COOL* Statistics 200 Fall 2018 Sections
Section L1 - Karle Flanagan
150 Animal Science Lab
Online Section - Ellen Fireman
Watch lectures on Compass.
Welcome to attend in-person lecture anytime.
Course Goals and Philosophy
Why everyone needs to know Basic Statistics:Statistics is a tool to make sense of large amounts of information. Common sense can only handle limited amounts of information. Until recently common sense was sufficient for most people because daily life didn't involve processing a large amount of data. Now large stores of information have become readily available. You can either choose to ignore the information available or you can choose to make sense of it, which means learning statistics.
What's special about Stat 200:Most people think statistics is boring and difficult. Statistics is to data, what grammar is to words. And like grammar, it's only interesting if it's used to understand something interesting. In Stat 200, we use statistics to research a topic we're all interested in - ourselves. We collect data on ourselves through anonymous surveys, largely on the sort of social questions on which students have shown intense interest. Having real questions that we want to answer motivates real understanding, not just memorizing some complicated rules. Statistics is a collection of real tools- the key is to understand which one to use when and why.
*This year we have a special project, doing research on a new class structure we developed, our Coordinated Online Learning (COOL) system seamlessly integrating online and in-person versions. We're studying which works better for which types of students.
3 Main Goals
- Using a conceptual, intuitive approach to understand a set of complex statistical methods.
- Determining whether predictors are also causes.
- Learning to use statistical software both to help us understand what the statistical methods are doing and to do the calculations for us.
We tie the methods together with a few unifying themes rather than get bogged down in the tedious calculations and tiny differences between some of them. We build a unifying framework for general predictive models.
Often we really care about whether X causes Y, not just whether X predicts Y. Would changing X change Y? To answer that question in the absence of a randomized experiment involves sorting out a tangle of possible confounders and causal links. As we examine real data sets, we investigate how to evaluate causal claims.
a. We have a simple, point and click data program to let you compare the same data sets with a variety of methods to see how the results they give differ and why. We can get a feel for which of the methods, which are usually all approximations to the real world, work better in which situations by analyzing the data using a variety of them. We focus on understanding the underlying concepts behind the methods so you can choose your methods wisely.
b. Once you do understand the needed statistics, performing useful applied calculations usually requires more flexible software than our data program. We have bonus exercises on using the R statistical language to acquire basic familiarity. For those who wish to become fluent in R programming, we now offer a free non-credit R course .