Types of Correlations and Quality of Evidence

Correlations

Overview of Correlations

  • Date of Discussion: Tuesday, February 17, 2026

  • Time of Discussion: 4:31 PM

Types of Correlations

Linear Correlation

  • Definition: Expresses the association or relationship between variables in the form of a straight line.

  • Bivariate Emphasis: Focuses on correlating 2 variables.

  • Level of Measurement: Different types of correlations depend on the levels of measurement.

Pearson Correlation

  • Type: Numeric to Numeric Correlation.

  • Definition: Refers to correlations measured at the interval or ratio level of measurement, consistent with the discussion from the previous session.

  • Zero Order Correlation: Involves correlating 2 variables directly.

Multivariate Linear Correlation

  • Definition: Correlates 3 or more variables simultaneously.

Non-Linear Correlation

  • Definition: Expresses the relationship between variables in a non-straight line.

  • Note: There are many types of non-linear correlations, but these are typically not covered at the undergraduate level, thus skipped in the discussion.

Pearson Correlation Explained

  • Insight: Indicates what percent of what was expected was actually observed.

  • Description of Pearson's r: Summarizes the proportion of the expected average linear association between variables X and Y as reflected in the empirical data collected from the sample.

    • Computation Formula:
      Pearson’s r=Empirical Covariance(S<em>xS</em>y)Theoretical Expectation\text{Pearson's } r = \frac{\text{Empirical Covariance} (S<em>x S</em>y)}{\text{Theoretical Expectation}}

  • Conceptual Framework: Keeps Pearson's r as a summary based on sample observations rather than as an inherent property of people or constructs.

Covariance in Pearson's R

  • Definition: The Pearson R is a ratio between the covariance that you empirically observed and the expected covariance.

Relational Evidence Quality Assessment

  • Main focus: Will be assessed on the differences in quality of evidence.

  • Elements to Assess:

    • Validity: Checks if the results obtained are due to chance.

    • Sensitivity (Detection Capability): Measures the probability that a tool can detect significant results; evaluates how effective it is at detection.

    • Credibility (Diagnostic Value): Indicates the percentage of the results that are false positives versus true positives.

    • Precision: Relates to the accuracy of measurement and results obtained.

Conclusion

  • Note: The discussion emphasizes the importance of understanding the nuances and implications of correlations in research and data analysis, including considerations of validity, sensitivity, credibility, and precision in drawing conclusions from empirical data.