What do you think?
Rate this book
Hypothesis Testing: An Intuitive Guide for Making Data Driven Decisions
The world produces more data than ever. Are you ready for it?
In today's data-driven world, you hear about making decisions based on data all the time. Hypothesis testing plays a crucial role in that process, whether you're in academia, business, or data science. Without hypothesis tests, you risk making bad decisions.
Chances are high you'll need to understand these tests to analyze your data and evaluate the work of others.
Build the knowledge for effective hypothesis testing! Know when to use each test, how to use them reliably, and how to interpret the results correctly!
Understand why you need hypothesis tests and how they work.Effectively use significance levels, p-values, confidence intervals.Select the correct type of test to answer your question.Learn how to test means, medians, variances, proportions, distributions, counts, correlations for continuous and categorical data, and find outliers.One-Way ANOVA, Two-Way ANOVA, and interaction effects.Check assumptions to obtain reliable results.Manage the error rates for false positives and false negatives.Understand sampling distributions, the central limit theorem, and statistical power.Know how t-tests, F-tests, chi-squared, and post hoc tests work.Learn about differences between parametric, nonparametric, and bootstrapping methods.Examples of many hypothesis tests.Access free downloadable datasets so you can try it yourself.
In today's data-driven world, you hear about making decisions based on data all the time. Hypothesis testing plays a crucial role in that process, whether you're in academia, business, or data science. Without hypothesis tests, you risk making bad decisions.
Chances are high you'll need to understand these tests to analyze your data and evaluate the work of others.
Build the knowledge for effective hypothesis testing! Know when to use each test, how to use them reliably, and how to interpret the results correctly!
Understand why you need hypothesis tests and how they work.Effectively use significance levels, p-values, confidence intervals.Select the correct type of test to answer your question.Learn how to test means, medians, variances, proportions, distributions, counts, correlations for continuous and categorical data, and find outliers.One-Way ANOVA, Two-Way ANOVA, and interaction effects.Check assumptions to obtain reliable results.Manage the error rates for false positives and false negatives.Understand sampling distributions, the central limit theorem, and statistical power.Know how t-tests, F-tests, chi-squared, and post hoc tests work.Learn about differences between parametric, nonparametric, and bootstrapping methods.Examples of many hypothesis tests.Access free downloadable datasets so you can try it yourself.
409 pages, Kindle Edition
Published September 17, 2020
90 people are currently reading
98 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 6 of 6 reviews
October 30, 2021
If you took statistics in college and need an easy-to-read, clear review of the most important concepts in statistics related to hypothesis testing, look no further. Every professor should teach like Jim frost.
Alternatively, if you are currently taking a course in college and need to understand things intuitively before they shove all those formulas down your throat, then this should be THE companion to your textbook.
Excellent work Mr. Frost! Highly recommended.
Now, get to work on a bayesian counterpart to this one :)
Alternatively, if you are currently taking a course in college and need to understand things intuitively before they shove all those formulas down your throat, then this should be THE companion to your textbook.
Excellent work Mr. Frost! Highly recommended.
Now, get to work on a bayesian counterpart to this one :)
October 28, 2024
This will be my reference point whenever I'll need to refresh my memory around the intricate topic of Hypothesis Testing.
The book introduces readers to different types of hypothesis tests, such as t-tests, ANOVA, and chi-square tests, with an emphasis on when and why to use each. Frost also covers one-tailed vs. two-tailed tests, Type I and Type II errors, and power analysis, making the book a comprehensive guide to the basics of testing.
Frost takes time to discuss common mistakes in hypothesis testing, such as data dredging and misinterpreting p-values. He emphasizes the importance of context and encourages readers to look beyond just the numbers to understand the data's story.
First instance this explanation of Type I error aka false positive is so on point and accessible to everyone
He succeeds in explaining very convoluted and difficult topics, such as why we do not accept the Null Hypothesis, in such a good manner
Frost highlights common pitfalls, like misinterpreting p-values or ignoring statistical power, that can lead to poor decision-making. This focus on interpretation is invaluable and helps readers avoid drawing incorrect conclusions from their data.
