Lecture 26: Latin Square Design Latin Square Design Example: Greenhouse We have seen how to use blocking to account for variability from a nuisance factor. What if we have two blocking variables? Less light Heater Warmer Cooler Doors More light 1 Windows Nuisance Factors Latin Square Design There is variability in warmth from side to side. There is variability in light from top to bottom. Need a design that can account for both of these nuisance factors. One set of treatments (these could be levels of a single factor or combinations made by crossing several factors of interest). 3 4 Latin Square Design Latin Square Design Two blocking (nuisance) factors. The number of levels for each blocking factor is determined by the number of treatments. Every combination of the two blocking factors occurs once. Treatments are crossed with each blocking factor. 5 2 6 1 Lecture 26: Latin Square Design Latin Square Design 3 Treatments A Latin Square plan (or layout) starts with a systematic arrangement of the treatments in rows (levels of one blocking variable) and columns (levels of the second blocking variable). A B C B C A C A B 7 4 Treatments A B C D B C D A C D A B 8 Comment Note that in a Latin Square plan that each treatment appears once in each row and once in each column. D A B C 9 Comment 10 Example: Movie Appeal Response: How many people, out of 50, would recommend the movie to a friend. Conditions: Four types of movies; Action, Sci Fi, Comedy or Drama. Experimental Units: Audiences of 50 people. A Latin Square plan is a systematic arrangement of treatments. How does one include randomization? 11 12 2 Lecture 26: Latin Square Design Blocking (Nuisance) Factors First Blocking Factor The people that attend movies may be different depending on the day of the week or the time of day. Day of the week. Because there are four movies (levels of the factor of interest) four days of the week; M, T, W, Th make up one blocking factor. 13 Second Blocking Factor 14 Randomization Select a plan at random from all possible plans (not practical). Randomize order of rows, columns and treatments. Time of day. Use four times of day; Early Matinee, Late Matinee, Early Evening, Late Evening. 15 Start with systematic plan 1 2 3 4 1 A B C D 2 B C D A 3 C D A B 4 D A B C 16 Randomize Columns Use JMP to randomly shuffle the numbers from 1 to 4. Random order: 2, 3, 4, 1 17 18 3 Lecture 26: Latin Square Design Randomize Columns 1 2 3 4 Monday 2 Tuesday 3 B C D A C D A B Randomize Rows Wednesday Thursday 4 1 D A B C Use JMP to randomly shuffle the numbers from 1 to 4. Random order: 4, 2, 1, 3 A B C D 19 Randomize Rows EM 4 LM 2 EE 1 LE 3 Monday Tuesday C A B D D B C A Randomly Assign Treatments Wednesday Thursday A C D B 20 B D A C Use JMP to randomly shuffle the numbers from 1 to 4. Random order: 1, 3, 2, 4 21 Comment Randomly Assign Treatments Monday Tuesday EM C Trmt 2 D Trmt 4 A Trmt 1 B Trmt 3 LM A Trmt 1 B Trmt 3 C Trmt 2 D Trmt 4 EE B Trmt 3 C Trmt 2 D Trmt 4 A Trmt 1 LE D Trmt 4 A Trmt 1 B Trmt 3 C Trmt 2 22 Wednesday Thursday 23 After randomizing rows, columns and treatment assignments the randomized plan still has the characteristic that each treatment appears once in each row and once in each column. 24 4