Lab 5: Lingering Effects of Sleep Deprivation?

Background: Researchers have established that sleep deprivation has a harmful effect on visual learning. But it is not clear how long these negative effects last. In a 2000 study Stickgold, James, and Hobson investigated whether subjects could “make up” for sleep deprivation by getting a full night’s sleep in subsequent nights. This study involved training and then pre-testing 21 subjects (volunteers between the ages of 18 and 25) on a visual discrimination task on Day 1. (A stimulus appeared on a computer screen and then the time, in milli-seconds, was measured until subjects could accurately report what they had seen.) Subjects were then randomly assigned to one of two groups: one group of 11 subjects was deprived of sleep on the night following training (night 1), and the other group of 10 subjects was permitted unrestricted sleep on that first night. Both groups were then allowed as much sleep as they wanted on the following two nights (nights 2 and 3), and were re-tested on Day 4. The researchers expected participants in both groups to improve on the visual discrimination task the second time around (i.e., have faster reaction times), but they hypothesized that the unrestricted sleep group would have more improvement than the restricted sleep group, assuming trying to "catch up" on sleep does not work for visual learning.  The research question is whether there tends to be larger improvement in performance on the task for people who are not sleep deprived than for people who are sleep deprived.

Goals: In this lab you will analyze whether sleep deprivation hinders subjects' abilty to improve on a visual discrimination task (basically reaction time) three days later. You will produce graphical and numerical summaries to effectively compare the two groups on this quantitative response variable. You will then carry out a simulation of the random assignment process assuming there is no detrimental effect from sleep deprivation to help you decide whether the observed difference in sample means between the two groups is statistically significant. You will also consider an alternative choice of statistic and explore its null distribution.


  • These webpages are best viewed "maximized" (click the Maximize tab in the upper right). You may need to use scroll bars to see more of the applet and/or instructions.

When you are ready, click Start to begin.

Start Lab