RSM 3/31/26

Overview of Regression Analysis

  • Regression analysis is a statistical method used to study the relationships between variables.

Key Concepts of Regression

  • Definition of Regression:
    • Regression quantifies relationships between two or more variables.
    • It can be used to describe how one variable (dependent) changes in response to another variable (independent).

Practical Applications in Hormone Analysis

  • Example: Radioimmunoassays used to analyze hormones, such as progesterone (P4).
    • Increasing concentration of hormone in a sample decreases the binding of radioactive hormones to antibodies, resulting in lower radioactivity readings.
    • Predicted Outcomes: More hormone leads to less radioactive hormone binding, thus influencing the gamma counts measured in the assay.

Regression in Sensitivity Improvement

  • Regression can improve the sensitivity of classical analyses by developing equations based on the relationship between variables.
    • Example: More color detection correlated with increased ammonia concentration measured by a spectrophotometer.

Covariation vs. Regression

  • Covariation: Describes how two variables change simultaneously but are independent of one another.
  • Regression: Demonstrates a cause-and-effect relationship where change in one variable (independent, X) causes a change in the second variable (dependent, Y).

Components of Regression Analysis

  • Cause and Effect Relationships:
    • The existence of a relationship must precede quantifying it through regression analysis.
    • The independent variable (X) is denoted as the source of variation, and the dependent variable (Y) is influenced by this change.
Regression Line
  • The regression line represents the best fit for the data points plotted in a graph.
    • Example 1: Ice cream sales vs. temperature.
    • Interpretations: Higher temperatures correlate with increased ice cream sales.
Regression Coefficient (Beta)
  • Denoted by β (beta):
    • Indicates the change in Y for a one-unit change in X:
    • The formula for regression is:
      Y=β0+β1XY = \beta_0 + \beta_1X
    • Interpretation: Y-intercept (β0\beta_0) is the predicted value of Y when X is 0.
Correlation vs. Regression
  • Correlation: Measures how two variables change at the same time without establishing a cause or effect relationship.
  • Example: A relationship between feed consumption (Y) and body weight (X) in chickens, where both may change without establishing direct causation.

Calculation of Regression Analysis

Example: Chicken Weight vs. Feed Consumption
  • Input Data: Measurements of feeding habits affected by body weight of chickens.
  • Cross product (X*Y) calculations determine how feed intake varies with weight differences.
Step-by-Step Calculation:
  1. Calculate the average: Sum of X divided by the count n (number of samples).
  2. Regression Equation:
    • extExpectedY=extMeanofY+bimes(X−extMeanofX)ext{Expected Y} = ext{Mean of Y} + b imes (X - ext{Mean of X})
  3. Slope Calculation: Derived from the sum of cross products and the variance in feed and weight of chickens.
  4. Fit data to find the slope (7.69) indicating feed consumption per weight increase.
Statistical Testing
  • Sum of Squares (SS): Used to measure the variation in Y that can be explained by the regression model.
  • Degrees of Freedom: The regression has one degree of freedom per factor (X).
Calculation Output:
  • F Statistic: Obtain from the ratio of regression mean square to error mean square:
    • F=extRegressionSSextErrorSSF = \frac{ ext{Regression SS}}{ ext{Error SS}}
  • P Values: Compare to significance level (e.g., 0.05) to determine statistical significance.
    • A significant P value (< 0.05) indicates that Y depends on X, confirming the regression model's validity.
    • Example: The significance of feed consumption related to weight changes in chickens.

Conclusion

  • Regression analysis effectively measures and predicts relationships between variables through quantifying how changes in one variable affect another. The analysis assumes a consistent biological relationship for valid and meaningful results across a variety of applications such as hormone measurement and agricultural studies.