The noncentrality interval for the association strength in a contingency
table, as effectsize::cramers_v(adjust = FALSE, alternative = "two.sided") computes it: the 2-sided interval for the noncentrality of
the chi-square statistic, each bound turned into V as
\(\sqrt{\lambda / (n (k - 1))}\), where k is the smaller of the number
of rows and columns. For a 2 by 2 table V equals phi. The interval reaches
0 when the test is far from significant. Until 0.4.3 this was a percentile
bootstrap, whose resampled V is 0 or above, so an exactly independent
table had the interval [0.05, 0.50] beside V = 0 (FR-24).
Arguments
- tab
A contingency table (from
table()) or a matrix of counts.- R
Kept so existing calls still run. Unused since 0.4.3.
- conf.level
Confidence level. Defaults to 0.95.
- seed
Kept so existing calls still run. Unused since 0.4.3, as the interval uses no resampling.
Value
A named numeric vector: estimate, lower, upper, with the
attributes level and method ("noncentral chi-square"). The bounds
are NA when the table holds fewer than 3 observations or has fewer than
2 rows or columns.
See also
sframe_eta_sq_ci(), sframe_bootstrap_ci()
Other Confidence interval functions:
sframe_bootstrap_ci(),
sframe_cohens_d_ci(),
sframe_eta_sq_ci()
Examples
sframe_cramers_v_ci(table(mtcars$am, mtcars$cyl))