I’m thrilled to release my new book! Hypothesis Testing: An Intuitive Guide for Making Data Driven Decisions.
In today’s data-driven world, we hear about making decisions based on the data all the time. Hypothesis testing plays a crucial role in that process, whether you’re in academia, making business decisions, or in quality improvement. Without hypothesis tests, you risk drawing the wrong conclusions and making bad decisions. That can be costly, either in business dollars or for your reputation as an analyst or scientist.
Painlessly learn how to use these tests in this 367-page book! If you like the clear writing style I use on my website, you’ll love this book! Throughout this book, I use the same clear, concise language. I focus on helping you grasp key concepts, methodologies, and procedures while deemphasizing equations.
Hypothesis tests allow you to use sample data to draw conclusions about an entire population—not just the small sample with which you’re working. Consequently, these tests play a vital role in making discoveries in science, making decisions based on data, and making predictions. Additionally, given the growing importance of decisions and opinions based on data, your ability to critically assess the quality of analyses that others present to you is more crucial than ever.
By reading this book, you will build a solid foundation for understanding hypothesis tests and become confident that you know when to use each type of test, how to use them properly to obtain reliable results, and how to interpret the results correctly. I present a wide variety of tests that assess characteristics of different data types. Chances are high that you’ll need a working knowledge of hypothesis testing to produce new findings yourself and to understand the work of others. The world today produces more analyses designed to influence you than ever before. Are you ready for it?
To accomplish these goals, I teach you how these tests work using an intuitive approach, which helps you fully understand the results. At the end of this post, you’ll find the full table of contents.
Buy it on Amazon (US site)!! Or go to my Web Store for other locations.
My Book Covers the Following Critical Hypothesis Testing Concepts, Methods, and Skills
This book enables you to build the skills and knowledge necessary for effective hypothesis testing, including the following:
- Why you need hypothesis tests and how they work.
- Using significance levels, p-values, confidence intervals.
- Select the correct type of hypothesis test to answer your question.
- Learn how to test means, medians, variances, proportions, distributions, counts, correlations for continuous and categorical data, and outliers.
- One-Way ANOVA, Two-Way ANOVA and interaction effects.
- Interpreting the results.
- Checking assumptions and obtaining reliable results.
- Manage the error rates for false positives and false negatives.
- Understand sampling distributions, central limit theorem, and statistical power.
- Know how t-tests, F-tests, chi-squared, and post hoc tests work.
- Learn about the differences between parametric, nonparametric, and bootstrapping methods.
- Examples of different types of hypothesis tests.
- Downloadable datasets so you can try it yourself.
For each hypothesis test I cover, you will learn what it tells you, understand its assumptions, know how to interpret the results, and work through examples with downloadable datasets.
Please consider buying my book and learn about hypothesis testing! I’m sure you’ll enjoy it and find it helpful! The full table of contents is below.
Buy it on Amazon (US site)!! Or go to my Web Store for other locations.
Thank you for the quick response! The thing that confuses me is the requirement for independent groups. For example, one or more people might be coincidentally selected in both polls.
You bet, Travis!
Accidentally including the same people in both samples is a potential concern that depends on the size of your sample relative to the size of the population you’re sampling.
We do have to assume that you’re drawing two random samples from the same population to use the test. If a few people happen to be in both samples by chance, that’s technically an assumption violation but it won’t affect the results much. However, if there are many in both samples, it is a problem.
If you’re drawing two random samples of 100 out of a town of 200, you’d expect 25 people to both samples by chance. That’s too many. However, if you’re drawing two random samples of 100 out of 1000, you’d expect 1 person to be in both samples by chance. That should be OK. I don’t know of any guidelines surrounding that to provide but 1 person in a total of 200 in both samples can’t affect the results too much. But you really need to understand the likelihood of people being in both samples. Of course, if you know the identities of the respondents, you can remove those that are in both samples.
