# Announcing Release 2.8 of the Real Statistics Resource Pack

This release provides the following new functionality:

New (Binary) Logistic Regression supplemental functions

In addition to the Logistic Regression data analysis tool, available in previous releases, the latest release provide similar functionality in the form of the following functions that can be used to build formulas for use in Excel spreadsheets.

• LogitCoeff – calculates the logistic regression coefficients for data in raw or summary form. Includes the standard errors, Wald statistic, p-value and 95% confidence interval
• LogitTest – calculates LL of the full and reduced models, the chi-square statistic and the p-value
• LogitRSquare – calculates LL of the full and reduced models and three versions of R2 (McFadden, Cox and Snell, Nagelkerke) as well as AIC and BIC
• LogitPred – calculates the probability of success for any inputted values of the independent variables based on the logistic regression model
• LogitPredC – like LogitPred, except that the model coefficients are used as input (instead of building the model from scratch)

New Multinomial Logistic Regression supplemental functions

The latest release provides supplemental Excel functions similar to the binary logistic regression functions described above, but using a multinomial regression model. This enables models to be built with more than two outcomes,

• MLogitCoeff – calculates the multinomial logistic regression coefficients for data in raw or summary form.
• MLogitParam – create a table for an outcome of the dependent variable versus the reference value, including the coefficients, standard errors, Wald statistic, p-value and 95% confidence interval
• MLogitTest – calculates LL of the full and reduced multinomial logistic regression models, the chi-square statistic and the p-value
• MLogitRSquare – calculates LL of the full and reduced models and three versions of R2 (McFadden, Cox and Snell, Nagelkerke) as well as AIC and BIC
• MLogitPred – calculates the probabilities of each outcome of multinomial logistic regression model for any inputted values of the independent variables
• MLogitPredC – like MLogitPred, except that the model coefficients are used as input (instead of building the model from scratch)
• MLogitSummary – converts raw data to summary form for use in building a multinomial logistic regression model
• MLogitExtract – used to create a submodel of a multinomial logistic regression model based on a reduced set of outcomes for the dependent variable

Improvements to the Frequency Table data analysis tool

• Convert raw data to a frequency table based on bins of fixed length
• Create descriptive statistics (mean, median, standard deviation, IQR, MAD, etc,) for data in a frequency table
• Convert data in a frequency table to raw data form
• Create a histogram from the frequency tables

Improvement to the Binary Logistic Regression data analysis tool

• Ability to specify that certain independent variables be removed from the model to create a reduced logistic regression model

Bug fixes

• Enables logistic regression models using the Logistic Regression data analysis tool when the input range contains only two rows of data
• Fixes an error in the NORM_CONF, NORM_LOWER and NORM_UPPER functions (thanks to Colin for identifying this error)

I hope that you find the new capabilities useful.

Charles

This entry was posted in Uncategorized. Bookmark the permalink.

### One Response to Announcing Release 2.8 of the Real Statistics Resource Pack

1. Colin says:

Sir
Great News! Thank your updating. “Real Statistics Resource Pack” really make people who love Excel and statistics more powerful. You are a good man.