Scribd Logo
Upload

Welcome to Scribd!

  • 52 views

    Sample Size Calculation For Two Independent Groups: A Useful Rule of Thumb

    Uploaded by

    abdeljelile

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Sample Size Calculation For Two Independent Groups: A Useful Rule of Thumb

    Uploaded by

    abdeljelile
    52 views4 pages

    Original Title

    Sample_Size_Calculation_for_Two_Independent_Groups

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    52 views4 pages

    Sample Size Calculation For Two Independent Groups: A Useful Rule of Thumb

    Uploaded by

    abdeljelile

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 4

    See discussions, stats, and author profiles for this publication at: https://www.researchgate.

    net/publication/287927836

    Sample Size Calculation for Two Independent Groups: A Useful Rule of Thumb

    Article  in  Proceedings of Singapore Healthcare · June 2011


    DOI: 10.1177/201010581102000213

    CITATIONS READS

    7 2,374

    1 author:

    John Carson Allen


    Duke-NUS Medical School
    307 PUBLICATIONS   3,558 CITATIONS   

    SEE PROFILE

    Some of the authors of this publication are also working on these related projects:

    Pubs 1Aug2017-31Jul2018 View project

    eczema View project

    All content following this page was uploaded by John Carson Allen on 14 June 2016.

    The user has requested enhancement of the downloaded file.


    Statistics

    Sample Size Calculation for Two Independent Groups: A Useful Rule


    of Thumb
    John C Allen Jr, PhD
    Duke-NUS Graduate Medical School, Singapore

    Introduction the false negative rate. Using conventional values


    There is a great little book by Gerald van Belle of α = 0.05 and β = 0.20, the ROT for calculating n,
    written “for practitioners of statistical science” the per-group sample size is simply
    called Statistical Rules of Thumb1. It is a collection
    of widely applicable simple rules that capture n 16 16 16
    = = = (1)
    Δ δ2
    the essence of key statistical concepts. One of ( ) σ
    2
    (effect size) 2

    the topics addressed is sample size calculation —


    in particular, the basic problem of calculating a Effect size (δ ) is the standardised difference —
    sample size when the aim is to compare means from the absolute difference Δ divided by the standard
    2 independent samples. deviation σ. The ROT is easily obtained from the
    2-sample normal approximation sample size
    With the widely available and readily accessible formula:
    software for computing sample size, both online
    and commercial, one might question the value of a 2 ( Ζ α ∕2 + Ζ 1–β )2 2(1.96 + 0.84)2 15.68 16
    n= = = ≈
    “rule of thumb” formula for calculating sample size. μ1 – μ2 Δ δ2 δ2
    While a thorough assessment of the conceivable
    ( σ ) 2
    ( σ ) 2

    factors and influences that could impact sample


    size is critical, what is often needed in the early A comparison of samples sizes calculated using
    planning stages is an expeditious calculation to the ROT versus a commercial software package
    allow a gross assessment of resource requirements (PASS©) is given in Table 1, overleaf (α = 0.05,
    and feasibility. This is where the statistical rule of 1 – β = 0.80). Overall, the agreement is very good.
    thumb (ROT) is often very useful. In discussions The total number of subjects required is 2n.
    between statistician and investigator, the ROT
    removes much of the algebraic and notational For α = 0.05, the numerators in equation (1)
    shroud and the problem is reduced to its elemental corresponding to 1 – β = 0.90 and 0.95, respectively,
    form. The complex is made simple. The benefit are 21 and 26. In addition, equation (1) can
    may not lie as much in the ease of calculation as be inverted to calculate either the detectable
    in facilitating communication between researcher difference Δ or effect size δ for a given sample size:
    and statistician.
    4σ 4
    Δ = or effect size = δ = (2)
    2 Independent Samples √n √n
    A sample size calculation with the intent of comparing
    means μ1 and μ2 from 2 independent groups 1 Sample
    usually begins by assuming normal distributions In the 1-sample case where a single mean
    and homogeneous variances σ12 = σ22 = σ 2, equal is compared to a known reference value, for
    sample sizes n1 = n2 = n and a 2-sided test. The false α = 0.05 and 1 – β = 0.80, 0.90 and 0.95, the
    positive rate is set to α and the power for detecting numerator in equation (1) becomes, respectively, 8,
    a difference Δ = | μ1 – μ2 | is set to 1 – β, where β is 11 and 13 — half the number per group required

