Statistical Thinking for Decision Makers Training Course

What Statistics Can Offer to Decision Makers for Government

• Descriptive Statistics
  • Basic Statistics - Identifying which statistics (e.g., median, average, percentiles) are most relevant to different data distributions
  • Graphs - The importance of accurate graphical representation and how it influences decision-making
  • Variable Types - Understanding which variables are easier to manage and analyze
  • Ceteris Paribus, Things Are Always in Motion - Recognizing that all factors are dynamic and can change over time
  • Third Variable Problem - Techniques for identifying the true influencing factor in complex data sets
• Inferential Statistics
  • Probability Value - Understanding the significance of the P-value in statistical analysis
  • Repeated Experiment - Interpreting results from repeated experiments to ensure reliability
  • Data Collection - Minimizing bias in data collection, while recognizing that it cannot be entirely eliminated
  • Understanding Confidence Level - The importance of confidence intervals in making informed decisions for government

Statistical Thinking for Government Decision Makers

• Decision Making with Limited Information
  • Determining when there is sufficient information to make a decision
  • Prioritizing goals based on probability and potential return (benefit/cost ratio, decision trees)
• How Errors Add Up
  • The Butterfly Effect - Small changes can have large impacts
  • Black Swans - Unpredictable events with significant consequences
  • Schrödinger's Cat and Newton's Apple in Government Operations - Analogies for understanding uncertainty and predictability
• Cassandra Problem - Measuring the Accuracy of Forecasts When Actions Change
  • Google Flu Trends - An example of how forecasting can go wrong
  • The Impact of Decisions on Forecast Relevance
• Forecasting Methods and Practicality for Government
  • ARIMA (AutoRegressive Integrated Moving Average) - A method for time series forecasting
  • Why Naive Forecasts Are Often More Responsive - The simplicity of naive models can sometimes lead to better predictions
  • How Far Should a Forecast Look into the Past? - Balancing historical data and current trends
  • Why More Data Can Sometimes Lead to Worse Forecasts - The potential pitfalls of overfitting with large datasets

Statistical Methods Useful for Government Decision Makers

• Describing Bivariate Data
  • Distinguishing Between Univariate and Bivariate Data - Understanding the differences in data analysis
• Probability
  • Why Things Differ Each Time We Measure Them? - Exploring variability and randomness in measurements
• Normal Distributions and Normally Distributed Errors - The importance of understanding normal distributions for accurate error analysis
• Estimation
  • Independent Sources of Information and Degrees of Freedom - Techniques for combining multiple sources of data effectively
• Logic of Hypothesis Testing
  • What Can Be Proven, and Why It Is Often the Opposite of What We Want (Falsification) - The principles of hypothesis testing in scientific inquiry
  • Interpreting the Results of Hypothesis Testing - Understanding how to draw meaningful conclusions from statistical tests
  • Testing Means - Methods for comparing means and assessing statistical significance
• Power
  • Determining a Good (and Cost-Effective) Sample Size - Strategies for optimizing sample size in research studies
  • False Positive and False Negative - The trade-offs between these errors and their implications for decision-making for government