Skip to contents

Confidence interval for a 2-group median difference

Source: R/statpsych1.R
ci.median2.Rd

Computes a distribution-free confidence interval for a difference of population medians in a 2-group design. Tied scores within each group are assumed to be rare.

For more details, see Section 2.12 of Bonett (2021, Volume 1)

Usage

ci.median2(alpha, y1, y2)

Arguments

alpha

alpha level for 1-alpha confidence

y1

vector of scores for group 1

y2

vector of scores for group 2

Value

Returns a 1-row matrix. The columns are:

  • Median1 - estimated median for group 1

  • Median2 - estimated median for group 2

  • Median1-Median2 - estimated difference of medians

  • SE - standard error

  • LL - lower limit of the confidence interval

  • UL - upper limit of the confidence interval

References

Bonett DG, Price RM (2002). “Statistical inference for a linear function of medians: Confidence intervals, hypothesis testing, and sample size requirements.” Psychological Methods, 7(3), 370–383. ISSN 1939-1463, doi:10.1037/1082-989X.7.3.370 .

Bonett DG (2021). Statistical Methods for Psychologists https://dgbonett.sites.ucsc.edu/.

Examples

y1 = c(70, 394, 43, 95, 62, 128, 2, 203, 81, 436, 85, 35, 
        156, 1, 3, 27, 63, 181, 184, 18)
y2 = c(102, 120, 78, 78, 417, 124, 86, 171, 176, 129, 230, 
       194, 100, 157, 306, 411, 164, 103, 193, 312)
ci.median2(.05, y1, y2)
#>  Median1 Median2 Median1-Median2       SE        LL        UL
#>     75.5   160.5             -85 37.11502 -157.7441 -12.25589

# Should return:
#  Median1 Median2 Median1-Median2       SE        LL        UL
#     75.5   160.5             -85 37.11502 -157.7441 -12.25589