This is an amazing book! However, I do a have question that I haven’t been able to answer. Imagine that I conduct a poll of the favorite pet of people in my town, with the following results:
Dog – 25%
Cat – 25%
Horse – 10%
Other – 40%
Margin of error: +/- 3%
After the poll, two things happen. First, several people that own horses move to town. Second, a nonprofit organization launches a campaign to adopt cats. So, I conduct another poll with these results:
Dog – 24%
Cat – 30%
Horse – 20%
Other – 26%
Margin of error: +/- 2.8%
Here’s my question: What would be the appropriate test to see if there is a statistically significant difference between “Cat” from the first poll to the second (i.e., 25% to 30%)? Please note: I’m not interested in the difference for “Dog,” “Horse,” or “Other” in this example – just “Cat.” Thank you in advance!
Hi Travis,
I’m glad you enjoyed the book and that it was helpful!
You could perform a 2-sample proportions test to compare the proportion of “cat” responses in the first and second poll. If the results are statistically significant, then you can conclude that the preference for cats has changed between the two polls.
I go over the 2-proportions test in my Hypothesis Testing book, so look at that for more information!
Thanks a lot Jim. That was helpful !!
Thanks a lot Jim. That was really helpful.
Being new to this kind of analysis, sometimes I face ambiguity with the statistical language. We draw two samples from the same population and then check for the difference in means and then we conclude that the difference is statistically significant at the population level. So indirectly aren’t we saying that the two samples are coming from two different populations? Seems a bit ambiguous to understand it intuitively.
Hi Animesh,
In your case, if you’re using independent samples, you’d be drawing two random samples from the same population. So, you’d have two samples but they’re representing one population but at different points in time (in your case).
Hi Jim,
For a stock market index (lets say S&P 500), if I calculate returns for a particular period (lets say each month for the period 2012 to 2014 i.e. a sample of 36 months ) and then I calculate returns for another period (lets say each month for the period 2015 to 2017 i.e. another sample of 36 months). Now , I need to understand the difference between the mean monthly returns of these two samples of 36 months each. I need to understand whether the difference is statistically significant. Each period is a sample drawn from the larger population of monthly returns of S&P 500. So, for Hypothesis testing can I assume that each sample is drawn from the population of monthly returns of S&P 500? In other words , each sample is coming from the same population . Or do , I need to assume that each sample is coming from a different population ? First sample from population of monthly returns for the period 2012 to 2014 and second sample from the population of monthly returns for the period 2015 to 2017. Could you please clarify this point ? Thanks.
Hi Animesh,
I haven’t done stock market research like that so I don’t know if there’s a standard they use for that type of research. However, I’d imagine you can assume it’s the same population for the two time periods. If you’re using independent samples (different companies), you’d use a 2-sample t-test. Alternatively, you could draw one random sample at the first time period and then reassess the same group of companies during the 2nd time period. In that case, you’d use a paired t-test to determine whether the change in values was significantly different from zero (or other value you set).
I hope that helps!
Hi Jim,
I have purchased you boos intro to stats which I find very useful in plain English. I plan to do so for the other tow but I have a question about Hypothesis Testing. Some big statistician like Dr. Wheeler asserts that Hypothesis Testing can;t be use in real production or manufacturing environment as the assumption of normal data does not exist, Do yuo agree with his work, if you are familiar with his.
Thanks in advance
Hi Mohamed,
Thanks so much for buying my Introduction to Statistics book. I’m glad you found it to be helpful! 🙂
I disagree with the idea that you can’t perform hypothesis testing on non-normal data. There are various hypothesis testing methods available for non-normal data. For one thing, with a sample size of only 20-30 per group, parametric tests become robust to non-normal data. Additionally, you can use transformations to make the data normal. Alternatively, you can use nonparametric tests and bootstrapping methods to perform hypothesis tests with non-normal data. There are a variety of methods to analyze non-normal data. I cover those methods in my Hypothesis Testing book.
I have not read Dr. Wheeler’s work so I can’t say if I agree or disagree in general. But, I do say that you can use hypothesis testing with non-normal data!
Do you plan to sell this book as a paperback as well?
Hi Laura,
Yes! It’s currently available as a paperback. Go to My Web Store to get the Amazon links. It’s also available at other online retailers. Some physical bookstores can order it as well!
Thanks for asking!
Thanks for writing this blog