# Entry Test for Statistical/Data Science class

If there is a Data Science class (for final year high school students, not necessarily from the same school) with the following syllabus:

• Python programming (first 3 meetings)
• Data cleaning and Twitter Data visualization using Matplotlib (4 meetings)
• Probability & Statistics Theory, Normal Distribution, and the Central Limit Theorem (2 meetings)
• Bootstrap Algorithm (2 meetings)
• Some Advanced methods: Clustering, classification and regression trees (CART) (4 meetings)

My question is, how should the entry test be? (the test questions)

or better not give test? since we will teach from basic

I have initial idea that is to give test that includes basic calculus such as application of derivatives (like graphing a function etc) to test their mental ability to visualize abstract math concept. I wonder better approaches. The test should not be restrictive.

• Is the purpose of the test to filter out students who will not benefit from the class? For example, are likely to not do well, no matter how much work they put in, and no matter how much help they get? Commented Jul 15, 2019 at 15:38
• @Jasper I think so. But I am quite disagree with that though. People attend school because they don't know and not "smart". Commented Jul 15, 2019 at 22:02

## 1 Answer

Consider the following papers:

These feature a multiple-choice diagnostic test used at the start of a statistics course, with various data and conclusions drawn from the results. The test is given in its entirety in each paper. The Johnson paper uses a 15-question test, and the Lunsford paper builds on it with an additional 5 questions. I've used this as the basis for a first-day diagnostic in my community-college elementary statistics course for several years. I find that it has a small amount of predictive power with regard to final exam scores (R^2 = 0.13, N = 238).