To calculate the quasi variances for a categorical variable with levels {1,...,p}, type the number "p" in the levels box, enter the p x p variance-covariance matrix for the levels of the categorical variable in the covariance matrix box, and optionally add estimates for the coefficients of the categories in the estimates box (this will allow for plotting of comparison intervals). Numbers should be seperated by spaces in all boxes, and rows on seperate lines for the covariance matrix; press the "Example" button for a demonstration.

Input to the "Covariance Matrix" can be either:

• Square Variance-Covariance matrix
• Lower Triangle of Variance-Covariance matrix
• One line list of standard errors, then lower triangle of correlation matrix.

The number of levels should be equal to the number of lines in the "Covariance Matrix". Currently scientific notation is not recognised.

Input to "Estimates" is optional - it should be a list seperated by spaces (not semi-colons or commas). If a valid list is given, a plot of estimates will be shown and an estimates column will be included in the table of quasi variances.

The Kuvee calculator implements the procedure suggested by Professor D. Firth to calculate quasi variances for categorical variables from a covariance matrix or standard errors with correlation matrix. For more information on quasi variances, see here.

If you have any suggestions or find any bugs, please send an email with details.

For more advanced users interested in minimising the maximum error in the quasi variances instead of the average error, please see the Kuvee2 page.

Covariance Matrix