Experiment Analysis
The purpose of this activity is to practice analyzing experiment results by applying the t test. The data you will analyze will be simulated data, randomly generated by a web service.
For this activity, you will need to use:
- our random data generator;
- a spreadsheet that can accept tab-separated-value data (such as Excel, OpenOffice, or Google Spreadsheet);
- GraphPad's t test calculator
The data generator requires two inputs: an experiment and an experimental design. For this activity, we will use the Point-and-click experiment, which simulates the time it takes for a user to point and click a particular target on screen using up to three different pointing devices (a mouse, a trackpad, and a trackball). The other input, the experiment design, is a string specifying the conditions and trials made by each user in the experiment.
The output of the data generator is tab-separated-value data, which you can copy and paste into a spreadsheet to rearrange and compute statistics, and then transfer to the t test calculator to perform the statistical test.
Between-Subjects
Generate data from the Point-and-Click experiment using the experiment design MMMM,MMMM,MMMM,PPPP,PPPP,PPPP. (How many users and trials is this?)
Move the data to your spreadsheet. It's a good idea to split the time column into two side-by-side columns, one for the mouse condition (M) and one for the trackpad condition (P).
Use the spreadsheet to compute the mean (called AVERAGE in most spreadsheets), standard deviation (STDEV), and standard error (STDEV/sqrt(n)) of each condition. If you graphed the two means with error bars, would the error bars overlap?
Run a t test on this data at the 5% significance level. Determine the p value, the value of the t statistic, and the degrees of freedom.
Within-Subjects
Generate data from the Point-and-Click experiment using the experiment design MMPP,MMPP,MMPP,MMPP. (What's the risk of this experiment design?)
Move the data to your spreadsheet. Line up each user's M and P trials side-by-side.
Run a paired t test on this data at the 5% significance level. Determine the p value, the value of the t statistic, and the degrees of freedom.