DMAR (version 1.0.0)

ss_power_smd: Sample Size or Power for a Standardized Mean Difference (Two Independent Groups)

Description

Determine the necessary per-group sample size to achieve a desired level of statistical power for the two-sample (independent groups) t-test on a standardized mean difference (Cohen's d; equivalently Hedges' g and Glass's g for sample size purposes). Alternatively, given a per-group sample size, return the realized statistical power.

Usage

ss_power_smd(
  smd,
  desired_power = 0.85,
  alpha_level = 0.05,
  n_1 = NULL,
  n_2 = NULL,
  directional = FALSE
)

Value

A data.frame with term and value columns. The design result comes first, followed by rows that echo the user-supplied planning inputs, so the assumptions the power was evaluated under travel with the result. The supposed_smd row is the supposed effect the plan is built on: a value the researcher posits, either a minimally important effect or a value believed to be true in the population, never a sample estimate. The tails row is 2 for a nondirectional test and 1 for a directional test.

When n_1 is NULL

Result rows necessary_n_per_group, actual_power, and noncentral_t_parm, then the planning inputs supposed_smd, desired_power, alpha_level, and tails.

When n_1 is specified

Result rows specified_n_1, specified_n_2, actual_power, and noncentral_t_parm, then the planning inputs supposed_smd, alpha_level, and tails (the supplied group sizes are the specified_n_1 / specified_n_2 rows).

Arguments

smd

Supposed standardized mean difference (Cohen's d) the design is planned against: a value the researcher posits for the population, either a minimally important effect or a value believed to be true in the population, never a sample estimate. Echoed in the returned table as the supposed_smd row.

desired_power

Desired statistical power (default 0.85)

alpha_level

Type I error rate (default 0.05)

n_1

Sample size for group 1 (if specified, the function returns the realized power; assumes n_2 = n_1 unless n_2 is also given)

n_2

Sample size for group 2 (defaults to n_1 when n_1 is supplied)

directional

Logical: TRUE for a one-sided test (in the same sign as smd), FALSE (default) for a two-sided test

Author

Ken Kelley [email protected]

Details

The two-sample t-statistic with pooled standard deviation follows a noncentral t-distribution with \(n_1 + n_2 - 2\) degrees of freedom and noncentrality parameter \(\lambda = \delta \sqrt{n_1 n_2 / (n_1 + n_2)}\), where \(\delta\) is the population standardized mean difference. For balanced designs (\(n_1 = n_2 = n\)) this simplifies to \(\lambda = \delta \sqrt{n / 2}\).

Power is computed as the probability that the absolute value of the test statistic exceeds the critical value(s) under the alternative; the function returns the per-group sample size for which power first reaches desired_power.

Kelley and Rausch (2006) develop the accuracy in parameter estimation approach to planning the sample size for the standardized mean difference, implemented in ss_aipe_smd.

References

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

Kelley, K., Maxwell, S. E., & Rausch, J. R. (2003). Obtaining power or obtaining precision: Delineating methods of sample size planning. Evaluation and the Health Professions, 26(3), 258--287. tools:::Rd_expr_doi("10.1177/0163278703255242")

Kelley, K., & Rausch, J. R. (2006). Sample size planning for the standardized mean difference: Accuracy in parameter estimation via narrow confidence intervals. Psychological Methods, 11(4), 363--385. tools:::Rd_expr_doi("10.1037/1082-989X.11.4.363")

Maxwell, S. E., Delaney, H. D., & Kelley, K. (2027). Designing experiments and analyzing data: A model comparison perspective (4th ed.). Routledge.

Maxwell, S. E., Kelley, K., & Rausch, J. R. (2008). Sample size planning for statistical power and accuracy in parameter estimation. Annual Review of Psychology, 59, 537--563. tools:::Rd_expr_doi("10.1146/annurev.psych.59.103006.093735")

See Also

ss_aipe_smd, ci_smd, smd, ci_nc_t

design_consequences for what a chosen design delivers: power, the Type S (sign) and Type M (exaggeration) errors of the significance filter, and the expected confidence interval width.

Other sample size for power: power_fisher_exact(), ss_aipe_mixed_effects(), ss_power_R2(), ss_power_R2_sensitivity(), ss_power_c(), ss_power_c_ancova(), ss_power_composite_ancova(), ss_power_composite_ancova_2group(), ss_power_composite_anova(), ss_power_composite_factorial_ancova(), ss_power_composite_factorial_ancova_het(), ss_power_composite_factorial_anova(), ss_power_composite_sem(), ss_power_contrast(), ss_power_equivalence_c(), ss_power_factorial_ancova(), ss_power_factorial_anova(), ss_power_indirect_effect(), ss_power_mixed_effects(), ss_power_one_way_anova(), ss_power_pcm(), ss_power_r(), ss_power_rc(), ss_power_reg_coef(), ss_power_reg_coef_sensitivity(), ss_power_rm_anova(), ss_power_sc(), ss_power_sem(), ss_power_split_plot_anova()

Examples

# Per-group sample size for d = 0.5, alpha = .05, power = .80, two-sided
ss_power_smd(smd = 0.5, desired_power = 0.80)

# Same with a directional (one-sided) test
ss_power_smd(smd = 0.5, desired_power = 0.80, directional = TRUE)

# Realized power given balanced n = 30 per group
ss_power_smd(smd = 0.5, n_1 = 30)

# Realized power for unbalanced (n_1 = 30, n_2 = 50)
ss_power_smd(smd = 0.5, n_1 = 30, n_2 = 50)