Statistics calculator (t-test, MV outliers)

I created this page so I could plug in values and get Cohen’s d for multiple comparisons, chi square p values, and cut-off values for multivariate (MV) outliers. Hopefully others can use it too! I’ve also included useful links to other statistics sites. Let me know if anything fails to work (for a while the spreadsheets wouldn’t allow people to input values, even though I had changed nothing in the source. I believe it’s fixed now).

Cut-off values for Multivariate outliers

This spreadsheet will calculate the cut-off values (above which the case may be a multivariate outlier) for Cook’s d (4/n-k-1), leverage (2k+2/N), and dfbeta (2/SQRT(N); Belsley, Kuh, & Welsch, 2013).  When I’ve got more time, I will provide definitions and links… for now, if you know what you’re looking for, this calculator should be useful ;-).  .

Independent samples t-test with Cohen’s d

This next one will calculate t (and associated p values for one- and two-tailed tests), and Cohen’s d:

Chi square probability calculator

(gives p for given chi square/degrees of freedom)

d for ANOVA/ANCOVA etc models:

To calculate d for ANOVA and other regression models, I divide the mean difference for any given comparison by the root mean square error. This spreadsheet allows you to plug in the MSE (output available on any statistical software) and mean difference:

Multiple Cohen’s d:

I use the spreadsheet below to copy-paste means and standard deviations from statistical software output tables. Pasting means and standard deviation values into blue cells will result in Cohen’s d in green cells.

Some links to other useful stats tools:

Fourmilab:  Calculates p value for a given chi square/ degrees of freedom value (I use this site when doing chi-square change tests)

Lots of useful tools:  I download the FZT computator, which calculates Steiger’s z / Hotelling’s t and Fisher’s exact z.

A web page with a Steiger’s z calculator for comparing correlations in the same sample.  I haven’t actually tested this one, but it looks legit.  (Steiger’s z is used to test if two correlations are statistically different.  For example, if you want to show that the correlation between x and y (e.g., r = .72) is significantly stronger than the correlation between y and z (e.g., r = .45), you would use Steiger’s z.)

A page that will calculate z to compare correlations in different samples.

Andrew Hayes‘s many macros, etc., for SPSS and SAS

More to come when I’ve got time… You can find more about me here. I will be accepting clients for freelance consulting on research design and data collection starting in 2020 (details).

In the meantime, if you’ve found my statistics calculators useful, please do buy me a coffee!