Skip to main content

Table 1 Summary of identified statistical tests and interpretation criteria for validation of dietary intake assessment methods

From: Application and interpretation of multiple statistical tests to evaluate validity of dietary intake assessment methods

Statistical test Facet of validity reflected Interpretation criteria
   Good outcome Acceptable outcome Poor outcome
Correlation coefficient [8,14-16,18,28-67,90] Strength and direction of association at individual level [8] ≥0.50 [2] 0.20 - 0.49 [2] <0.20 [2]
Paired t-test/ Wilcoxon signed rank test [8,22,23,25,27,28,33,34,36,48,49,52-56,60,62,65,66,68,69,91] Agreement at group level [8] P > 0.05 [8]   P ≤ 0.05 [8]
Percent difference [8,22,23,25,27,28,33,34,49,52-56,60,65-72,91] Agreement at group level (size and direction of error) [8]   0.0 - 10.0% [77] >10%
Cross-classification (tertiles/ quartiles or quintiles) [8,22,31,32,35-38,41,42,44-51,55-61,63-69,91] Agreement (including chance), at individual level [8] ≥50% in same tertile/quartile [2] ≤10% in opposite tertile/quartile [2]   <50% in same tertile/quartile [2] >10% in opposite tertile/quartile [2]
• In same tertile
• In opposite tertile
Weighted Kappa statistics (coefficient) [8,24,26,30,40,43,54,58,59,63,64,66-69,91] Agreement (excluding chance) at individual level [8] ≥0.61 [2] 0.20 - 0.60 [2] <0.20 [2]
Bland Altman analysis: Correlation between mean and mean difference) [6,21,33,34,37-39,43,50,53,54,61,63,69,76,92] Presence, direction and extent of bias at group level [6,76] P > 0.05 [6]   P ≤ 0.05 [6]