RM lecture 10

Null Hypothesis Testing in Dutch and Belgian Groups

  • Objective: Apply testing of the null hypothesis to a study involving Dutchmen and Belgians.

Probability Distribution of Test Statistic

  • Concept: Determine the probability distribution of test statistics under the assumption that the null hypothesis is true.

    • Process: Collect data and compute the value of the test statistic.

Data Collection Example

  • Observed Value: Counted number of occurrences - value found to be 14 (exact context to be verified).

  • Comparative Analysis: Assessment based on the distribution of observed counts under the null hypothesis.

    • Key Question: Is the observed value in the rejection area of the distribution?

Important Factors in Testing

  • **Significant Factors Identified: **

    • Mean Differences: The perceived difference between the populations under study.

    • Sample Size: Number of participants from both groups (Dutchmen and Belgians).

    • Dispersion: Variability of observations relative to the mean; is measured through standard deviation.

Defining the Test Statistic

  • Definition: A variable that allows testing of the null hypothesis, typically using a specific formula.

    • Variants: Two primary types of t-tests, with focus on one variant today.

T-Test for Unpaired Observations
  • Application: Used when observations come from different groups (in this instance, Dutchmen and Belgians).

    • Focus: Testing whether there is a significant difference between the means of the two groups.

Calculation of T-Statistic
  • Numerator: Difference between group means (important for significance testing).

  • Denominator: Standard deviation of the groups - the smaller the spread, the more confidence in the difference being statistically significant.

    • Implication: As standard deviations (s<em>1s<em>1 and s</em>2s</em>2) decrease, resulting t-values become more extreme, leading to stronger evidence against the null hypothesis.

Impact of Sample Size
  • Larger Sample Size: Results in smaller sample variances contributing to more confident estimations of population parameters.

    • Mathematical Expression: The combined sample counts from both groups follows extsqrt(n<em>1+n</em>2)ext{sqrt}(n<em>1 + n</em>2).

Understanding Distribution of the Test Statistic

  • Distribution Characteristic: Approximately normal distribution when the null hypothesis is true; most t-values cluster around zero.

  • Extreme Values: Not impossible but rare to observe t-values far from zero; examples include t-values of negative 1 or negative 2.8 observed only in large samples.

  • Historical Insight: Nineteenth-century mathematicians deduced reliable statistics should be based on larger sample sizes to assure distributional accuracy.

T-Distribution

  • Family of Distributions: The shape of the t-distribution varies based on the degrees of freedom associated with the sample size.

    • Trend: As the sample size increases, distribution tails thin until they resemble the standard normal distribution when reaching infinity in sample size.

  • Critical Value for Two-Sided Testing: Reference border values at approximately ±1.96.

  • T5 Distribution Variable: Involves degrees of freedom; for instance, 18 in this context, impacting critical value determination.

Critical Values and Comparisons
  • Table Interpretation: Based on sample size affecting distribution; critical t-values are necessary to reject the null hypothesis effectively.

    • Example Calculation: Observed t-value of +/- 2.1 may differ significantly from reference distributions leading to wrong conclusions.

P-Value and Statistical Significance

  • Importance of P-Value: Indicates the probability of observing a more extreme value than the calculated t-value under the null hypothesis.

    • Example Numerical Result: Obtaining a p-value of 0.005 implies strong evidence against the null hypothesis when compared to significance level .

  • Convenience of Computational Tools: Modern technology aids in estimating critical values and p-values to simplify hypothesis testing.

Interpretation of Results

  • Mean Comparison: Highlighting mean response times, e.g., Belgians scoring 260 milliseconds based on averaged conditions (immediate vs. delayed recall).

  • Contextual Experiment Design: Participants randomly assigned to conditions to evaluate the effect of timing on recall abilities.

Ethical Consideration in Experimental Design

  • Historical Reflection: Significant changes in research practices over the past century, shaped by ethical discussions regarding experimentation methods.

  • Example Case Study: Stanford Prison Experiment showcases dramatic shifts in behavior based on environment; guards exhibited cruelty while subjects became submissive, reflecting the extent of situational influences on human behavior.