# ind.oneway.second: A one-way design with independent samples using published... In rpsychi: Statistics for psychiatric research

## Description

`ind.oneway.second` conducts a one-way design with independent samples, namely one-way randomized-group analysis of variance, using published work.

## Usage

 ```1 2``` ```ind.oneway.second(m, sd, n, unbiased = TRUE, contr = NULL, sig.level = 0.05, digits = 3) ```

## Arguments

 `m` a numeric vector contains the means (length(`m`) >= 2) `sd` a numeric vector contains the sample/unbiased standard deviations (length(`sd`) >= 2) `n` a numeric contains the sample size (length(`n`) >= 2) `unbiased` `sd` contains unbiased standard deviations (`unbiased` = TRUE, default) or sample standard deviations (`unbiased` = FALSE) `contr` a matrix or vector contains the contrast weights `sig.level` a numeric contains the significance level (default 0.05) `digits` the specified number of decimal places (default 3)

## Details

This function conducts a one-way design with independent samples, namely one-way randomized-group analysis of variance, using published work. If you do not specify `contr`, all possible pairwise contrasts will be calculated. Statistical power is calculated using the following specifications:

(a) small (η^2 = 0.01), medium (η^2 = 0.06), and large (η^2 = 0.14) population effect sizes, according to the interpretive guideline for effect sizes by Cohen (1992)

(b) sample size specified by `n`

(c) significance level specified by `sig.level`

## Value

The returned object of `ind.oneway.second` contains the following components:

 `anova.table` returns a ANOVA table containing sums of squares, degrees of freedom, mean squares, F values `omnibus.es` returns a omnibus effect size which is a η^2, and its' confidence interval `raw.contrasts` returns raw mean differences, their confidence intervals, and standard errors `standardized.contrasts` returns standardized mean differences for the contrasts (Hedges's g), their approximate confidence intervals for population standardized mean differences, and standard errors `power` returns statistical power for detecting small (η^2 = 0.01), medium (η^2 = 0.06), and large (η^2 = 0.14) population effect sizes

## Author(s)

Yasuyuki Okumura
Department of Social Psychiatry,
National Institute of Mental Health,
National Center of Neurology and Psychiatry
yokumura@blue.zero.jp

## References

Cohen B (2000) Calculating a factorial ANOVA from means and standard deviations. Understanding Statistics, 1, 191-203.

Cohen J (1992) A power primer. Psychological Bulletin, 112, 155-159.

Kline RB (2004) Beyond significance testing: Reforming data analysis methods in behavioral research. Washington: American Psychological Association.

`ind.oneway`, `samplesize.etasq`
 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19``` ```##Kline (2004) Table 6.3 dat <- data.frame(y = c(9,12,13,15,16, 8,12,11,10,14, 10,11,13,11,15), x = rep(factor(c("a","b","c")), each=5) ) ##contrast 1: a - c, contrast 2: 1/2(a + c) - b my.cont <- matrix(c(1,0,-1,1/2,-1,1/2), ncol=3, nrow=2, byrow=TRUE) ind.oneway.second(m = tapply(dat\$y, dat\$x, mean), sd = tapply(dat\$y, dat\$x, sd), n= tapply(dat\$y, dat\$x, length)) ind.oneway.second(m = tapply(dat\$y, dat\$x, mean), sd = tapply(dat\$y, dat\$x, sd), n= tapply(dat\$y, dat\$x, length), contr = my.cont) ```