The Scientific Study of Nutrition

Pellagra

Disease characterized by dermatitis, diarrhea, and dementia, which was discovered to be caused by niacin deficiency (Vit B3) through the scientific method

Components of a good scientific experiment

-Falsifiable Hypothesis

-Quantifiable Data

-Appropriate Experimental Population

-Proper Experimental Controls

-Reasonable analysis and interpretation

Falsifiable Hypothesis in a Scientific Experiment

Hypothesis should be falsifiable so that evidence can be gathered in support or against hypothesis.

Ideally, you should be trying to prove against every way your hypothesis can be wrong

Quantifiable Data in a Scientific Experiment

Quantitative metrics should be measured reliably and repeatedly

Qualitative metrics should be quantified through standardized tools

Appropriate Experimental Population in Scientific Experiment

Experimental population should be relevant and large enough for data from experiment to be applicable

Proper Experimental Controls in Scientific Experiment

A control group and/or a placebo group should be used in an experiment so that the experimental group can be compared with baseline

Reasonable Analysis and Interpretation in Scientific Experiment

There should be a clearly defined criteria for the hypothesis to be proven true/false. The correct statistical approaches should be used. The applicability of the research to a broader context should be appropriate without being "overhyped". Other explanations for the result from the experiment should be ruled out

Observational Studies

Large populational studies that observe correlations between variables (diet, health, disease).

Cannot determine causation without experimental studies being conducted

Cross-Sectional Study

A study design that analyzes a population at a specific point in time

Cohort Study

A study design that analyzes a population without the outcome and follows the population over time to see if they develop the outcome

Case-Control Study

A study design that compares a population with or without the outcome to determine differences between the two groups that may be related to the outcome

Experimental Studies

Studies in which a variable is directly manipulated (treatment) to determine a casual relationship.

Strongest Experimental Study Design

Double-blind randomized control trial

Subjects are assigned randomly to a experimental and control group with both the participants and the research being unaware of which group each participant is in

Reliable Nutritional Information Sources

-Information from government entities/organizations (Health Canada)

-Professional associations associated with experts

-Widely adopted textbooks

-Peer-reviewed research paper from high quality journals

Systematic Reviews

A systematic synthesis of research findings about a clinical question, providing a reproducible answer.

Systematic reviews analyze multiple studies, reducing impact of confounding variables and abnormal fluke effect sizes

Data Dredging

The inappropriate (sometimes deliberately so) use of data mining to uncover any relationships in data without background theory or hypotheses. Increases the likelihood of false associations being made (from random chance).

Common Misleading Method of Presenting Data

-Truncated axes

-Misleading use of log scales

-Inappropriate/mixed units

-Fabricated trends

-In proper use of pie charts, 3D graphs, and pictograms

Truncated Axes

Showing only a part of the axis can exaggerate changes or make differences seem more pronounced than they are.

P-value

Represents the mathematical statistical significance of data.

P value

R^2-value

Represents how close the data is to the trend line

Greater R^2 value = more closer to trend line = stronger association