This module uses the Online Reaction time testing program written by University of Washington Faculty Eric H. Chudler
techred home > experimental design

Statistical Experiments

wb_incandescent

Data analytics as a field contains many data collection and anayltic methods, many of which center around the need to determine causal relationships between variables. Many studies, both academic and "practical" use experimental design to tease out what actions or behaviors lead to a given outcome.

This module invites you to design an experiment around human reaction times and use basic statistical inference to make a claim to the treatment's impact on reactiontime. We'll brush up on our spreadsheet skills and review the fundamentals of sampling distributions and the Student's T-Test.

wb_incandescent

  1. The official documentation for Libre Office Calc (400+ page PDF document)
  2. Intro to statistics with Libre Calc
  3. Online Reaction time testing program written by University of Washington Faculty Eric H. Chudler
  4. Experimental design template spreasheet
  5. Tinkering with normal distributions and random numbers

directions_walkWorker incentive design

You are a management team at a national nonprofit called UrbanBranch which coordinates with local organizations to plan regionally appropriate trees in urban spaces throughout the united states. You coordinate two teams of Tree Coordinators: the team Blue coordinators and the Team Tan coordinators.

Goal: Develop tree planting targets for each team which honor the recommended a scheme of allocating bonus dollars collectively to 60% of the highest performing members of a given team.

A team member is represented by a pen with rubber band of a particular color. The number of trees that each employee on each coordinates getting planted for each month is represented by the number of millimeters that the pen travels from its launch location.

Products to produce as a team

  1. A number of trees that represents the threshold on which to base the team's bonus. Ideally, you want a threshold that will, on average, award bonuses to the top-performing 60% of the team. Since each time has a different distribution, you'll need a cutoff for team Blue and team Tan
  2. A summary of how you arrived at your particular threshold levels, including any screen shots of your use of the Lock 5 stats tool

wb_incandescentStation 2: Cell phones and reaction time

  1. Check out the Online Reaction time testing program written by University of Washington Faculty Eric H. Chudler. See how it works, and what kind of data it puts out.
  2. Design an experimental condition that you believe will impact the reaction time of a subject participating in your study. You could design treatments that you believe will lengthen reaction time, such as by asking the subject to send a text message while waiting for the light to change, or by displaying distracting images alongside the test screen.
  3. Populate the sample spreadsheet template linked in the resources with your experimental data
  4. Conduct your experiment with as many class members as you can
  5. Use the average() and stdev() functions to calculate summary statistics on each experimental group
  6. Use the ttest() function with the listed function settings to calculate the p-value of the analytic procedure
  7. Write a claim about your conclusion with respect to reaciton times
  8. Post your results spreadsheet to your cloud drive folder

wb_incandescentStation 3: Product quality

You work for a micro-chip manufacturer and have received two batches of chips from your foundry (the place that converts a digital chip design into an actual chip). Your task is to choose which of the two batches is a higher quality.