---

Below is an overview of what updates have been made from one edition to the next. Full details are logged in our GitHub repository for changes made from the 2nd to the current edition.

---

Availability of Past Editions

All editions are included as extras in the Leanpub purchase. Reminder: you may set the price to Free; contributions are 100% optional. We believe that students should have "forever access" to the book they used in their college course.

---

3rd to 4th Edition Updates

The 4th Edition was released on May 1st, 2019. The textbook price was updated from $14.99 for the 3rd Edition to $20 for the 4th Edition, which we believe will be a sustainable price point that helps support OpenIntro as it scales into new subjects.

Complete visual redesign. Complete style redesign of chapter, section, example, guided practice, and term box layouts. Sections now always start at the top of a page, making them easier to find. Page size is now 8.5" x 11" (up from 8" x 10"), allowing us to reduce the page count.

Structural and pedagogy. The content from Chapter 1 in the Third Edition is now Chapters 1 and 2. We eliminated normal probability (quantile-quantile) plots and replaced them by histograms. See the related change around the skew condition change below. Negative binomial and Poisson distributions are now in their own sections. Proportions are now used to introduce inference concepts, and the Inference for Categorical Data chapter comes before the Inference for Numerical Data chapter. We now only use two-sided hypothesis tests throughout the textbook. Small sample proportion inference in the Inference for Categorical Data chapter has been moved into an online extra. The skew condition check for inference for means has been simplified and been replaced with rules of thumb check for outliers. There are now two significant multiple regression case studies in the main text.

Data updates. Many older data sets have been replaced with new data sets. We've consolidated data sets information from the main text into a Data Appendix (Appendix B).

Exercise overhaul. Exercises are now at the end of sections, with a handful of additional review exercises at the end of each chapter. Dozens of new exercises, as well as removing many older exercises.

Price. The PDF is still free. B&W paperback retail price is $20, which is up from $14.99 for the 3rd Edition. You may see an increase to the price on Amazon relative to other retailers and bookstores, e.g. $25-30, if we see Amazon applying fee structures that handicap our ability to later sell full-color prints at reasonable prices when we try a new printer.

If you have questions related to the Fourth Edition, see this thread in the Public Forums and ask away!

The 3rd Edition was made available thru mid-December 2019 to allow a half year for instructors to switch to the new edition.

---

2nd to 3rd Edition Updates

The 3rd Edition of OpenIntro Statistics was released in July 2015. Pricing was changed from $9.94 for the 2nd Edition (purely at cost) to $14.99 for the 3rd Edition (provided a margin to cover project costs). Below are the significant changes. Full details are logged in our GitHub repository.

The t distribution is used exclusively in the Inference for Numerical Data chapter (Ch 5).

Some data sets are replaced with newer or more interesting data sets.

Some exercises are revamped, and some entirely new exercises have been added. The number of exercises has increased by a small amount.

Video icons were added to sections where videos are available, and these are hyperlinked in the PDF.

Section 8.2, which focuses on variable selection in multiple regresion, now highlights the usage of adjusted R-squared instead of p-values for variable selection. Both approaches are still featured, but adjusted R-squared is featured more prominently.

Section 3.2.2 in the Second Edition, which covers calculations for the normal probability plot, was eliminated. Additionally, the R function qqline() is now used throughout the textbook for the normal probability plot, whereas a custom function was previously used.

The within-text exercises (the ones with solutions in the footnotes) are now called Guided Practice to eliminate duplicate "exercise" numbering with end-of-chapter exercises.

We dropped the df > 1 condition for the chi-square test. The reasoning on why it was introduced is subtle: a 1-prop test can be turned into a chi-square test with df = 1, but then we only need 5 successes / failures for the 2-prop test, so the conditions don't match. Anyways, we removed this condition for simplicity.

The section on calculating power (Section 4.6) was removed and a new section was added in Chapter 5 that covers power in the context of an experiment. This new section should be much easier to follow, and the example is also a much more common power calculation context.

A reference to Why 0.05? was added.

There were other minor changes, e.g. fixing typos identified in the Second Edition.

The 2nd Edition was made available for 10 months after the release of the 3rd Edition.

---

1st to 2nd Edition Updates

The 2nd Edition of OpenIntro Statistics was released in August 2012. Pricing was changed from $9.02 for the 1st Edition to $9.94 for the 3rd Edition (both editions were priced at-cost). Below are the significant changes of the update from the 1st to the 2nd Edition. The blog post announcing the updates may be found here.

New data. Many of the data sets, some just one or two years old, have been swapped out for newer data and studies. We've worked hard to ensure that OpenIntro Statistics remains fresh and current.

Updated Chapter 1. Data collection is now featured ahead of the summaries and graphics sections. We include a new research study with surprising results to lead off the textbook and engage students. Two new data sets featuring email and census data take the place of the possum and cars data sets that are present in the First Edition. An important new subsection has also been added that includes intensity maps to highlight the structure of spatial data.

Chapter 5 and 6 updates. These chapters previously were structured around large and small samples. Chapter 5 will now introduce inference methods for numerical data, and Chapter 6 will feature methods for categorical data (proportions, contingency tables). The section on ANOVA will be moving from Chapter 8 to Chapter 5 and will remain a "special topic" section.

Sample size conditions have evolved slightly. We now set 30 as the standard for using the normal approximation for numerical data and 5 as the minimum expected cell count size for chi-square. With these changes, we also add more strict and explicit conditions regarding skew and table size to ensure the methods remain rigorous and appropriate.

New logistic regression section. Chapter 8 includes a brand new section featuring logistic regression in the context of developing a filter for email spam. Our final spam filter isn't on par with Gmail's spam filter, but we make a surprising amount of headway in just 9 pages. This new section also highlights ideas and methodology students could see in a second applied statistics course, which makes this new logistic regression section a nice closing act for the content of OpenIntro Statistics and one that is within reach of an advanced or honors introductory statistics course.

The 1sts Edition was made available for one academic year after the release of the 2nd Edition (discontinued in mid-2013).