Another example is the difficult explanation of what is a P-value.
Of course you cannot simply too much complex concept, but you can try to make them more accessible as Frost does.
At last he explains thoroughly Central Limit Theorem and why it is so important: the distribution of sample means (or sample sums) from a population will tend to follow a normal distribution, regardless of the population's original distribution, as long as the sample size is sufficiently large. This is true even if the data itself doesn’t follow a normal distribution, which is a powerful feature of the CLT.
In hypothesis testing, the CLT justifies the use of normal distribution-based methods (like z-tests or t-tests) to assess sample means or proportions when making data-driven decisions. For instance, it allows researchers to test if a sample mean significantly differs from a hypothesized population mean even if they don’t know the underlying distribution of the data.
In fact, Even if the underlying data is skewed, multimodal, or has any other shape, the distribution of the sample means will approach a normal (bell-shaped) distribution as the sample size increases.
The book introduces readers to different types of hypothesis tests, such as t-tests, ANOVA, and chi-square tests, with an emphasis on when and why to use each. Frost also covers one-tailed vs. two-tailed tests, Type I and Type II errors, and power analysis, making the book a comprehensive guide to the basics of testing.
Frost takes time to discuss common mistakes in hypothesis testing, such as data dredging and misinterpreting p-values. He emphasizes the importance of context and encourages readers to look beyond just the numbers to understand the data's story.
First instance this explanation of Type I error aka false positive is so on point and accessible to everyone
A fire alarm provides a good analogy for the types of hypothesis testing errors. Ideally, the alarm rings when there is a fire and does not ring in the absence of a fire. However, if the alarm rings when there is no fire, it is a false positive, or a Type I error in statistical terms. Conversely, if the fire alarm fails to ring when there is a fire, it is a false negative, or a Type II error.
He succeeds in explaining very convoluted and difficult topics, such as why we do not accept the Null Hypothesis, in such a good manner
You learned that we do not accept the null hypothesis. Instead, we fail to reject it. The convoluted wording encapsulates the fact that insufficient evidence for an effect in our sample isn't proof that the effect does not exist in the population. The effect might exist, but our sample didn't detect it-just like all those species scientists presumed were extinct because they didn't see them.
Finally, I admonished you not to use the common practice of seeing whether confidence intervals of the mean for two groups overlap to determine whether the means are different. That process can cause you to overlook significant results and miss out on important information about the likely range of the mean difference. Instead, assess the confidence interval of the difference between means.
Frost highlights common pitfalls, like misinterpreting p-values or ignoring statistical power, that can lead to poor decision-making. This focus on interpretation is invaluable and helps readers avoid drawing incorrect conclusions from their data.
Another example is the difficult explanation of what is a P-value.
Of course you cannot simply too much complex concept, but you can try to make them more accessible as Frost does.
P-values indicate the strength of the sample evidence against the null hypothesis. If it is less than the significance level, your results are statistically significant.
This is because it is the probability of observing a sample statistic that is at least as extreme as your sample statistic when you assume the Null Hypothesis is correct,.
P-values are the probability that you would obtain the effect observed in your sample, or larger, if the null hypothesis is correct. In simpler terms, p-values tell you how strongly your sample data contradict the null. Lower p-values represent stronger evidence against the null.
If the p-value is less than or equal to the significance level, you reject the null hypothesis and your results are statistically significant. The data support the alternative hypothesis that the effect exists in the population. When the p-value is greater than the significance level, your sample data don't provide enough evidence to conclude that the effect exists.
At last he explains thoroughly Central Limit Theorem and why it is so important: the distribution of sample means (or sample sums) from a population will tend to follow a normal distribution, regardless of the population's original distribution, as long as the sample size is sufficiently large. This is true even if the data itself doesn’t follow a normal distribution, which is a powerful feature of the CLT.
In hypothesis testing, the CLT justifies the use of normal distribution-based methods (like z-tests or t-tests) to assess sample means or proportions when making data-driven decisions. For instance, it allows researchers to test if a sample mean significantly differs from a hypothesized population mean even if they don’t know the underlying distribution of the data.