    138 Proceedings of Singapore Healthcare  Volume 20  Number 2  2011


    Sample Size Calculation for Two Independent Groups: A Useful Rule of Thumb

    Table 1. Sample size calculated using the Rule of Thumb (ROT) versus commercial software (PASS©).
    Effect Size 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
    PASS© 393 176 100 64 45 33 26 21 17
    n per Group
    ROT 400 178 100 64 45 33 25 20 16

    for a 2-sample test. This serves as a reminder that In approach (B) comparing time-averaged
    comparing 2 groups requires 4 times the total differences, the sample size will be labelled
    number of subjects as when comparing a single −
    nμ(r , ρ). The standardised difference δ = Δ ∕ σ is a
    mean to a known reference. meaningful effect size in the average response
    between the 2 groups. For α = 0.05 and β = 0.20, the
    2 Independent Samples with Repeated sample size nμ(r , ρ) can be obtained by multiplying
    Measurements n in equation (1) by the factor [1 + ρ (r – 1)] ∕ r , i.e.
    Extending the use of equation (1), suppose we
    wish to determine n for 2 independent samples [1 + ρ (r – 1)]
    nμ(r , ρ) = n • . (4)
    for which r repeated measurements will be made r
    on each subject over a period of time. Two analysis
    approaches can be considered: (A) a comparison of Inspection of (4) indicates that low values of the
    slopes of 2 linear trends; or (B) a comparison of time intra-class correlation ρ (the closer to 0 the better)
    averaged differences. Within- and between-subject will reduce n for detecting an effect of size δ.
    variance components are σw2 and σ b2, respectively, Spacing of measurement times is irrelevant.
    with total variance σ 2 = σw2 + σ b2 . It is assumed
    that each subject is measured at the same times Repeated measurement studies are likely to have
    tj = ( j = 1, ..., r ) and s t2 = Σ j (tj – t̄)2 ∕ r is the missing follow-up visit data due to various reasons
    within-subject variance of the tj. The only other (e.g. non-compliance). As a “hedge” against the
    critical piece of information is an estimate of the potential for a substantial loss of power, the
    intra-class correlation ρ = σ b2 ∕ σ 2 , the common proportion of patients anticipated to be lost
    correlation between measurements taken on the to follow-up should be taken into account in
    same subject, where 0 ≤ ρ ≤ 1. calculating sample size.

    In approach (A) comparing slopes, the sample size Summary


    will be labelled nβ (r , ρ). The standardised difference In calculating sample size for comparing means
    δ = ( β 1 – β 2 ) ∕ σ is a meaningful effect size relative of independent samples, a good starting point
    to the difference in slopes. For α = 0.05 and is simply 16 ∕ (effect size)2. In the absence of
    β = 0.20 the sample size nβ (r , ρ) can be obtained specific information about Δ or σ, effect sizes of
    by multiplying n in equation (1) by the factor δ = 0.30, 0.50 and 0.80 could be regarded as “small”,
    (1 – ρ) ∕ rs t2 , i.e. “medium” and “large”, respectively. In longitudinal
    studies, the intra-class correlation affects sample
    (1 – ρ) size calculation differently, depending on the
    nβ (r , ρ) = n • . (3)
    rs t2 analysis approach. In applying formulas (3) and (4),
    respectively, the ROT is as follows: for comparing
    Inspection of (3) indicates that high values of the slopes (within-subject average rate of change over
    intra-class correlation ρ (the closer to 1 the better) time), high values of the intra-class correlation (ρ
    and more repeated measurements with wider close to 1) are desirable as they will result in smaller
    spacing of measurement times will reduce n for sample sizes for detecting a given effect size, all
    detecting an effect of size δ. else constant; for comparing time-averaged mean

    Proceedings of Singapore Healthcare  Volume 20  Number 2  2011 139


    Statistics

    differences (averaged across subjects), low values of Reference:


    the intra-class correlation (ρ close to 0) are desirable 1. Van Belle G. Statistical Rules of Thumb. 1st ed. New York,
    NY: Wiley Interscience. 2002. 248 p.
    for minimising sample size. In the absence of any
    information at all about ρ, the recommendation
    is to try to choose a value of ρ that will err on the
    conservative side of whichever approach is being
    used, (A) or (B).

    140 Proceedings of Singapore Healthcare  Volume 20  Number 2  2011

    View publication stats

    You might also like