A Mini-Lesson in Statistics: What Causes Treatment Groups to be Deemed 'Not Statistically Different'?

Abstract

The theory behind a "not-statistically significant difference" (NS) in group comparison research is crucial for proper study design. Proprietary company-sponsored research whose purpose is to demonstrate null intervention effects is widespread. Therefore, a discussion of null effects in research results was undertaken.

Type I and Type II errors are explained, and reasons for "NS" results detailed. Clinical differences, power, statistical tests chosen, and values of a and D are included in the discussion. Examples from studies of the effects of artificial baby milk gift packs are offered. If the aim of research is to demonstrate NS between treatments, or if the conclusion of published research is NS, then the research must demonstrate a design which ensures ethical and appropriate levels of both Type I and Type II errors.

Keywords

power calculations statistical analysis Type II errors

Get full access to this article

View all access options for this article.

References

1. For your information. J Hum Lact 1994; 10:61-61.

2. Snell BJ , Krantz M , Keeton R , Delgado K , Peckman C : The association of formula samples given at hospital discharge with the early duration of breastfeeding. J Hum Lact 1992; 8:67-72.

3. Erratum. J Hum Lact 1992; 8:135-135.

4. Labbok M , Krasovec K : Towards consistency in breastfeeding definitions. Stud Fam Plann 1990; 21:226-230.

5. Moher D , Dulberg CS , Wells GA : Statistical power, sample size and their reporting in randomized controlled trials. J Am Med Assoc 1994; 272:122-124.

6. Howie PW , Forsyth JS , Ogston SA , Clark A , DuvFlorey C : Protective effect of breast feeding against infection. Br Med J 1990; 300: 11-16.

7. Hassard TH: What sample size will I need (Chapter 9); In Understanding Biostatistics. Toronto: Mosby Year Book, Inc., 1991; pp. 167-182.

8. Rubin DH , Leventhal JM , Krasilnikoff PA , Juo HS , Jekel JF , Weile B , et al: Relationship between infant feeding and infectious illness: a prospective study of infants during the first year of life. Pediatrics 1990; 85:464-471.

9. Hopkinson J : Breast-feeding and infectious illness, power, and confidence intervals (letter). Pediatrics 1991; 88:1079-1080.

10.

10. Mulford C : Infant feeding and infectious disease (letter). Pediatrics 1990; 86: 806-807.

11.

11. Frank DA , Wirtz SJ , Sorenson JR , Heeren T : Commercial discharge packs and breast-feeding counseling: effects on infantfeeding practices in a randomized trial. Pediatrics 1987; 80:845-854.

12.

12. Frank DA : Commercial discharge packs and breastfeeding counseling: summary of a study. J Hum Lact 1989; 5:7-10.

13.

13. Bergevin Y , Dougherty C , Kramer M : Do infant formula samples shorten the duration of breast feeding? Lancet 1983;1 (8334): 1148-1151.