Validity of a experiments
First we consider the validity of the experiment the are three main validates. statistical validity, internal validity, and external validity
- In statistical validity three main hypothesis are made – normality, Homogeneity of the variance, independent of the variance
- Internal validity is the how far relationship between independent and dependent variables.
- External validity – results of the experiment can apply to the wide range of experiments.
Observation and the Experiments
In observation base studies observe subject and measure variables of interest without applying treatment.
In experiment investigators apply treatments to the experiment units, and the observe the effect to the subject, in a randomized experiment investigators control the assignment of treatment to experiment units using the chance mechanism (such as random number generators)
Two main factors effect the quality and the quantity of the information of the data, collected for the research.
- Size of the samples
- The magnitude of the background noise.
Most important considerations of the picking a design
- Choosing how many independent variables
- how many level of independent variables
- how to assign experiments unit to levels of independent variables
Different Experimental Designs
- CRD (Completely randomized Design) – the simplest of all designs, treatments are allocated to the experimental units completely at random, adjacent subjects could have potentially same treatments.
- RCBD (Randomized Complete Block Design) – Blocking or Grouping of subjects with similar characteristics.
- LSD (Latin Square Design) – equal number of rows, columns and treatments
- Stratified sampling – population is divided into levels that represent clearly defined groups of units within the population
- Cluster sampling – modified basic random sampling design also used in heterogeneity of the population.
- Systematic sampling – samples are equal space either spatially or temporary.
Scientific Experiments are conducted
- To confirm certain theories about natural phenomena
- To find out relationships between or among certain variables
Sample data do not contain complete information due to background noise of the data.
Background noise due to;
- Variability within groups
- Individual differences
- Minor differences in experimental sessions
Biological data will always be “NOISY”