This Java applet allows introductory statistics students to gain practice in identifying what kind of methodology to apply to a given situation. The format is web-based and there are 13 possible answers for users to choose from. These choices range from using a confidence interval or a hypothesis test for one proportion, one mean, two proportions, two independent means, and paired means. It also has options for a prediction interval from regression and two chi-square optionis for goodness of fit or independence. Incorrect choices prompt some feedback to guide the student towards the appropriate choice.
Type of Material:
This material can be used in a variety of ways. An instructor can use the materials in class with a video projector/laptop and internet hookup. A question shown on a screen can be discused by students in small groups. Another option is to have the students themselves access the questions when they are in a computer classroom or lab. Students can also use these materials for individual practice while away from campus as long as they have an internet account.
Users must have internet access and be able to install Java tools on their machine. Some users may also need to allow popups.
Identify Major Learning Goals:
This activity helps students learn to identify the appropriate inference procedure to apply to a given problem or situation. This helps students distinguish between one and two sample situations, independent verses dependent situations, among others.
Target Student Population:
The target population seems to be students in a very comprehensive one or two semester introductory statistics class.
Although the questions are generic in nature - these materials might be helpful to any student in business,
psychology, sociology, education, or biostatistcs who might be in a statistics course for the first time. This may include graduate level students in departments other than statistics.
Prerequisite Knowledge or Skills:
The skills needed is content knowledge of estimation and hypothesis testing for one and two samples, for both proportions and means. In the two sample case, this means both independent and dependent samples. Content knowledge of regression with prediction intervals (which is not always a standard topic in a first semester general statistics course) is also required. Lastly, knowledge of chi-square tests for goodness of fit and independence and, thus obviously, the difference between the two, which is also not necesarily emphasized in an introductory statitsics course.
Evaluation and Observation
This applet provides a good source of drill and practice for students in determining the appropriate statistical methods. This is often one of the most difficult things for students to learn. Students often master the particular procedures, but evaluating the sitaution and deciding which procedure is appropriate is much more difficult. One of the ways to help students learn to do this is simply practice. This applet is designed to give students the opportunity to practice, and it provides that experience.
The situations to be examined are short and very easy to read. The variety of methods to be used gives students practice identifying different statistical techniques.
The only inference procedure for quantitative vs. quantiative relationships is "prediction interval for value of y given value of x" and for those cases the distinction between an confidence interval for the mean "y" vs. a prediction interval for an individual "y" is ignored - a substantial flaw.
"Guidance" given for incorrect choices feels too mechanical at times. For example, an error for incorectly choosing for/against a chi-square test just gives a message that this is not/is a chi-square test. Similarily, the message for the regression interval just states whether there is or is no "value of x given here". When the "correct" choice is a test, the prompt given when a student chooses a CI is "You want to determine if something is true. Try again with a hypothesis test". Not a good message to send that hypothesis testing is finding out what is "true".
Sometimes, the situations described are too terse. For example, one question says,
"500 people were asked what is their favorite movie and who is their favorite author. Based on the data can you conclude that there is a relationship." The answer is chi-square for independence. I don't like this question as it is not clear at all how data would be collected so that someone could do a chi-square test. Are the movies/authors divided into categories? In my opinion, this question would only confuse students more, instead of helping them learn.
Because of the vagueness of the wording on many questions, I would be hesitant to have students try these on their own at home. It might end up confusing them more and not helping them learn. If I can be there doing this stuff with the class, I can give hints or can clarify how the data would be collected to help them get to the right answer.
Potential Effectiveness as a Teaching Tool
This tool gives students access to practice something that is extremely difficult to learn. In addtion, it gives instructors a resource as these kind of questions take a long time to develop.
This item is probably only effective near the end of a course when students have some familiarity with all of the procedures. It would be less effective if one or more of the inference procedures (e.g. prediction intervals in regression or chi-square goodness of fit) are not covered. The effectiveness of this tool would be greatly enhanced by adding the ability to limit the applet to only a few topics.
Ease of Use for Both Students and Faculty
Very straightforward to use.
No record is kept of number of correct/incorrect choices, so it may be difficult for students to track their achievement level.
Other Issues and Comments:
The stats community needs more opportunities like this for students to practice. The author has done a great job in getting this process started. The challenge is that there is so much variety in what people teach in the intro stats course that customization
is almost impossible to go without. Right now this applet has limited appeal because not many instructors get through all those topics in one course and without a way to limit questions to just the topics that have been covered will preclude most instructors from using this applet. I would love to use this in my own class, but right now I don't get to several of the methods among the 13 choices in the first course. With more and more education research showing that students forget most of what we teach in 6 months anyway, I have been concentrating on emphasizing less topics, but trying to go to a deeper and more lasting understanding.
I do think for those instructors that do get through all the methods, this would be a very helpful tool.