Quality of Life Research

, Volume 15, Issue 3, pp 349–356 | Cite as

Application of Robust Statistical Methods for Sensitivity Analysis of Health-Related Quality of Life Outcomes

  • Jennifer L. Beaumont
  • Lisa M. Lix
  • Kathleen J. Yost
  • Elizabeth A. Hahn


Background: Researchers often use conventional parametric procedures to test hypotheses of health-related quality of life (HRQL) mean equality across patient groups. However, these techniques are sensitive to the presence of skewed distributions and unequal group variances, which may characterize many HRQL measures. Purpose: To conduct a sensitivity analysis of conventional and robust approaches to test hypotheses of mean equality on HRQL measures for hematopoietic stem cell transplantation survivors and a healthy comparison group. Methods: The methods applied were the conventional parametric procedure of least-squares analysis of variance applied to the raw scores, the conventional parametric procedure applied to transformed data, and a robust approximate degrees of freedom parametric procedure utilizing trimmed means and Winsorized variances. Results: The choice of analysis method affected the conclusions about the null hypothesis of mean equality. More commonly observed, however, was a substantial difference in the value of the F-statistic and standard errors which was particularly evident in the measures with greater degrees of skewness and heterogeneity of variances. Conclusions: Robust statistical tests should be incorporated into sensitivity analyses when analyzing HRQL data.


Assumption violations Robust statistical tests Trimmed means 



analysis of variance


approximate degrees of freedom


Center for Epidemiologic Studies Depression Scale


Duke-UNC Social Support


effect size


healthy comparison


Functional Assessment of Chronic Illness Therapy-Fatigue


Functional Assessment of Chronic Illness Therapy-Spiritual Well-Being


Functional Assessment of Cancer Therapy-Physical Well-Being


health-related quality of life


hematopoietic stem cell transplantation


least squares


Medical Outcomes Study-Family Functioning


Medical Outcomes Study-Sleep Problems


Perceived Health Questionnaire


Posttraumatic Growth Inventory


standard deviation


standard error


Medical Outcomes Study 36-item Short Form


SF-36 General Health Perceptions


SF-36 Mental Health Index


SF-36 Physical Functioning


SF-36 Social Functioning


Sickness Impact Profile-Alertness Behavior


State-Trait Anxiety Inventory-Trait Anxiety


UCLA Loneliness


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Julious, SA, George, S, Machin, D, Stephens, RJ 1997Sample sizes for randomized trials measuring quality of life in cancer patientsQual Life Res6109117CrossRefPubMedGoogle Scholar
  2. 2.
    Rose, MS, Koshman, ML, Spreng, S, Sheldon, R 1999Statistical issues encountered in the comparison of health-related quality of life in diseased patients to published general population norms: Problems and solutionsJ Clin Epidemiol52405412CrossRefPubMedGoogle Scholar
  3. 3.
    Ramsey, SD, Berry, K, Moinpour, C, Giedzinska, A, Andersen, MR 2002Quality of life in long term survivors of colorectal cancerAm J Gastroenterol9712281234PubMedGoogle Scholar
  4. 4.
    Lix, LM, Keselman, JC, Keselman, HJ 1996Consequences of assumption violations revisited: A quantitative review of alternatives to the one-way analysis of variance F testRev Educ Res66579619Google Scholar
  5. 5.
    Rasmussen, JL 1986An evaluation of parametric and non-parametric tests on modified and non-modified dataBrit J Math Stat Psychol39213220Google Scholar
  6. 6.
    Blair, RC, Sawilowsky, SS, Higgins, JJ 1987Limitations of the rank transform statistic in tests for interactionsCommun Stat Simul Comput1611331145Google Scholar
  7. 7.
    Sawilowsky, SS, Blair, RC, Higgins, JJ 1989An investigation of the type I error and power properties of the rank transform procedure in factorial ANOVAJ Educ Stat14255267Google Scholar
  8. 8.
    Wilcox, RR 1995ANOVA: The practical importance of heteroscedastic methods, using trimmed means versus means, and designing simulation studiesBrit J Math Stat Psychol4899114Google Scholar
  9. 9.
    Keselman, HJ, Wilcox, RR, Lix, LM 2003A generally robust approach to hypothesis testing in independent and correlated groups designsPsychophysiology40586596CrossRefPubMedGoogle Scholar
  10. 10.
    Mudholkar, GS, Srivastava, DK 2000A class of robust stepwise alternatives to Hotelling’s T2 testsJ Appl Stat27299619Google Scholar
  11. 11.
    Reed, JF,III 2003Solutions to the Behrens–Fisher problemComput Meth Progr Biomed70259263CrossRefGoogle Scholar
  12. 12.
    Yuen, KK 1974The two-sample trimmed t for unequal population variancesBiometrika61165170Google Scholar
  13. 13.
    Lix, LM, Keselman, HJ 1998To trim or not to trim: Tests of location equality under heteroscedasticity and nonnormalityEduc Psychol Meas58409429Google Scholar
  14. 14.
    Andrykowski, MA, Bishop, M, Hahn, EA,  et al. 2005Long-term health-related quality of life, growth, and spiritual well-being after hematopoietic stem-cell transplantationJ Clin Oncol23599608PubMedGoogle Scholar
  15. 15.
    SAS Institute Inc2002SAS/STAT, Version 8.2SAS Institute, Inc.Cary, NCGoogle Scholar
  16. 16.
    SAS Institute Inc2002SAS/IML User’s Guide, Version 8.2SAS Institute, Inc.Cary, NCGoogle Scholar
  17. 17.
    Westfall, PH, Young, SS 1993Resampling-Based Multiple TestingWileyNew YorkGoogle Scholar
  18. 18.
    Wilcox, RR, Keselman, HJ 2001Using trimmed means to compare k measures corresponding to k independent groupsMultivar Behav Res36421444Google Scholar
  19. 19.
    Box, GEP 1954Theorems on quadratic forms applied in the study of analysis of variance problems, I. Effect of inequality of variance in the one way classificationAnn Math Stat25290302Google Scholar
  20. 20.
    Pesarin, F 2001Multivariate Permutation TestsWileyNew YorkGoogle Scholar

Copyright information

© Springer 2006

Authors and Affiliations

  • Jennifer L. Beaumont
    • 1
  • Lisa M. Lix
    • 2
  • Kathleen J. Yost
    • 1
    • 3
  • Elizabeth A. Hahn
    • 1
    • 3
  1. 1.Center on Outcomes, Research and Education (CORE)Evanston Northwestern HealthcareEvanstonUSA
  2. 2.Department of Community Health SciencesUniversity of ManitobaWinnipegCanada
  3. 3.Department of Preventive Medicine, Feinberg School of MedicineNorthwestern UniversityChicago

Personalised recommendations