Experimental Simulating Program ESP Experimental Simulating Program USER GUIDE Prepared by: Dr. Ricaro V. Nunes, Universidade Estadual do Oeste do Parana Dr. Gene M. Pesti, University of Georgia Msc. Jomara Broch, Universidade Estadual dp Oeste do Parana Version 1.0

Determining experimental power for planning purposes: Introduction Determining experimental power for planning purposes: By it's very nature is a risky business; Experiments are planned because we don't know what the results will be! In planning, we hope that the means and variances of control and treated groups will be similar to previous experiments; There is always the possibility that any imposed treatments will affect individuals differently, changing the variation between treatment groups.

Experimental errors often occur even before field trials: Introduction Experimental errors often occur even before field trials: One of the main mistakes made may be related to the number of repetitions that we will be used; To assist in the proper choice of the minimum number of replicates per treatment in a given assay: - a spreadsheet has been developed ; Remember this spreadsheet is a tool to assist in choosing the ideal number of replicates; The results will not always be satisfactory because of variation between experiments

In order to use this spreadsheet: We must carry out a literature review regarding the purpose of the work being proposed; Identify the following points of each parameter that will be measured: - Mean and Standard deviation of the Treatment Control (TC) - Difference expected between the means of the CT and the Group to be studied.

For Example the Mean and Standard Deviation for Broiler Growth Experiment Average weight at 42 days of age: 2950 grams Standard deviation: ± 115 Expected mean difference: 5%

IT IS UNDERSTOOD - Reference or Control treatment as: - Standard It is important to emphasize that the conditions in the experiments predicted from are similar to the conditions in the facilities to be used; - After simulating some experiments, we will perform some tests.

Home page Introduction the E S P Instructions to User Simulations: 6 replicates 10 replicates 20 replicates

Instructions for the User: For example the data

Body Weight Simulation You can simulate using 6, 10 or 20 replicates

Only enter the values in the cells in black Body Weight Simulation

Maximum desired difference between means Simulated values appear Data to Be Entered Average (grams) Standard deviation Simulated values appear Tests are performed

Output Data CV of the Average and SD Mean Difference Mean, SD and CV 5% different from Control 20 Simulated data Treatment (Control) 20 Simulated data Treatment (Test) Actual Mean Difference about 5%

Output Data Each row is an experiment simulation (20) Randomly generated data for each replicate (six) Mean and SEM of each simulated treatment

Output Data Difference (P) observed between the treatments by the t-test, with 6 replicates Probability <0.05 and <0.01 Total Number of significant simulations Percentage of finding difference between two treatments according to the input data

With 6 replicates With 10 replicates We would have 35% different at P<0.05 and 5% at P<0.01 With 20 simulated experiments we could find: With 10 replicates With P<0.05 ~ 65% different With P<0.01 about 40% different

With 20 replicates In this example, we can find almost 100% difference between two treatments with 20 replicates

Important points ❶ With ESP students can see how changing Means Standard Deviation Number of Replicates Influence Experimental outcomes and Statistical Inference