Course content
WELCOME
TO
BIOSTATISTICS!
Course content
1. Overview and Definitions
2. Experimental design
3. Data types and representation of data
4. Measurement of central tendency and
dispersion
5. Normal distribution
6. Hypothesis testing
7. The t distribution and comparison of means
Course content
8.
Calculation of 95% confidence intervals
9.
Basic analysis of variance
10. Correlation and regression techniques
11. chi-squared tests
12. Mann-Whitney U tests (non-parametric
methods)
13. Multivariate testing
Definitions
Population:
The entire collection of items that forms the focus of a study.
Random sample:
Subset of the population that is randomly selected, and
analysed appropriately for the information it contains. During
selection, each item in the population had an equal opportunity
to appear in the sample.
An experiment:
A deliberate course of action that is aimed at satisfying carefully
stated objectives.
Definitions
Response variable:
That which is actually measured, as a function of the
treatments
Parameter:
A measurement on a population that characterises one of it’s
features. For example, the average of a specific response
variable for a population - the average height of a group of
people, the average weight etc.
Definitions
Experimental unit (EU):
• The smallest division of experimental material to which a
treatment can be assigned in a single act of randomization.
• Is a physical entity, e.g., cow, pot in the greenhouse, area of
indigenous flora
• or can be a group of individuals, e.g., litter of pigs, number of
plants in a tray.
• The EU is the smallest entity receiving a single treatment,
provided two such entities could receive different treatments.
Definitions
Sampling unit (SU):
• The unit of experimental material on which an observation is
recorded.
• It is the physical entity on which the measurement is made.
• The experimental unit is often the same as the sampling
unit. Often, the EU may be divided into two or more SU’s.
EU = unit to which treatments are applied
SU = unit upon which response variable is measured.
Definitions
Experimental error:
• Variation which naturally exists among experimental units
treated alike.
• Is a characteristic of all experimental material.
• It is the presence of random variation which makes statistics
an integral part of all research endeavors.
• Is basically a collective term often used to describe variation
resulting from all sources of variation unaccounted for in the
experiment.
Definitions
Sensitivity:
The ability of an experiment to detect real differences.
Confounding:
The mixing together of effects. Effects of two independent
variables on a dependent variable are said to be confounded
when they cannot be distinguished from one another in the
statistical analysis. Confounding often obscures the true effects
of treatments on the response variable.
Definitions
Replication:
• The assignment of more than one EU to the same treatment
(or treatment combination).
• Is one complete set of treatments.
• Increasing the number of replications results in greater
sensitivity by reducing the standard error of the difference
between treatment means.
Definitions
Randomization:
• This is the cornerstone of experimental design, and assures
validity of the estimate.
• Refers to assignment of treatments to EU’s so that all units
have an equal chance or receiving a treatment.
• It protects against systematic bias caused by subjective
assignment of treatments, and also validates the statistical
assumption that observations (or errors) are independently
distributed random variables.
Definitions
Local control:
The techniques used to reduce or control experimental error,so
that the over-all precision of the experiment is increased.
This can be done by:
1) handling experimental material so that effects of inherent
variability are reduced,
2) Refining experimental techniques for administering
treatments and measuring responses
3) the most appropriate design structure is selected.
Definitions
Why should we study and use stats?
There are two main purposes:
DESCRIPTIVE
INFERENTIAL
We generalise
We summarise the important
characteristics of a set of data
from a set of data
describing a sample,
to the larger population