One- and two-variable data analysis Knowledge-Based Study

Study notes

1. What is One-Variable Data?

   One-variable data, often called univariate data, involves measurements or observations on a single characteristic for each individual or item in a sample or population.

   The primary goal of univariate data analysis is to describe the distribution of that single variable by examining its center, spread, and overall shape.

   Key Representation Tools

   • Dot plots and histograms: Show the frequency distribution and shape (symmetric, skewed left, skewed right, or uniform).
   • Box plots: Display the five-number summary (minimum, first quartile, median, third quartile, and maximum), highlighting spread and potential outliers.

   Summarizing Univariate Data Numerically

   Numerical summaries allow us to quantify where data values cluster and how much they vary.

   Measures of Center

   • Mean: The arithmetic average, calculated as the sum of all values divided by the count: $\bar{x} = \frac{\sum x_i}{n}$. The mean is sensitive to extreme values or outliers.
   • Median: The middle value when data points are ordered from least to greatest. The median is resistant to outliers.

   Measures of Spread

   • Range: The difference between the maximum and minimum values.
   • Interquartile Range (IQR): $IQR = Q_3 - Q_1$, representing the spread of the middle 50% of values. Like the median, it is resistant to outliers.
   • Standard Deviation: Measures the typical distance of data points from the mean: $s = \sqrt{\frac{\sum (x_i - \bar{x})^2}{n - 1}}$.

   What is Two-Variable Data?

   Two-variable data, or bivariate data, involves pairs of numerical measurements collected from the same individual or subject, usually written as ordered pairs $(x, y)$.

   Unlike univariate analysis, which focuses on distribution, bivariate analysis investigates whether a relationship or association exists between the two variables.

   Explanatory and Response Variables

   • Explanatory variable (independent variable, $x$): The variable hypothesized to explain or predict changes in the other. It is placed on the horizontal axis.
   • Response variable (dependent variable, $y$): The outcome variable being measured or predicted. It is placed on the vertical axis.

   Visualizing and Modeling Relationships

   The primary graphical display for bivariate numerical data is the scatter plot.

   Describing an Association

   When analyzing a scatter plot, consider four essential characteristics:

   • Direction: Positive (as $x$ increases, $y$ tends to increase) or negative (as $x$ increases, $y$ tends to decrease).
   • Form: Linear (forming a roughly straight band) or non-linear (curved or clustered).
   • Strength: How closely the points follow the underlying pattern (weak, moderate, or strong).
   • Outliers or influential points: Individual observations that deviate markedly from the overall pattern.

   Linear Trend and Correlation

   For linear relationships, the linear correlation coefficient ($r$) quantifies direction and strength, ranging from $-1$ to $+1$. A line of best fit, $\hat{y} = a + bx$, can be used to model the trend and predict values of $y$ for given values of $x$.

   Summary of Key Differences

   • Purpose: Univariate analysis describes how a single variable is distributed; bivariate analysis determines how two variables co-vary or relate.
   • Visual displays: Univariate uses histograms, dot plots, and box plots; bivariate uses scatter plots and residual plots.
   • Numerical metrics: Univariate relies on mean, median, IQR, and standard deviation; bivariate relies on covariance, correlation coefficient ($r$), and slope/intercept of regression lines.

   Remember that observing a strong association between two variables in bivariate data indicates correlation, but it does not establish causation.