2016
DOI: 10.1080/13645579.2016.1155379
|View full text |Cite
|
Sign up to set email alerts
|

Normal enough? Tools to aid decision making

Abstract: When undertaking quantitative hypothesis testing, social researchers need to decide whether the data with which they are working is suitable for parametric analyses to be used. When considering the relevant assumptions they can examine graphs and summary statistics but the decision making process is subjective and must also take into account the robustness of the proposed tests to deviations from the assumptions. We review the contemporary advice on this issue available to researchers and look back to the root… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
4
0

Year Published

2017
2017
2021
2021

Publication Types

Select...
6

Relationship

1
5

Authors

Journals

citations
Cited by 6 publications
(4 citation statements)
references
References 32 publications
(76 reference statements)
0
4
0
Order By: Relevance
“…We first examined the data for deviations from normality and variance of the distribution. Unless there were substantial deviations with regard to both distributional characteristics, we used parametric analytic techniques because, for example, ANOVA is robust to violations of its assumptions (Spencer et al, 2017 ).…”
Section: Methodsmentioning
confidence: 99%
“…We first examined the data for deviations from normality and variance of the distribution. Unless there were substantial deviations with regard to both distributional characteristics, we used parametric analytic techniques because, for example, ANOVA is robust to violations of its assumptions (Spencer et al, 2017 ).…”
Section: Methodsmentioning
confidence: 99%
“…The implication of such small samples is their reduced statistical power to find small but significant effects (Loken and Gelman, 2017). Additionally, small samples make it difficult to ensure randomization across all demographic characteristics or possible control variables, and to parse the study population for statistical analysis across multiple variables (Spencer et al, 2017). Nonetheless, significant effects such as those we have demonstrated with a small sample indicate that the treatment effect is likely larger than the equivalent result with a larger sample (Friston, 2012).…”
Section: Discussionmentioning
confidence: 85%
“…There is a suggestion of some clustering at around 0.18 g (180 mg). Multivariate analysis 23 was performed to ascertain if this correlated with initial ROSE assessment (insufficient, benign or malignant—note that three passes with ROSE assessments of equivocal/benign or suggestive of malignancy were excluded as they formed groups too small for analysis), needle type and gauge (standard or Procore, 22 or 25) and pass number.…”
Section: Resultsmentioning
confidence: 99%