TY - JOUR
T1 - Statistical and nonstatistical significance
T2 - Implications for health care researchers
AU - Baghi, Heibatollah
AU - Noorbaloochi, Siamak
AU - Moore, Jean B.
PY - 2007/4
Y1 - 2007/4
N2 - Quality improvement professionals have to decide whether a change has led to improvement. This is typically done through testing the statistical significance of the findings. In this article, we explore controversies surrounding statistical significance testing with attention to contemporary criticism of bad practice resulting from the misuse of statistical significance testing. Most statistical significance tests use tests (eg, F, χ) with known distributions with the P values used as the main evidence to evaluate whether tests are statistically significant. The primary conclusion of this article is that the P value alone as a measure of statistical significance does not give sufficient information about testing of hypotheses. When it is coupled with other measures, however, such as the point estimation of the effect size and the use of a confidence interval around it, the combination of these statistics can provide a more thorough explanation of statistical testing. This article offers recommendations for process improvement investigators as to when to appropriately apply and not to apply statistical significance testing.
AB - Quality improvement professionals have to decide whether a change has led to improvement. This is typically done through testing the statistical significance of the findings. In this article, we explore controversies surrounding statistical significance testing with attention to contemporary criticism of bad practice resulting from the misuse of statistical significance testing. Most statistical significance tests use tests (eg, F, χ) with known distributions with the P values used as the main evidence to evaluate whether tests are statistically significant. The primary conclusion of this article is that the P value alone as a measure of statistical significance does not give sufficient information about testing of hypotheses. When it is coupled with other measures, however, such as the point estimation of the effect size and the use of a confidence interval around it, the combination of these statistics can provide a more thorough explanation of statistical testing. This article offers recommendations for process improvement investigators as to when to appropriately apply and not to apply statistical significance testing.
KW - Effect size
KW - P value
KW - Quality improvement
KW - Statistical significance
KW - Statistical testing
UR - http://www.scopus.com/inward/record.url?scp=34248326457&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=34248326457&partnerID=8YFLogxK
U2 - 10.1097/01.QMH.0000267447.55500.57
DO - 10.1097/01.QMH.0000267447.55500.57
M3 - Review article
C2 - 17426608
AN - SCOPUS:34248326457
SN - 1063-8628
VL - 16
SP - 104
EP - 112
JO - Quality management in health care
JF - Quality management in health care
IS - 2
ER -