Homework #2

(due: section 501 - Wednesday, 8/12; section 502 - Thursday 8/13)

Attempt all questions. Question #1 and #2 are each worth 10 points and #3 is worth 20 points.
  1. A usability engineer would like to develop a model to predict the usability of a GUI from the organization of GUI objects (i.e. menus, list boxes, radio buttons etc.). The engineer uses a widely used index as a measure of organization of the GUI objects, and time required to complete a specific set of tasks as the measure of usability.

    The engineer collects data for several GUI designs and computes the following statistics:

    If the correlation coefficient is 0.94 and the simple linear regression model is appropriate.

    1. State the the regression model.
    2. Estimate the coefficients of the regression line.
    3. Interpret your estimates.

  2. A DePaul administrator believes that DePaul graduates are offered better starting salaries than UIC graduates. She compiles a list of seniors with job offers from both schools and randomly selects ten from each list to be interviewed.

    Assume that these data represent the starting salaries in thousands of dollars for the twenty graduates interviewed.

    1. Would you analyze these data using a paired or two independent sample approach. Justify your answer.
    2. The administrator is familiar with regression and would like you to analyze these data using regression. Write the SAS code to do this. Provide the data step and the procedures in the procedure step to conduct a thorough analysis.
      Note:You do not need to analyze the data but you should make sure that the code works and is complete.

  3. You have been asked to analyze a SAS report which was produced to test the null hypothesis that a particular testing methodology is equal in effectiveness to some other testing methodology (say white box vs. black box testing) against the two sided alternative that they are not equal.

    Conduct a thorough analysis using the methods outlined in Knafl, chapter 4.
    Note: Your analysis should include a post hoc analysis.