Please enable JavaScript.

Coggle requires JavaScript to display documents.

design of experiment - Coggle Diagram

- - - - antibody labelling: positive/negative standard; threshold, background; normalization
      - electrophysiology
        
        synaptic current amplitude; area; decay; time to peak; frequency of occurence
      - WB
        
        positive/negative threshold; band intensity; MW range
    - - screening experiment
        
        test the procedure of planned analysis
    - - limited number
      - only to improve quality not quantity
  - - - no treatment/placebo
        
        select either and use alternatively
      - positive control
      - negative control
      - after control
        
        for reversible effects, eg. washout of drug
    - - decide befor experiment
      - same sample before and after treatment
      - 2 samples treated in exactly the same way
      - normalized paired
        
        eg: normalized EPSP amplitude
  - - - avoid selection bias
    - - ideally random
        
        use whenever possible
        
        variants of random sampling
        
        stratified random sampling
        
        bootstrap sampling
        
        same chance for all elements to be sample
      - systematical
        
        number elements and take every eg: 5th
        
        results in simple random list
        
        problematic if variable periodic or list effect
        
        results only valid for this system
        
        document your system meticulously
      - cluster sampling
        
        geographically
        
        quarters or blocks in cities
      - avoid
        
        grab 1st mouse you get
        
        interview next person you meet
        
        cultures leftover by others
      - sample size
        
        never below 4
        
        depends on variability of parameter measured (may need screening experiment
        
        n = 16(STD/difference)^2, alpha 5%, power of 80%
        
        always set sample size in advance
        
        lower random false positive rate
        
        temporal multiple comparison are possible and can be corrected for
  - - - collect screening data
      - check
        
        STD of appropriate control group
        
        take a look at sample data
        
        variability
    - - establish on your own
  - - - usually 4-8 per group
      - effect < scatter n=64
      - effect > scatter n=4
    - - if smaller than scatter
        
        unlikely matter
      - small scatter, stable system or large effect
        
        likely biologically important