I have two groups (A and B). I want to determine statistical significance between the two groups for each individual item (7 items measuring participation). This way I can be able to tell whether item 1 made more significant impact on overall participation in group A than B. The data is normally distributed and homogeneous variance was obtained for overall participation. It’s a continuous data.
Can this be done?
What is the most appropriate statistical test to do this?
Am not thinking of a paired sample t test for each item since we are testing for significance between the groups for the same item.
I have data for each item as well as averages for each item that looks like this:
Group A. Group B
item 1. 14. 36
item 2. 56. 44
Overall Part. 68. 73
Thank you all
A commonly used test for statistical significance is the t-test. It assesses whether the means of two groups are statistically different from each other. For larger sample sizes or when assumptions of normality and equal variances are met, the z-test (or normal distribution test) can also be appropriate. These tests help determine if observed differences between groups or conditions are likely due to real effects or simply due to random chance.
Hello Fati Osuman. I am trying to understand what your data looks like. My guess is that it looks something like this:
Grp item1 item2 item3 item4 item5 item6 item7
1 33 23 30 64 32 49 29
1 78 37 46 31 37 49 49
1 33 45 37 62 52 54 48
1 55 59 45 52 37 38 63
1 65 60 65 52 42 32 64
2 68 34 61 64 62 54 57
2 65 52 36 53 55 69 52
2 58 38 43 45 68 38 53
2 57 49 57 53 62 53 28
2 59 43 43 56 64 53 71
Am I correct? If so, how many observations do you have in each group? And how are the 7 items measured? Is it sensible and defensible to use means and SDs of the 7 items for descriptive purposes? If not, why not? Thank you for clarifying.
I'm not sure I understand the question... Why isn't it just 7 independent t-tests ?
Or, is it the case that have only one observation per item per group ?
Thank you all. Yes that is how the data looks like. I have 56 observations in group A and 31 in group B. It’ is continuous from 0- 100 @Sal Mangiafico Bruce Weaver
Thank you for clarifying, Fati Osuman. If means & SDs make sense for description, then you could use t-tests, perhaps with some kind of adjustment for multiplicity (e.g., Benjamini-Hochberg). Given the fairly large difference in sample sizes for the two groups, you may wish to use the Welch-Satterthwaite t-test (because the greater the difference in the sample sizes, the less robust Student's t-test is to variance heterogeneity). HTH.
PS- You can read about the Benjamini-Hochberg FDR method here:
http://www.biostathandbook.com/multiplecomparisons.html
- Badges
- Science topic
- Similar topics
- Chemistry
- Biochemistry
- Biochemical Techniques
- Blotting Techniques
- Immunoblotting
- Dyes