Hi All
I have 4 groups, i tested body weight on day 0 and day 60.
I would like to know that which group had the highest body weight increasing in 60 days intervention compared to others. Does the body weight increasing of this group statistically higher than other groups' increasing.
What statistical test should be used?
Thank you
Hello Yihong Chen
In my opinion, it is appropriate to use analysis of covariance (One-way ANCOVA) in which the zero day weight is considered as covariate.
You have one continuous variable (increase of weight) and a categorical variabile (groups).
It's easy to check if there is no difference among group means (ANOVA test).
If you find that there are statistically significantly differences between them and/or you want a more comprenhensive comparison analysis. One approach is to perform pairwise t-tests (performing multiple t-test between all possible combinations of groups).
You can find attached an example on how you can present the results.
By the way, you can perform a Levene’s test to check if the variabilities between families are equal or at least one groups' variability differs from the rest. And plots (e.g. boxplots) are always a good way to present the results.
Use one-way ANOVA with a post hoc test.
One-way ANOVA will indicate if the weight gains (difference between day 0 and 60) are significantly different between the 4 groups (Yes-P < 0.05 or No-P > 0.05). However, this won't tell you specifically whether Group A is significantly higher than Group B. For this purpose, you will need a post hoc test (given that the One-way ANOVA result is significant).
Post hoc tests determine pairwise differences (e.g. whether any one group is significantly higher than another). There are different post hoc tests available. One of the common ones is Tukey's test. Adding a post hoc test is usually a simple option when performing one-way ANOVA in statistical software.
Some more info about one-way ANOVA and post hoc tests:
https://statistics.laerd.com/statistical-guides/one-way-anova-statistical-guide-4.php
Siqi Li has the best answer above. it is inappropriate to use t tests as a post hoc test or pairwise comparisons as it is giving you the same type of information as the Anova. I am curious why you did not measure body weights at 10, 20, 30, 40 or 50 days to see if there are more acute changes in body weights. Body weights are so easy to measure and take little time.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA