I want to analyze a performance task over a period of time in animal study, where a particular group of animals is observed for learning curve over a period of 5 days. Is there any trend analysis statistics or other similar analysis that I can perform for this type of dataset? I would like to know how significant the treatment group is as compared to untreated group. Any input or suggestions are welcome.
Hi Shah,
Probably you could use the repeated measures Two-Way ANOVA. Than you will take in account the performance of the particular group of animal along time and also you can compare both groups. Use as independent variables Time (5 days) and group treatment. You also can add the post-hoc of Bonferroni to compare both groups in each day.
I am interested in the answer to this question as well. Wouldn't a repeated measures ANOVA have too many variables, assuming you group performance by day? Imagine we had 20 days instead of 5.
Are there other ways to approach this?
It may be possible to apply growth curve modeling now used in parts of developmental psychology.
Hello Espen,
I don't think so. If you increase the number of days, you would increase the number of dependent variables (dependent variables you can't control because is the number of observation that you have, so you have to include all), but the independent variables that you use for compare the dependent variable will continue to be two, time (20 days) and treatment.
About the Growth Curve, I never used before, so I can't add useful information about it.
Regards
Hey,
you could also compute the slope of the learning curves for each animal and then compare the values between groups with a simple t-test. This way, you can asses if the animals of one group learn faster (steeper curve or higher slope) or if they start with higher performance levels at the beginning already.
But the mixed two way ANOVA would be the classical way, I think...
Best wishes,
Sarah
Fit a sigmoidal and compare parameter estimates. Here's an example from nicotine self-admin in rats: https://www.researchgate.net/publication/221746104_A_new_criterion_for_acquisition_of_nicotine_self-administration_in_rats
Article A New Criterion for Acquisition of Nicotine Self-administrat...
Mixed effect analysis provides a more parsimonious approach because you are concentrating on slope/intercept instead of day-by-day comparison: see https://www.researchgate.net/publication/224282503_Mixed_effects_modeling_of_Morris_water_maze_data_Advantages_and_cautionary_notes
This is appropriate for any learning curve, not just Morris water maze, of course.
Article Mixed effects modeling of Morris water maze data: Advantages...
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA