I am pleased to announce Release 5.7 of the Real Statistics Resource Pack. The new release is now available for free download at for Excel 2007, 2010, 2013 and 2016 (Windows) environments.
Release 5.7 will be available for Excel 2011 and Excel 2016 for Mac in about a week. The Real Statistics 1B and 2A examples workbooks have also been updated for compatibility with the new release. The Real Statistics website will be updated over the course of the next several days to reflect the new capabilities. If you are getting value from the Real Statistics website or software, I would appreciate your donations to help offset the costs of the website by going to. The following is a summary of the new features in Release 5.7.
The first few new features provide additional support for missing data and unbalanced models. Mann-Whitney Exact Test Support for the Mann-Whitney Exact Test has been completely revised. This test now supports much larger sample sizes. The recommended limits are now 300 for the size of the smaller sample and 1,000 for the larger sample (instead of only about 30 between the two samples in the previous versions of Real Statistics). This means that when the Two samples and Non-parametric options of the T Tests and Nonparametric Equivalents data analysis tool have been chosen, the output has now been revised to show the results of the exact test most of the time.
It also means that the following Real Statistics functions have been revised to better support the Mann-Whitney Exact test. PERM2DIST( x, n1, n2, cum) = value of the Mann-Whitney version of the two-sample permutation distribution at x based on n1 and n2 elements; returns the pdf value if cum = FALSE and the cdf value if cum = TRUE (default). PERM2INV( p, n1, n2) = inverse of the Mann-Whitney version of the two-sample permutation distribution at p; i.e. The least value of x such that PERM2DIST( x, n1, n2, TRUE) ≥ p. Note that the slow argument has been eliminated since it is no longer necessary. You will need to modify any existing spreadsheets that use the slow argument by dropping the value of this argument from any formula that uses PERM2DIST( x, n1, n2, cum, slow) or PERM2INV( p, n1, n2, slow).
A separate security update release is available for Mac users running OS X El Capitan and Yosemite. Separately from the Mac OS update, Apple has released iOS 10.3.2 for iPhone and iPad, watchOS 3.2.2 for Apple Watch, and tvOS 10.2.1 for Apple TV. Always back up a Mac before installing any system software update. Includes several improvements to Directory Service and Client Management, which are described in the About Mac OS X Server 10.5.7 Update article. For a complete list of fixes and enhancements, check out Apple’s official support article detailing the release. As always, let us know if the update brings any unpleasant surprises.
Note that PERM2DIST( x, n1, n2, cum, FALSE) and PERM2INV( p, n1, n2, FALSE) return the Wilcoxon Ranked Sum versions of the two-sample permutation distribution and its inverse. The previous version of the PERM2DIST and PERM2INV functions was based on the Wilcoxon Ranked Sum version of the test, while the new version is based on the Mann-Whitney test unless the FALSE argument is added. In addition, the MANNDIST, MANNINV and MANNEXACT functions have been replaced by the following functions: MWDIST( x, n1, n2, tails) = p-value of the Mann-Whitney exact distribution at x based on n1 and n2 elements, where tails = 1 or 2 (default) MWINV( p, n1, n2, tails) = inverse of the Mann-Whitney exact distribution at p; i.e. The least value of x such that MWDIST( x, n1, n2, cum, tails) ≥ p, where tails = 1 or 2 (default) MWEXACT(R1, R2, tails) = p-value of the Mann-Whitney exact test on the data in ranges R1 and R2, where tails = 1 or 2 (default) Note that MWDIST( x, n1, n2, tails, FALSE) and MWINV( p, n1, n2, FALSE) return the Wilcoxon Ranked Sum versions of the Mann-Whitney distribution and its inverse. The following new array function has also been added: PERM2DIST( n1, n2, cum) returns a column array with the p-values of the Mann-Whitney exact test for values of U from 0 to n1.
n2 when cum = TRUE (default) and the frequency values when cum = FALSE. In addition, the Mann-Whitney Table of critical values has been expanded on the website, although this table is really no longer needed since the MWINV function returns a wide range of critical values, including all the values from the table.
Also SRDIST and SRINV should now be used instead of the SRankPROB and SRankCRIT functions. Note that the SRankPROB and SRankCRIT functions are still available for backwards compatibility, but they will no longer be supported or included in the reference materials on the website. The values output by these functions will now be calculated using SRDIST and SRINV. Mann-Whitney Simulation The following new function creates a simulated version of the exact test. This function can be useful for very large samples where the exact test can’t be used or where there are many ties. MWSIMUL(R1, R2, lab, iter): returns a column array with the U value for R1 and R2 (i.e.
Release 5.7 For Mac Pro
U = MANN(R1, R2)), the p-value for the left tailed test, the p-value for the right tailed test and the p-value for the two-tailed test; if lab = TRUE (default FALSE) then an extra column of labels is appended to the output; iter = # of iterations in the simulation (default 10,000). Other Mann-Whitney Test Improvements The following functions replace the MANNTEST and MANNTEST functions: MWTEST(R1, R2, tails, ties, cont) = p-value of the Mann-Whitney U test for the samples contained in ranges R1 and R2 using the normal approximation. Tails = # of tails = 1 or 2 (default). If ties = TRUE (default) the ties correction is applied. If cont = TRUE (default) a continuity correction is applied.
MWTEST(R1, R2, lab, tails, tails, ties, cont, exact, iter): returns a column array with the values U-stat, z-stat, r effect size and three types of p-values for the samples contained in ranges R1 and R2 using the normal approximation, exact test and simulation. Tails = 1 or 2 (default). For the normal approximation, if ties = TRUE (default) the ties correction factor is applied, while if cont = TRUE (default) a continuity correction is applied. If exact = TRUE (default FALSE) then the p-value of the exact test is output and if iter ≠ 0 then the p-value of the simulation version of the test is output where the simulation consists of iter samples (default 10,000). If lab = TRUE (default FALSE) then an extra column of labels is appended to the output.
Note that the MANNTEST and MANNTEST functions are still available for backwards compatibility, but they will no longer be supported or included in the reference materials on the website. The T Tests and Nonparametric Equivalents data analysis tool has also been revised so that when the Two independent samples and Non-parametric options are chosen (representing the Mann-Whitney test), the output is similar to that obtained from the MWTEST function. In particular, the p-values of the normal approximation, exact test and simulation test are output. Wilcoxon Signed-Ranks Exact Test Support for the Wilcoxon Signed-Ranks Exact Test.
Has been completely revised. This test now supports much larger sample sizes. The recommended limits are now 1,000 for the size of the sample (instead of only about 30 in the previous versions of Real Statistics). This means that when the One sample or Paired samples and Non-parametric options of the T Tests and Nonparametric Equivalents data analysis tool are chosen, the output has now been revised to show the results of the exact test most of the time. It also means that the following Real Statistics functions have been revised to better support the Wilcoxon Signed-Ranks Exact test.
Release 5.7 For Mac Free
PERMDIST( x, n, cum) = value of the one-sample permutation distribution at x based on a sample with n elements; returns the pdf value at x if cum = FALSE and the cdf value if cum = TRUE (default) PERMINV( p, n) = inverse of the permutation distribution at x; i.e. The least value of x such that PERMDIST( x, n, TRUE) ≥ p Note that the slow argument has been eliminated since it is no longer necessary.
You will need to modify any existing spreadsheets that use the slow argument by dropping the value of this argument from any formula that uses PERMDIST( x, n, cum, slow) or PERMINV( p, n, slow). The following functions have also been added: SRDIST( x, n, tails) = p-value of the signed-ranks exact distribution at x based a sample with n elements where tails = 1 or 2 (default) SRINV( p, n, tails) = inverse of the signed-ranks exact distribution at p; i.e. The least value of x such that SRDIST( x, n, tails) ≥ p, where tails = 1 or 2 (default) In addition, the SRANKPairEXACT and SRANKEXACT functions have been replaced by the following function. SREXACT(R1, R2, tails) = p-value of the paired signed-ranks exact test on the data in ranges R1 and R2 where tails = 1 or 2 (default) SREXACT(R1, med, tails) = p-value of the one-sample signed-ranks exact test on the data in range R1 for the hypothetical median med where tails = 1 or 2 (default) In addition, the Signed-Ranks Table of critical values has been revised and expanded on the website, although this table is really no longer needed since the SRINV function returns a wide range of critical values, including all the values from the table. Also SRDIST and SRINV should now be used instead of SRankPROB and SRankCRIT. Note that the SRANKPairEXACT, SRANKEXACT, SRankCRIT and SRankPROB functions are still available for backwards compatibility, but they will no longer be supported or included in the reference materials on the website.
The values output by SRANKPairEXACT and SRANKEXACT will now be calculated using SREXACT. Similarly, the values output by SRankCRIT and SRankPROB will now be calculated using SRDIST and SRINV.
Signed-Ranks Simulation The following new function creates a simulated version of the exact test. This function can be useful for very large samples where the exact test can’t be used or where there are many ties. SRSIMUL(R1, R2, lab, iter): returns a column array with the T-stat for R1 and R2 (i.e. T = SRANK(R1, R2)), the p-value for the left tailed test, the p-value for the right tailed test and the p-value for the two-tailed test; if lab = TRUE (default FALSE) then an extra column of labels is appended to the output; iter = # of iterations in the simulation (default 10,000). SRSIMUL(R1, med, lab, iter): returns a column array with the T-stat for R1 and med (i.e. T = SRANK(R1, med)), the p-value for the left tailed test, the p-value for the right tailed test and the p-value for the two-tailed test; if lab = TRUE (default FALSE) then an extra column of labels is appended to the output; iter = # of iterations in the simulation (default 10,000). Other Signed-Ranks Test Changes and Improvements The SRANK function has been enhanced to support both the paired samples signed-ranks test (when R1 contains two columns or when R2 is present) and the one sample signed-ranks test (when R1 contains one column and the second argument has a numeric value).
SRANK(R1, R2) = T-stat for the paired sample Signed-Ranks test for the samples in the one-column ranges R1 and R2 SRANK(R1) = T-stat for the paired sample Signed-Ranks test for the samples in the two-column range R1 SRANK(R1, med) = T-stat for the one sample Signed-Ranks test for the sample in the one-column range R1 based on a hypothetical median med (default 0) The SRTEST function has been enhanced to support both the one sample and paired sample versions of the signed-ranks test, using the same approach as for SRANK. In addition, SRTEST now supports ties and a continuity correction. SRTEST(R1, R2, tails, ties, cont) = p-value of the paired sample Signed-Ranks test for the samples in ranges R1 and R2 using the normal approximation. Tails = 1 or 2 (default).
If ties = TRUE (default) the ties correction is applied. If cont = TRUE (default) a continuity correction is applied. When R1 contains two columns, this function should take the format SRTEST(R1, tails, ties, cont). SRTEST(R1, med, tails, ties, cont) = p-value of the one sample Signed-Ranks test for the sample in ranges R1 based on the hypothetical median med (default 0) using the normal approximation. Tails = 1 or 2 (default). If ties = TRUE (default) the ties correction is applied.
If cont = TRUE (default) a continuity correction is applied. Note that SRANK can now be used in place of SRankPair. Similarly, SRTEST can now be used in place of the SRTestPair function. SRANKPair and SRTestPair are still available for backwards compatibility, although these functions will no longer be supported or included in the reference materials on the website. The SRTEST array function, which already supported both the one sample and paired signed-ranks test, has been revised as follows: SRTEST(R1, R2, lab, tails, tails, ties, cont, exact, iter): returns a column array with the values T-stat, z-stat, r effect size and three types of p-values for the paired samples contained in ranges R1 and R2 using the normal approximation, exact test and simulation. Tails = 1 or 2 (default).
For the normal approximation, if ties = TRUE (default) the ties correction factor is applied, while if cont = TRUE (default) a continuity correction is applied. If exact = TRUE (default FALSE) then the p-value of the exact test is output and if iter ≠ 0 then the p-value of the simulation version of the test is output where the simulation consists of iter samples (default 10,000). If lab = TRUE (default FALSE) then an extra column of labels is appended to the output.
Release 5.7 For Macos Server
SRTEST(R1, lab, tails, tails, ties, cont, exact, iter): the paired samples version of the SRTEST when R2 contains two columns. SRTEST(R1, med, lab, tails, tails, ties, cont, exact, iter): the one-sample version of the SRTEST function The T Tests and Nonparametric Equivalents data analysis tool has also been revised so that when the One sample or Two-paired samples and Non-parametric options are chosen (representing the Signed-Ranks test), the output is similar to that obtained from the SRTEST function. In particular, the p-values of the normal approximation, exact test and simulation test are output. Studentized Range Q The table of critical values for the studentized range q distribution has been expanded to support up to 100 groups.
The QCRIT( k, df, alpha, tails, interp) function has also been enhanced to accept values of k from 2 to 100.