Cramer's V is calculated as:

where

N |
is the total number of observations |

min(i-1)(j-1) |
is the minimum of row dimension minus 1 and column dimension minus 1 |

Since V is based on the Chi-square statistic, it can be interpreted as reflecting the squared departure of the expected from the observed frequencies, on a scale from 0 to 1 (or slightly less than 1). It's mostly useful for comparing results from several tables. For more details concerning this statistic, see Bishop, Fienberg, Holland (1975, p. 385-387).