Skip to main content

Table 4 Statistical test outcomes and interpretation for energy and nutrient intakes derived from the test data set 1

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

Nutrient

Spearman correlation (r value)

Wilcoxon signed rank test (P value)

Percentage difference (%)

Cross-classification (Tertiles)

Weighted Kappa statistics (value)

Bland – Altman 2,3 Spearman Correlation (r value)

 

Association (strength & direction)

Agreement

Agreement (size & direction of error)

Agreement (including chance)

Agreement (excluding chance)

Presence, direction and extent of bias

 

% in same tertile

% in opposite tertile

Level of validation

Individual

Group

Group

Individual

Individual

Individual

Group

Energy (kJ)

0.26

P > 0.05

−9.8

46.8

19.2

0.20

P < 0.001

Validly interpretation

Acceptable

Good

Good

Poor

Poor

Acceptable

Biased

Protein (g)

0.23

P < 0.01

−19.1

42.6

23.4

0.12

P < 0.05

Validly interpretation

Acceptable

Poor

Acceptable

Poor

Poor

Poor

Biased

Fat (g)

0.01

P > 0.05

−6.9

34.0

31.9

−0.01

P > 0.05

Validly interpretation

Poor

Good

Good

Poor

Poor

Poor

Not biased

Carbohydrates (g)

0.40

P > 0.05

−1.4

50.0

17.4

0.25

P < 0.01

Validly interpretation

Acceptable

Good

Good

Good

Poor

Acceptable

Biased

Folate (mcg)

0.40

P < 0.05

33.0

53.2

8.5

0.30

P < 0.01

Validly interpretation

Acceptable

Poor

Poor

Good

Good

Acceptable

Biased

Vitamin A (mcg)

0.15

P < 0.01

−22.9

34.0

14.9

0.03

P > 0.05

Validly interpretation

Poor

Poor

Poor

Poor

Poor

Poor

Not biased

Iron (mg)

0.38

P > 0.05

−24.8

51.1

23.4

0.29

P < 0.01

Validly interpretation

Acceptable

Good

Poor

Good

Poor

Acceptable

Biased

  1. 1Energy and nutrient intake derived from a QFFQ and four repeated 24-hour recalls conducted in 18–65 year old adults (n = 47, 11 males & 36 females) in a rural area in the Eastern Cape.
  2. 2% in LOA: Energy: 93.6%; protein: 95.7%, fat: 97.9%, carbohydrate: 95.7%; folate: 95.7%, vitamin A: 89.4%; iron: 98%.
  3. 3Upper & lower limits of agreement (LOA) and DRI for females aged 18 to 55: energy (kJ): −13406 & 10694 (EER: 10093 kJ); protein (g): −94.6 & 59.4 (RDA: 46 g); fat: −125.4 & 96.6 (no DRI); folate (mcg): −649 & 925.8 (RDA: 400mcg); vitamin A (mcg): −876 &576.4 (RDA: 700mcg); iron (mg): −22,4 &10.4 (RDA: ????mg).
  4. Interpretation criteria for statistical tests.
  5. Wilcoxon signed rank test: Good: p > 0.05; Poor: ≤ 0.05 [9].
  6. Percentage difference: Good: 0.0 – 10.9%; Acceptable: 11.0 – 20.0%; Poor: > 20.0% [92].
  7. Correlations coefficient (Spearman): Good: ≥ 0.50; Acceptable: 0.20 – 0.49; Poor < 0.20[2]
  8. Cross-classification (Tertiles) (% in same tertile): Good: ≥ 50%; Poor: < 50% [2].
  9. Cross-classification (% in opposite tertile): Good: ≤ 10%, Poor: > 10% [2].
  10. Weighted Kappa statistics: Good: ≥ 0.61; Acceptable: 0.20 – 0.59; Poor: < 0.20 [2].
  11. Bland-Altman - Correlation coefficient (Spearman): Good: P > 0.05; Poor: P ≤ 0.05 [2].