Hi, I struggle to understand what kind of statistic should I use for my experiment. I have a 2x3 design (two cat variables, one has two levels, and another has three), but I have eight groups of participants and each group exposed to its unique combination of those variables.
Just as an example: let's say one variable is the time of the day - morning, day, evening.
Another variable is the temperature - cold, warm.
The dependent variable will be continuous - what time of the day we would see more bees.
We have eight groups of participants who were counting bees in the following conditions:
1. cold morning, cold, day, cold evening
2. cold morning, cold day, warm evening
3. cold morning, warm day, cold evening
4. cold morning, warm day, warm evening
and four more for the warm mornings.
The question is what time of the day we would see more bees and how the temperature would moderate the effect of the time on the quantity of the bees.
It's loos like I need something more complex than a simple factorial analysis.
Max Kulbida for your experiment as 2x3 design (two cat variables, one has two levels, and another has three) you have to use (Mainly studied and designed in Statistics) is Factorial Experimental Design. In this Design we mainly deal variable(s) have some other factors. I am strongly recommend to apply it.
Good Luck.
Hi Max Kulbida ,
Use cross-tabs with X2 (Chi-squared) to test the independence hypothesis. But mind the conditions of that test.
Thanks,
Ali
---------------------------------------------
Recent publications:
Conference Paper Wartime Crisis Perceptions Predict Attitudes Towards Cheatin...
Article Email is Evil! Behavioural Responses towards Permission-base...
Article Validating a New TQM-Benchmarking Measurement Model in an In...
Article Assessing Patients’ Perception of Health Care Service Qualit...
Article Total Quality Management Boosters and Blockers in a Humanita...
Article Political Advertising Effectiveness in Wartime Syria
Hello Max,
I see at least two ways to handle the data in an anova framework.
Method A: Use a 2 x 3 factorial design, as you suggest, with 4 sets of observations (bee counts per observation period from group X) per cell. To feel confident with this choice, you must presume that groups do not differ systematically in their likelihood to over-count or under-count bees in an observation period.
Method B: Add a 3rd factor, group (assuming that your group memberships are constant across the three observation periods assigned to each group), so that the "full" model would have been a 8 x 2 x 3 design. Unfortunately, you'll have only 24 observations, as opposed to the 48 that a full factorial design would expect (if each cell is to have 1 data point). So this is like an incomplete randomized block design. You can still estimate main effects and the two-way interaction of time and temperature...it's obviously not possible to estimate all possible two-way interactions and a three-way interaction.
Method C: As for method B, except add observation period (1st, 2nd, 3rd) for each group as a factor. That leads to an even more incomplete block design, though with suppressed interactions, can still allow for tests of main effects.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Filing