Skip to contents

Returns the words a published labelling scheme assigns to a Bayes factor, for authors whose venue asks for them. This is the only function in apabayes that turns evidence into words. No other function calls it, and nothing attaches its output to a number unless the user does.

Usage

apa_bf_label(x, scheme, markup = NULL)

Arguments

x

Numeric vector of Bayes factors as BF10; values of 0 or more, NA and Inf allowed.

scheme

The scheme, chosen explicitly: "jeffreys" or "raftery". There is no default; see Details.

markup

NULL or one of "md", "latex", "typst", "plain". NULL uses getOption("apabayes.markup"), then the knitr output format, then "md". Decides the minus sign (U+2212 outside "plain") and the infinity symbol.

Value

A character vector of length(x); NA in gives NA_character_ out.

Details

Verbal categories are interpretation aids, not part of the statistic. The Bayes factor is a continuous measure of relative evidence, and the guidelines this package follows recommend reporting the number itself with its direction made explicit (Tendeiro et al., 2024; van Doorn et al., 2021; Heck et al., 2023). Use a label only where a venue requires one, name the scheme in the text, and never let the label replace the number.

The category is looked up on the evidence in favour of whichever hypothesis the value supports (x for H1, 1 / x for H0), and the result names that hypothesis: "moderate evidence for H~1~". A value of exactly 1 gives "no evidence for either hypothesis". Bounds are inclusive on the upper side: a Bayes factor of 3 is the last value in the lowest category, as effectsize::interpret_bf() reads them.

Thresholds and words (BF in favour, upper bound inclusive):

  • "jeffreys": up to 3 anecdotal, 10 moderate, 30 strong, 100 very strong, above 100 extreme.

  • "raftery": up to 3 weak, 20 positive, 150 strong, above 150 very strong.

Both tables are taken from effectsize::interpret_bf() (version 1.0.3, rules "jeffreys1961" and "raftery1995"), whose help page cites Jeffreys (1961) and Raftery (1995). The primary sources were not consulted when this function was written; check the thresholds against them before relying on the attribution in a manuscript. The scheme of Lee and Wagenmakers (2013), which uses the same five words as the "jeffreys" table above, is not offered until its table has been verified against the book.

References

Heck, D. W., Boehm, U., Böing-Messing, F., Bürkner, P.-C., Derks, K., Dienes, Z., Fu, Q., Gu, X., Karimova, D., Kiers, H. A. L., Klugkist, I., Kuiper, R. M., Lee, M. D., Leenders, R., Leplaa, H. J., Linde, M., Ly, A., Meijerink-Bosman, M., Moerbeek, M., ... Hoijtink, H. (2023). A review of applications of the Bayes factor in psychological research. Psychological Methods, 28(3), 558–579. doi:10.1037/met0000454

Jeffreys, H. (1961). Theory of probability (3rd ed.). Oxford University Press.

Raftery, A. E. (1995). Bayesian model selection in social research. Sociological Methodology, 25, 111–163.

Tendeiro, J. N., Kiers, H. A. L., Hoekstra, R., Wong, T. K., & Morey, R. D. (2024). Diagnosing the misuse of the Bayes factor in applied research. Advances in Methods and Practices in Psychological Science, 7(1). doi:10.1177/25152459231213371

van Doorn, J., et al. (2021). The JASP guidelines for conducting and reporting a Bayesian analysis. Psychonomic Bulletin & Review, 28(3), 813–826. doi:10.3758/s13423-020-01798-5

See also

apa_bf(), which prints the number.

Examples

apa_bf_label(c(0.2, 1, 5, 50, Inf), scheme = "jeffreys")
#> [1] "moderate evidence for H~0~"        "no evidence for either hypothesis"
#> [3] "moderate evidence for H~1~"        "very strong evidence for H~1~"    
#> [5] "extreme evidence for H~1~"        
apa_bf_label(5, scheme = "raftery", markup = "plain")
#> [1] "positive evidence for H1"