250 likes | 384 Views
Statistics 400 - Lecture 18. Last Day: ANOVA Example, Paired Comparisons Today: Re-visit boy’s shoes...Randomized Block Design. Example (Boys Shoes). Company ran an experiment to determine if a new synthetic material is better than the existing one used for making the soles of boys' shoes
E N D
Last Day: ANOVA Example, Paired Comparisons • Today: Re-visit boy’s shoes...Randomized Block Design
Example (Boys Shoes) • Company ran an experiment to determine if a new synthetic material is better than the existing one used for making the soles of boys' shoes • Experiment was run to see if the new, cheaper sole wears at the same rate at which the soles wear out
Example (Boys Shoes) • 10 boys were selected at random • Each boy was given a pair of shoes • Each pair had 1 shoe with the old sole (Sole A) and 1 shoe with the new sole (sole B) • For each pair of shoes, the sole type was randomly assigned to the right or left foot
Can we use a 2-sample t-test or ANOVA here? • Would the 2-sample t-test or ANOVA detect a significant difference?
Paired or Matched Pairs T-test • Situation: • Two measurements made on same experimental unit • Compute difference (say B-A) in observation on the same experimental unit • Analyze differences using a 1-sample t-test • Because we analyze the differences using a 1-sample t-test, what must we assume about the difference?
Just looked at comparing means for two treatments applied to the same experimental unit (see boys’ shoes example) • Used a matched pairs T-test and analyzed the differences to see if there was a significant difference in the treatment means • When more than 2 treatments are applied to the same experimental unit, the experiment is called a randomized block experiment
Example • An experiment was performed to investigate the impact of soil salinity on the growth of salt marsh plants (C. Schwarz, 2001) • Plots of land at 4 agriculture field stations were used to grow plants in this environment • Six different amounts of salt (in ppm) are to be investigated • The plots of land were divided into 6 smaller plots • Each of the 6 smaller plots were treated with a different amount of salt and the bio-mass at the end of several months recorded
The application of the 6 treatments to the smaller plots are done randomly • Like the Boys’ Shoes Example, each experimental unit has received more than 1 treatment • Here each unit receives 6 treatments
Observations • Notice that the 4th plot gives smaller results than the other plots • Due to a block (plot effect) • Similar to the way a boy wears his shoes • Are the observations independent?
Randomized Block Design • Situation: • Have k treatments • Have b blocks • Each of the k treatments appears in each of the b blocks • The treatments within block are assigned to the within block units in random order
Structure of Data • Have k treatments in b blocks • Denote ith treatment from the jth block as yij
Model: • Model for comparing k treatments from a randomized block design: • for i =1, 2, …, k and j =1, 2, …, b • where is the overall mean, and • is the i th treatment effect • is the jth block effect • eij has a distribution • Want to test: