Experimental Design - PowerPoint PPT Presentation

1 / 13
About This Presentation
Title:

Experimental Design

Description:

Title: Experiments Author: Seed Last modified by: Seed Created Date: 1/29/2004 9:07:10 AM Document presentation format: On-screen Show Other titles – PowerPoint PPT presentation

Number of Views:96
Avg rating:3.0/5.0
Slides: 14
Provided by: See100
Category:

less

Transcript and Presenter's Notes

Title: Experimental Design


1
Experimental Design
  • Dr MP Seed
  • m.p.seed_at_qmul.ac.uk

2
Objectives
  • To be able to design an experiment taking into
    account
  • Suitable controls
  • Suitable comparisons
  • Suitable statistics
  • Error reduction
  • Power
  • If animals are used, benefit/suffering
  • Final Data Presentation

3
Science is a process of discovery through
hypothesis testing
4
Formulation of Hypothesis or objective Experimenta
l Unit Control Variability Control of
bias Treatment choice (independent variables) End
Point choice (dependent variable) Choice of
design Sample size Statistical analysis
planning Pilot study Protocols and SOPs (Standard
Operating Procedures) Statistical
Analysis Interpretation Data Presentation Choice
of model
5
Common Errors Lack of design Ad Hoc
approach Historical controls Clue variable
numbers Factorial Design unbalanced Experiment
size Blocking Genetic heterogeneity
(Inbred/outbred animals strains) Statistical
analysis errors Assay errors
6
Statistics Parametric The Bloody Obvious
Test (t-Test) ANOVA Followed by Post Hoc test
Bonferroni (No more than 4 comparisons
unless corrected Neuman Keuls test (does
not give confidence limits) Tuley Kramer
(Gives confidence Limits) Comparison of groups
to one control- Dunnetts Test
7
T-test Compares two groups Reduced reliability
with multiple comparisons Bonferroni also, use
correction factor or do not use for more than 4
comparisons.
8
ANOVA Post Hoc Testing Bonferroni, Newman
Keuls, Tukey Kramer Parametric
9
  • Non Parametric ANOVA
  • Kruskall Wallace (NOT Neuman Keuls as mentoned
    in lecture!!)
  • Followed by Post Hoc test eg Dunns test

10
Power analysis Effect size 10-20 Standard
deviation Significance level plt0.05 Power
level 80-90 Sample size unknown? Alternative
hypothesis 1 / 2 tailed?
11
Useful Packages Graphpad Prizm
(Desktop) Minitab (Desktop) SPS Contain useful
guides
12
Reading Material
13
  • Handout
  • Basic Principles of the Design of Animal
    Experiments. Festing et al, The Design of Animal
    Experiments, Laboratory Anumal Handbooks No14
    Chapter 1.
Write a Comment
User Comments (0)
About PowerShow.com