Hypothesis Testing Specialist Needed
Budget: $30 – $250 USD
I'm looking for a statistician with expertise in Bayesian analysis to help me test hypotheses.
Ideal Skills and Experience:
- Proficient in Bayesian analysis
- Able to carry out hypothesis testing
- Advanced understanding of statistical methods and tests
- Strong analytical skills and attention to detail
- Excellent communication skills to explain complex statistical concepts in simple terms.
This is my problem:
I am planning an experiment on brain stimulation and effort measured with cardiac response.
These is my variables:
2 between conditions:
- Right stimulation stimulation
- Left stimulation stimulation
2 within conditions:
- Task fixed and easy
- Task unfixed
This is the analysis that I would like to achieve
To evaluate the influence of tDCS on PEP, we will utilize Bayes Factors comparing two hypotheses (Masson, 2011; Richter, 2016; Wagenmakers, 2007).
The null hypothesis (H0) states that there is no effect of tDCS on effort-related cardiovascular indicators, while the alternative hypothesis (H1) suggests that tDCS does affect effort-related cardiovascular indicators. We will employ a contrast model of -3 (right stimulation task unfixed), +1 (left stimulation task unfixed), +1 (right stimulation task fixed and easy), +1 (left stimulation task fixed and easy), to investigate the impact of tDCS on effort intensity. Specifically, this contrast model examines whether right stimulation stimulation in the unfixed task demand, as opposed to left stimulation stimulation conditions and the right stimulation stimulation condition in the fixed and clear task demand lead to higher effort intensity.
My problem: Running contrast in mixed design is not recommended according to some statistician since it mix between and within variance to calculate p value. What I should do according to what I find in literature for testing contrast in mixed design is to have a matrice with two vectors [+1 -1; -1 +1] (this would take into account the experimental design) instead of the one vector above that I want to achieve [-3 +1 +1 +1].
The problem is that the matrice with two vectors is not reflecting exactly what I want to achieve and I will loose power if I run contrast that way.
Can you confirm that my reflection is correct ? Or are there way to test contrast with one vector [-3 +1 +1 +1] instead of the two recommended ? As I’m using Bayes and not ineferential statistics, does this solve the problem ?
I would like a clear answer supported by scientific literature to back up the response.
Ideal Skills and Experience:
- Proficient in Bayesian analysis
- Able to carry out hypothesis testing
- Advanced understanding of statistical methods and tests
- Strong analytical skills and attention to detail
- Excellent communication skills to explain complex statistical concepts in simple terms.
This is my problem:
I am planning an experiment on brain stimulation and effort measured with cardiac response.
These is my variables:
2 between conditions:
- Right stimulation stimulation
- Left stimulation stimulation
2 within conditions:
- Task fixed and easy
- Task unfixed
This is the analysis that I would like to achieve
To evaluate the influence of tDCS on PEP, we will utilize Bayes Factors comparing two hypotheses (Masson, 2011; Richter, 2016; Wagenmakers, 2007).
The null hypothesis (H0) states that there is no effect of tDCS on effort-related cardiovascular indicators, while the alternative hypothesis (H1) suggests that tDCS does affect effort-related cardiovascular indicators. We will employ a contrast model of -3 (right stimulation task unfixed), +1 (left stimulation task unfixed), +1 (right stimulation task fixed and easy), +1 (left stimulation task fixed and easy), to investigate the impact of tDCS on effort intensity. Specifically, this contrast model examines whether right stimulation stimulation in the unfixed task demand, as opposed to left stimulation stimulation conditions and the right stimulation stimulation condition in the fixed and clear task demand lead to higher effort intensity.
My problem: Running contrast in mixed design is not recommended according to some statistician since it mix between and within variance to calculate p value. What I should do according to what I find in literature for testing contrast in mixed design is to have a matrice with two vectors [+1 -1; -1 +1] (this would take into account the experimental design) instead of the one vector above that I want to achieve [-3 +1 +1 +1].
The problem is that the matrice with two vectors is not reflecting exactly what I want to achieve and I will loose power if I run contrast that way.
Can you confirm that my reflection is correct ? Or are there way to test contrast with one vector [-3 +1 +1 +1] instead of the two recommended ? As I’m using Bayes and not ineferential statistics, does this solve the problem ?
I would like a clear answer supported by scientific literature to back up the response.