test can be an alternative of the other tests if the data sizes are large.
-------------------------
This was my answer
Z test is used to difference between mean when researcher know the sigma ( SD of the population ) , old wrong concept in 1970 said if you make all tests on all sample then it means you can apply it to all population. A z-test is a statistical test used to determine whether two population means are different when the variances are known and the sample size is large. The test statistic is assumed to have a normal distribution, and nuisance parameters such as standard deviation should be known in order for an accurate z-test to be performed.
But the concept depends on probabilities not on samples.
Z test is used in design as design for 6 sigma (DFSS) , design of experiment , others.
Z test is like T test and F test need hypothesis 2x2 table that include types of error.
P value will be significant if it is < 0.05
Also please check this link:-
https://www.danielsoper.com/statcalc/calculator.aspx?id=48
Thank you for your attention and your idea. Well, the answer for my question is there are some misunderstanding when at first I think that the ratio " effects size and standard error " is used to compare between them. Actually that ratio is used to calculate z statistics which is used as a significant test for REM.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size