Experiments are used to study about causal relationship between independent variables and dependent variables. We can manipulate one or more independent variables and then measure their effect on the dependent variables. Then the Experimental Design means a method for collecting data and testing hypotheses related to the causal relationship between the independent variables and the dependent variables
Statistical Tool for Agricultural Research (STAR) is a computer program developed by International Rice Research Institute (IRRI) for data management and basic statistical analysis of experimental data. Experimental design with STAR not only used for agricultural purposes but you can also use it for business or government purpose so this application can really help you as data analyst, UX researcher, business analyst, government researcher or any other position related to analytics role. Specifically for this experimental design will be closely related to the ANOVA test
Why use STAR ?
– STAR provides layout of experimental designs commonly used in research
– User friendly
– Open source and free access
– STAR provides descriptive analytics, ANOVA, T-test, Chi-square, Multivariate Analysis, Correlation, Regression
Completely Randomized Design (CRD)
Completely randomized design is the simplest type of experimental design . The reasons behind the use of a completely randomized design are as follows:
1. The experimental unit used is homogeneous or there are no other factors that affect the response outside of the factors being tried or studied.
2. External factors that can affect the experiment can be controlled. For example, an experiment conducted in a laboratory
Randomized Complete Block Design / Randomized Block Design (RBD)
A randomized block design is a type of experiment where participants who share certain characteristics are grouped together to form blocks, and then the treatment (or intervention) gets randomly assigned within each block
ANOVA Test with STAR
The following are the steps in conducting ANOVA testing with STAR :
– table format may not be in the form of cross tabulation
– save data in CSV format
– import data from Project Explorer tab : Data >> Import Data
– then select Analyze >> Analysis of Variance >> Completely Randomized Design / Randomized Complete Block Design
Test Result and Interpretation
Factor hypothesis ANOVA Model Result Interpretation :
p-value = 0.0076
p-value < 0.05, reject the null hypothesisHo: The population means of the factor are equal
Ha: There is one or more population means of the factor are not equal
The options tab also provides functions to perform descriptive analysis, homogeneity test (Bartlett Test) and normality test (Shapiro-Wilk)
Assumption Test Result Interpretation :
homogeneity test p-value = 0.7324, normality test p-value = 0.7256
p-value < 0.05, reject the null hypothesisBased on these tests, the data is proven to be homogeneous and normally distributed
In this test, post hoc testing is also automatically carried out with Least Significant Difference (LSD) as the default for CRD but we can use another post hoc test such as Duncan Multiple Range Test (DMRT) or Tukey’s Honest Significant Difference (HSD)
Post Hoc Test Result Interpretation :
– B and D are not significantly different (LSD and DMRT)
– C and D are not significantly different (LSD)
– A and D are not significantly different (LSD)
– A, C and D are not significantly different (DMRT)
Based on all the Experimental Design that have been carried out, it was found that there are variety with different result. Variety B has the highest mean score, B and D are not significantly different but A and C are significantly different with B
That’s the application of the Experimental Design with STAR, hopefully it’s easy to understand by everyone who needs this explanation
Read the full article here
