Thursday HW

Tonight, please complete the "Experiments Wrap Up: Blocking" homework provided in class (or linked below). This homework summarizes pretty much everything we need to know about blocking, so it's an extra-important one! 

• Experiments Wrap Up: Blocking (Click Me for HW!)
• For question 2 your task is to identify the item numbers that would make up each block; you don't need to write any sentences here, simply list the item numbers that would be used for each block.
• For example, in a, researchers believe the effectiveness of the detergent might be different based on the color of clothes--so we would first want to group our clothing items by color (we can ignore any info about type and water temp).
• Which items (by #) create each block? (you should have 2 blocks)
• For the back side (2007 free response)
• For a, your job is again to create the blocks, identified by window number
• You want to block by some variable that we would expect to be related to what's being measured--"heat gain"
• For b, you have to write a procedure--use your notes from Friday/Tuesday (or the blog post below)...
• But, we don't want a completely randomized procedure, we want a blocked design, so we'll have to randomly assign treatments within our blocks!
• Always Check Your Answers!
• 1.) Factor: type of detergent
• 1.) Response Variable = cleanliness rating on a scale of 0 (very dirty) to 10 (very clean)
• 1.) Experimental Units = the clothes we are washing
• 2a.) Blocks:
• Dark: 1, 2, 3, 4, 9, 10, 15, 16
• Light: 5, 6, 7, 8, 11, 12, 13, 14
• 2b.) Need 3 blocks
• 2b.) Warm Block: 1, 2, 5, 10
• 2b.) Cold Block: 4, 7, 8, 9, 11, 12, 15, 16
• 2b.) Hot Block: 3, 6, 13, 14
• 2c.) Need 4 blocks...
• Pants/Light: 6, 7, 13, 14
• Pants/Dark: 2, 10, 15, 16
• Shirt/Light: 5, 8, 11, 12
• Shirt/Dark: 1, 3, 4, 9
• Hot/Light: 6, 13, 14
• Hot/Dark: 3
• 2007 Free Response Part A:
• *Note: be sure to state the variable you will block by and explain why you chose that variable--to explain, suggest how your blocking variable might be related to the response variable (what we're measuring).
• We could block by the "side of the house" (North/South/East/West) because windows on different sides of the house are exposed to different amounts of sunlight; since they are exposed to different amounts of sunlight windows on different sides of the house may vary in terms of their amount of heat gain.
• In this case we use 4 blocks:
• North = 1, 12
• West: 11, 10, 9, 8
• South: 6, 7
• East: 2, 3, 4, 5
• OR...
• We could block by whether or not the window is next to a door or not. Windows that are near a door may see less heat gain since the door is another opening (for potentially cold air) in the house. 
• In this case we use two blocks.
• Windows by door: 1, 12, 6, 7
• Windows not by door: 2, 3, 4, 5, 8, 9, 10, 11
• 2007 Free Response Part B:
• If blocked by "side of the house"
• 1.) Write the # of each window in the West block on a piece of paper.
• 2.) Put these papers in a hat and mix them up to randomize.
• 3.) Choose 2 papers without replacement; the windows with these two numbers will have window type A installed.
• 4.) The other two windows will have window type B installed.
• 5.) Repeat this process for the East block.
• 6.) For the North and South blocks you repeat steps 1 and 2. Then, choose one paper without replacement; this window will have window type A installed and the other window will have B installed.
• If blocked by "window is next to a door or not"
• 1.) Write the number of each of the 8 windows that are not by a door on pieces of paper.
• 2.) Put these papers in a hat and mix them up (to randomize).
• 3.) Choose 4 papers without replacement; the windows with these 4 numbers will have window Type A installed.
• 4.) The remaining 4 windows will have window Type B installed.
• 5.) Now, write the numbers of the 4 windows that are next to doors on pieces of paper.
• 6.) Repeat step 2.
• 7.) Choose 2 papers without replacment; the windows with these 2 numbers will have type A installed.
• 8.) The remaining 2 windows will have type B installed.

Experimental Design Free Response: Common Questions

• Identify factor (explanatory variable), treatments, response variable, experimental units/subjects
• Explain a process to randomly assign treatments (random # generator or #'s in a hat)
• Determine if a given context represents is an experiment or observational study.
• Can we determine cause and effect from a given study? (determine if it's an experiment or not; if it's an experiment we can establish cause + effect; if it's not an experiment, we can't)
• Identify a potential confounding variable/explain why a variable might be a confounding variable
• Choose a variable to block by and explain why
• Create blocks based on this criteria
• Explain how to randomly assign treatments in a blocked design
• Explain if there is/is not a control group, how you know, and explain why it is/would be useful to have a control group
• Determine if a study is single/double blinded OR explain how a study could be single/double blinded