Choosing the right statistical test

Ask questions about projects relating to: biology, biochemistry, genomics, microbiology, molecular biology, pharmacology/toxicology, zoology, human behavior, archeology, anthropology, political science, sociology, geology, environmental science, oceanography, seismology, weather, or atmosphere.

Moderators: AmyCowen, kgudger, MadelineB, Moderators

Locked
heatlamp
Posts: 1
Joined: Fri Oct 28, 2016 2:25 pm
Occupation: Student

Choosing the right statistical test

Post by heatlamp »

My project has the following set up:

1 Control Group
6 Test groups

Each group has 20 subjects (mealworms). I am looking to see how many of each group develop into pupae and then progress to adult.

My endpoint is percent recovery.

Scientific journals have used the ANOVA test to compare the groups, but this does not seem correct because my data is nominal in nature - alive (1) vs. dead (0).

I am wondering what statistical analysis would be the most appropriate?
catherineM99
Former Expert
Posts: 50
Joined: Sun Aug 14, 2016 4:41 pm
Occupation: Student

Re: Choosing the right statistical test

Post by catherineM99 »

Dear heatlamp,

Survival Analysis is a part of statistics that focuses on the length of time until an event occurs, and specifically analyzes the length of time until the death of a biological organism. This sounds appropriate for your experiment; I would do a Google search to discover specifically what kind of survival analysis fits best with your experiment. The Wikipedia page for Survival Analysis does a good job at summarizing the different methods: https://en.wikipedia.org/wiki/Survival_analysis

Furthermore, what is your independent variable/treatment? This is essential to identify and include in order to have a factor that influences the mortality rate of the pupae. Secondly, in statistics the individuals that the experiment is being tested on are only called subjects if they are human and are otherwise considered experimental units.

Hope this helps!
~Catherine
heidigr
Former Expert
Posts: 6
Joined: Mon Sep 19, 2016 2:53 pm
Occupation: Other Adult

Re: Choosing the right statistical test

Post by heidigr »

Hello,

In addition to what Catherine said, seems like a simple Chi-2 test could be done for group vs status.
Also consider comparing pairwise comparisons between control group and each of the 6 test groups.
For these latter 6 pairwise comparisons multiply the p-value by 6, this is a crude Bonferroni adjustment for multiple comparisons.

~Heidi
Locked

Return to “Grades 9-12: Life, Earth, and Social Sciences”