Download Statistical Testing in Biology: Hypothesis Testing and Significance - Prof. Michael Angill and more Study notes Biology in PDF only on Docsity! Testing Hypotheses • Comparing samples • Statistical tests • Biological significance Strong Inference 1. Devise alternative hypotheses. 2. Devise a crucial experiment. 3. Carry out the experiment so as to get a clean result. 4. Recycle the procedure, making sequential hypotheses to refine the possibilities that remain. Platt (1964), Science, 146: 347-353 Let’s look at the data from lab. Factors affecting Biological Conclusions • Natural variance of the data – For a given sample, a greater variance in the trait increases the probability of observing a false relationship. • Number of replicates – For a given variance, a larger sample from the population decreases the probability of observing a false relationship. Degrees of freedom (df) The number of independent pieces of information used to estimate a parameter. In practice, the df is the # of observations minus the # of parameters estimated for a calculation. Examples Mean: df = N Variance: df = N – 1 t statistic (used to compare two means): df = ??? The P value • Probability of obtaining the difference between sample means (or a more extreme difference), when the population means do not differ. • Historically, scientists have based their conclusions on an arbitrary cutoff: – If P < 0.05, one would conclude the samples come from populations with different means. – If P > 0.05, one would conclude the samples come from populations with similar means. The P value X To get a P value, most people assume their data follow a normal distribution (or bell curve). Recall, the standard deviation tells us about the spread of data that are normally distributed. Tests to Compare Means • 2 means, normal distributions – t test • >2 means, normal distributions – F test • 2 means, other distributions – U test Which test would you use for your fly data? Testing our Hypotheses Let’s do some statistical tests. Statistical significance vs. Biological significance • Any two populations will differ in every trait. To determine statistical significance, you merely need to measure a trait with enough precision in enough individuals. • Not every difference between populations is important. To determine biological significance, you need to have some idea of how the difference affects the organisms. Let’s explore the effect of sample size.
