How to interpret bar diagrams and graphs in the composite index of anthropometric failure and behavioral feeding problems when the p-value is not significant?

Interpreting Non-Significant Results in CIAF and Behavioral Feeding Problem Studies

Understanding Non-Significant P-Values

When bar diagrams and graphs show a p-value that is not significant (p > 0.05) in studies examining the composite index of anthropometric failure (CIAF) and behavioral feeding problems, this means you cannot conclude that a true difference or association exists between the groups being compared—only that no statistically significant difference was detected in your sample. 1, 2

Critical Interpretation Principles

• A non-significant p-value (p > 0.05) indicates that the observed differences could reasonably have occurred by chance alone, with more than a 5% probability 2
• You cannot conclude that groups are equivalent or similar based solely on p > 0.05; this would require a study specifically designed to test equivalence 1
• The absence of statistical significance does not prove the absence of an effect—it may reflect insufficient sample size, high variability, or a true lack of difference 1

Examining the Bar Diagrams and Graphs

Focus on Effect Sizes and Descriptive Statistics

• Always report and examine the actual magnitude of differences (mean differences, percentages) alongside the p-value, as the clinical or practical importance may differ from statistical significance 1
• Look at the mean values and standard deviations displayed in the graphs to assess the actual size of differences between groups, regardless of statistical significance 1
• Standard deviations indicate the variability within each group—large standard deviations relative to the mean suggest high variability, which makes detecting true differences more difficult 1

Evaluating CIAF Categories

• CIAF disaggregates children into seven categories (no failure, wasting only, wasting and underweight, wasting/underweight/stunting, underweight and stunting, stunting only, underweight only), providing more precision than conventional single indices 3, 4, 5
• When comparing CIAF prevalence between groups (e.g., children with vs. without behavioral feeding problems), examine the distribution across all seven categories in your bar diagrams, not just overall prevalence 4, 5
• Even with non-significant overall p-values, specific CIAF subcategories (such as multiple anthropometric failures) may show clinically meaningful patterns that warrant attention 4, 6

Interpreting Behavioral Feeding Problem Associations

When P-Values Are Not Significant

• If the association between behavioral feeding problems and CIAF shows p > 0.05, you can only state that "no statistically significant association was observed in this sample" 1
• Do not conclude that behavioral feeding problems have no impact on nutritional status; the study may have been underpowered to detect the association 1
• Consider whether the sample size was adequate—studies with fewer than 50 participants per group have limited power to detect true differences 1

Examining Confidence Intervals

• Confidence intervals provide crucial information about the precision and range of plausible values for the true effect, even when p-values are not significant 1, 2
• Wide confidence intervals that cross the null value (e.g., odds ratio CI includes 1.0) indicate imprecision and suggest the study cannot rule out either beneficial or harmful associations 2
• If confidence intervals are not reported in your graphs, this represents a limitation in the presentation of results 1, 2

Common Pitfalls to Avoid

Misinterpretation Errors

• Never use the term "trend" to describe a p-value close to but above 0.05 (e.g., p = 0.06); a result is either statistically significant at your pre-specified threshold or it is not 1
• Do not conclude that behavioral feeding problems and CIAF are "unrelated" or "independent" based solely on p > 0.05 1
• Avoid stating that groups are "similar" or "the same" when p > 0.05—only equivalence studies can make such claims 1

Contextual Considerations

• CIAF typically reveals higher overall malnutrition prevalence (40-64%) compared to conventional single indices, as it captures children with multiple anthropometric failures 3, 4, 5, 6
• Behavioral feeding problems may be associated with specific CIAF subcategories (e.g., wasting patterns) rather than overall CIAF, requiring examination of disaggregated data 4
• Consider that maternal factors (height, education), household wealth, and child age are consistently strong predictors of CIAF that may confound associations with feeding behaviors 3, 6

Practical Reporting Approach

What to State in Results

• Report the exact p-value (e.g., p = 0.12) rather than just "p > 0.05" or "not significant" 1
• Present mean CIAF prevalence with standard deviations for each group being compared 1
• Include the sample size for each group in your bar diagram labels 1
• If available, report confidence intervals for the difference between groups 1, 2

What to State in Discussion

• Acknowledge that the lack of statistical significance may reflect true absence of association, insufficient sample size, or high variability 1
• Discuss whether the observed effect sizes (even if not statistically significant) have clinical or public health relevance 1
• Consider whether unmeasured confounders (such as socioeconomic factors, maternal nutrition, or environmental sanitation) might obscure true associations between feeding behaviors and anthropometric outcomes 1, 3, 6