In fact, Even if the underlying data is skewed, multimodal, or has any other shape, the distribution of the sample means will approach a normal (bell-shaped) distribution as the sample size increases.
July 25, 2024
Practical Hypothesis Testing Needs Humans!
This is an excellent book discussing statistical hypothesis testing. It takes a practical perspective, illustrated with plots and graphs, to convey ideas. His discussion at the end of the book regarding nonparametric methods and bootstrapping a sample were short but useful indicating when and why these methods can be used.
I especially liked the chapters on interpreting p-values and types of error and statistical power. There are numerous examples of research that confuse p-values with the probability of a type 1 error when in fact it is the probability of getting the results for the given sample if the null hypothesis is true (i.e. no change). Actually the probability of a type 1 error is the significance level alpha set before any results are gathered and is based on the researcher’s prior domain knowledge. So it’s an educated guess. Further the author demonstrates the effect of sample size on results by showing how too small a sample can yield exaggerated results (since it takes more extreme values to make the results statistically significant) and very large samples can make small differences statistically significant. This discussion underlies the fact that study replication (with appropriate sample size) is essential before coming to any reliable conclusions. It also shows the time, expense, and effort genuinely required to perform viable scientific research. Sadly today’s scientific research tends to play fast and loose with statistical significance undermining confidence in published studies which often turn out to be “just so� stories wrapped in technical jargon.
One other observation about the book is the author repeatedly called upon the analyst to exercise significant domain knowledge to evaluate the statistical results drawn from various methods (e.g. statistical power, p-value quality, significance level, outlier identification, continuous variable probability distribution identification, etc) in order to produce correct results. It was striking to me to see this insistence on human expertise to provide overall guidance of result evaluation and interpretation in an age where company management often insists on replacing humans with automation at every conceivable turn to reduce costs and increase profits. Taking this author seriously would question the quality of the results of this management approach.
This is an excellent book discussing statistical hypothesis testing. It takes a practical perspective, illustrated with plots and graphs, to convey ideas. His discussion at the end of the book regarding nonparametric methods and bootstrapping a sample were short but useful indicating when and why these methods can be used.
I especially liked the chapters on interpreting p-values and types of error and statistical power. There are numerous examples of research that confuse p-values with the probability of a type 1 error when in fact it is the probability of getting the results for the given sample if the null hypothesis is true (i.e. no change). Actually the probability of a type 1 error is the significance level alpha set before any results are gathered and is based on the researcher’s prior domain knowledge. So it’s an educated guess. Further the author demonstrates the effect of sample size on results by showing how too small a sample can yield exaggerated results (since it takes more extreme values to make the results statistically significant) and very large samples can make small differences statistically significant. This discussion underlies the fact that study replication (with appropriate sample size) is essential before coming to any reliable conclusions. It also shows the time, expense, and effort genuinely required to perform viable scientific research. Sadly today’s scientific research tends to play fast and loose with statistical significance undermining confidence in published studies which often turn out to be “just so� stories wrapped in technical jargon.
One other observation about the book is the author repeatedly called upon the analyst to exercise significant domain knowledge to evaluate the statistical results drawn from various methods (e.g. statistical power, p-value quality, significance level, outlier identification, continuous variable probability distribution identification, etc) in order to produce correct results. It was striking to me to see this insistence on human expertise to provide overall guidance of result evaluation and interpretation in an age where company management often insists on replacing humans with automation at every conceivable turn to reduce costs and increase profits. Taking this author seriously would question the quality of the results of this management approach.
May 29, 2022
Excellent refresher of (or introduction to) hypothesis testing. The writing is clear, intuitive, and at the same time sufficiently in-depth (for a statistics book that sports only a tiny handful of simple equations).
November 12, 2020
Must read for novices
You need to understand this stuff in your gut. Formulas are good after that. Start with this book. Easy pleasant reading.
You need to understand this stuff in your gut. Formulas are good after that. Start with this book. Easy pleasant reading.
June 27, 2024
definitely the book you need to read if you want to know statistics in a fun way. top quality!
Displaying 1 - 6 of 6 reviews
Can't find what you're looking for?
Get help and learn more about the design.