Package 'JustifyAlpha'

Justify your alpha level by avoiding the Lindley paradox or aiming for moderate or strong evidence when using anova.

Description

Justify your alpha level by avoiding the Lindley paradox or aiming for moderate or strong evidence when using anova.

Usage

ftestEvidence(evidence, df1, df2, paired = FALSE, printplot = FALSE)

Arguments

evidence	Desired level of evidence: "Lindley" to avoid the Lindley Paradox, "moderate" to achieve moderate evidence and "strong" to achieve strong evidence. Users that are more familiar with Bayesian statistics can also directly enter their desired Bayes factor.
df1	Numerator degrees of freedom.
df2	Denominator degrees of freedom.
paired	If true a within subjects design is assumed.
printplot	If true prints a plot relating Bayes factors and p-values.

Value

numeric alpha level required to avoid Lindley's paradox.

References

Maier & Lakens (2021). Justify Your Alpha: A Primer on Two Practical Approaches

Examples

## Avoid the Lindley paradox for an anova with 1 numerator and 248 denominator degrees of freedom.
ftestEvidence("lindley", 1, 248)

---

Justify your alpha level by minimizing or balancing Type 1 and Type 2 error rates.

Description

Justify your alpha level by minimizing or balancing Type 1 and Type 2 error rates.

Usage

optimal_alpha(
  power_function,
  costT1T2 = 1,
  priorH1H0 = 1,
  error = "minimize",
  verbose = FALSE,
  printplot = FALSE
)

Arguments

power_function	Function that outputs the power, calculated with an analytic function.
costT1T2	Relative cost of Type 1 errors vs. Type 2 errors.
priorH1H0	How much more likely a-priori is H1 than H0?
error	Either "minimize" to minimize error rates, or "balance" to balance error rates.
verbose	Print each iteration of the optimization function if TRUE. Defaults to FALSE.
printplot	Print a plot to illustrate the alpha level calculation.

Value

Returns a list of the following alpha = alpha or Type 1 error that minimizes or balances combined error rates, beta = beta or Type 2 error that minimizes or balances combined error rates, errorrate = weighted combined error rate, objective = value that is the result of the minimization, either 0 (for balance) or the combined weighted error rates. plot_data = data used for plotting (only if printplot = TRUE) plot = plot of error rates depending on alpha (only if printplot = TRUE)

References

Maier & Lakens (2021). Justify Your Alpha: A Primer on Two Practical Approaches

Examples

## Optimize power for a independent t-test, smallest effect of interest
## d = 0.5, 100 participants per condition
res <- optimal_alpha(power_function = "pwr::pwr.t.test(d = 0.5, n = 100,
sig.level = x, type = 'two.sample', alternative = 'two.sided')$power")
res$alpha
res$beta
res$errorate

---

Justify your alpha level by minimizing or balancing Type 1 and Type 2 error rates.

Description

Justify your alpha level by minimizing or balancing Type 1 and Type 2 error rates.

Usage

optimal_sample(
  power_function,
  errorgoal = 0.05,
  costT1T2 = 1,
  priorH1H0 = 1,
  error = "minimize",
  printplot = FALSE
)

Arguments

power_function	Function that outputs the power, calculated with an analytic function.
errorgoal	Desired weighted combined error rate
costT1T2	Relative cost of Type 1 errors vs. Type 2 errors.
priorH1H0	How much more likely a-priori is H1 than H0?
error	Either "minimize" to minimize error rates, or "balance" to balance error rates.
printplot	Print a plot to illustrate the alpha level calculation. This will make the function considerably slower.

Value

Returns a list of the following alpha = alpha or Type 1 error that minimizes or balances combined error rates, beta = beta or Type 2 error that minimizes or balances combined error rates, errorrate = weighted combined error rate, objective = value that is the result of the minimization, either 0 (for balance) or the combined weighted error rates, samplesize = the desired samplesize. plot = plot of alpha, beta, and error rate as a function of samplesize (only if printplot = TRUE)

References

Maier & Lakens (2021). Justify Your Alpha: A Primer on Two Practical Approaches

Examples

## Optimize power for a independent t-test, smallest effect of interest
## d = 0.5, desired weighted combined error rate = 5%
res <- optimal_sample(power_function = "pwr::pwr.t.test(d = 0.5, n = sample_n, sig.level = x,
type = 'two.sample', alternative = 'two.sided')$power",errorgoal = 0.05)
res$alpha
res$beta
res$errorrate
res$samplesize

---

Launch the Justify your alpha shiny app.

Description

Launch the Justify your alpha shiny app.

Usage

runApp()

---

Justify your alpha level by avoiding the Lindley paradox or aiming for moderate or strong evidence when using a t-test.

Description

Justify your alpha level by avoiding the Lindley paradox or aiming for moderate or strong evidence when using a t-test.

Usage

ttestEvidence(
  evidence,
  n1,
  n2 = 0,
  one.sided = FALSE,
  rscale = sqrt(2)/2,
  printplot = FALSE
)

Arguments

evidence	Desired level of evidence: "Lindley" to avoid the Lindley Paradox, "moderate" to achieve moderate evidence and "strong" to achieve strong evidence. Users that are more familiar with Bayesian statistics can also directly enter their desired Bayes factor.
n1	Sample size in Group 1.
n2	Sample size in Group 2. Leave blank for a one-sample or paired-sample
one.sided	Indicates whether the test is one sided or two sided.
rscale	Scale of the Cauchy prior
printplot	If true prints a plot relating Bayes factors and p-values.

Value

numeric alpha level required to avoid Lindley's paradox.

References

Maier & Lakens (2021). Justify Your Alpha: A Primer on Two Practical Approaches

Examples

## Avoid the Lindley paradox for a two sample t-test with 300 participants per condition
ttestEvidence("lindley", 300, 300)

---