Turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- How does JMP calculate p-values?

Topic Options

- Start Article
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Oct 29, 2015 5:12 PM
(7229 views)

What is the underlying equation/algorithm that JMP uses when it calculates p-value?

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

The *p*-value is the probability of a quantile in some distribution. One uses the *cumulative distribution function* (CDF) for the distribution model (e.g., normal distribution) to obtain the probability for a given quantile (e.g., data value or sample statistic). The CDF is the definite integral from negative infinity to the quantile x of the distribution *probability density function* (PDF). How these quantiles are used depends on the question and the distribution.

You can use the probability functions in a column formula or in a script window. For example, the lower one-tailed *p*-value for a *t* quantile of -3 with a sample size of 10 is:

t Distribution**(** -3, 9 **)**

**0.00747818195520711**

For example, the *p*-value for the *F* quantile of 4 used in a one-way ANOVA for testing the mean of 3 groups, each with 8 observations is:

1 - F Distribution**(** 4, 2, 21 **)**

**0.0337383327150672**

I hope that this information helps. I can't say more without more details about your question. Why do you want to know the 'equation/algorithm?' How will you use this information?

Learn it once, use it forever!

1 REPLY 1

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

The *p*-value is the probability of a quantile in some distribution. One uses the *cumulative distribution function* (CDF) for the distribution model (e.g., normal distribution) to obtain the probability for a given quantile (e.g., data value or sample statistic). The CDF is the definite integral from negative infinity to the quantile x of the distribution *probability density function* (PDF). How these quantiles are used depends on the question and the distribution.

You can use the probability functions in a column formula or in a script window. For example, the lower one-tailed *p*-value for a *t* quantile of -3 with a sample size of 10 is:

t Distribution**(** -3, 9 **)**

**0.00747818195520711**

For example, the *p*-value for the *F* quantile of 4 used in a one-way ANOVA for testing the mean of 3 groups, each with 8 observations is:

1 - F Distribution**(** 4, 2, 21 **)**

**0.0337383327150672**

I hope that this information helps. I can't say more without more details about your question. Why do you want to know the 'equation/algorithm?' How will you use this information?

Learn it once, use it forever!