D582 Statistics — Listen, Pause, Recall

Nominal Variables: Represent names and labels used for qualitative categorization without any inherent numerical or quantitative order.
Ordinal Variables: Represent categories characterized by a defined, meaningful order or rank, though the intervals between categories are not necessarily equal or measurable.
Discrete Variables: Numerical variables that consist of countable values, typically representing counts of distinct entities.
Continuous Variables: Numerical variables that result from measurements and can take on an infinite number of possible values within a given interval or range.
Interval Scale: Quantitative data possessing equal intervals between successive values, but lacking a true zero point (a value of zero does not represent the complete absence of the characteristic).
Ratio Scale: Quantitative data characterized by equal intervals between values and the presence of an absolute, true zero point (where zero represents the complete absence of the quantity).

Bar Chart: Used to compare discrete categories or qualitative groups.
Histogram: Used to display numerical data that has been grouped into contiguous intervals (bins).
Box Plot: Visualizes the distribution of a dataset by displaying quartiles, the median, and identifying outliers.
Scatterplot: Displays the relationship and pattern of association between two numerical variables plotted on Cartesian coordinates.
Frequency Polygon: Used to compare relative frequencies across multiple groups or distributions simultaneously.
Normal Distribution: Characterized graphically as a bell-shaped, perfectly symmetric curve.
Positively Skewed Distribution: A distribution whose tail extends toward the right (longer right tail, skewed toward higher values).
Negatively Skewed Distribution: A distribution whose tail extends toward the left (longer left tail, skewed toward lower values).
Bimodal Distribution: A distribution that exhibits two distinct local peaks (modes).

Systematic Sampling: A probability sampling technique where elements are selected from an ordered sampling frame at regular, fixed intervals, such as choosing every hundredth ( $100^{\text{th}}$ ) person.
Cluster Sampling: A method where the entire population is divided into groups (clusters), and entire groups are randomly selected in their entirety for observation.
Convenience Sampling: A non-probability sampling design based on selecting individuals who are easiest to reach or most readily available to the researcher.

Empirical Rule ( $68$ , $95$ , $99.7$ Rule):
- Exactly one standard deviation ( $\pm 1\sigma$ ) from the mean captures $68\%$ of the data.
- Exactly two standard deviations ( $\pm 2\sigma$ ) from the mean capture $95\%$ of the data.
- Exactly three standard deviations ( $\pm 3\sigma$ ) from the mean capture $99.7\%$ of the data.
- Foundational sequence to recall: $68$ , $95$ , $99.7$ .

Purpose: A confidence interval serves to estimate an unknown population parameter based on sample statistics.
Effect of Sample Size: To achieve a narrower confidence interval, increase the sample size ( $n$ ).
Effect of Confidence Level: Selecting a higher confidence level results in a wider confidence interval.

Decision Rules Based on P-Value and Alpha ( $\alpha$ ):
- When the p-value is less than alpha ( $p < \alpha$ ), reject the null hypothesis ( $H_0$ ).
- When the p-value is greater than alpha ( $p > \alpha$ ), fail to reject the null hypothesis ( $H_0$ ).
Interpretation Warning: Failing to reject the null hypothesis ( $H_0$ ) does not mean the null hypothesis has been proven true; it merely indicates insufficient evidence against it.

Independent Two-Sample T-Test: Used when evaluating two independent groups measured on a numerical outcome variable.
Determining Test Directionality via Alternative Hypothesis ( $H_a$ ):
- If the alternative hypothesis states "greater than" ( $>$ ), conduct a right-tailed test.
- If the alternative hypothesis states "less than" ( $<$ ), conduct a left-tailed test.
- If the alternative hypothesis states "not equal to" ( $\neq$ ), conduct a two-tailed test.

Application: Used when comparing the means of three ( $3$ ) or more groups.
Decision Criteria:
- If the calculated $F$ -statistic is greater than the critical $F$ -value ( $F > F_{\text{critical}}$ ), reject the null hypothesis ( $H_0$ ).
- If the calculated $F$ -statistic is less than the critical $F$ -value ( $F < F_{\text{critical}}$ ), fail to reject the null hypothesis ( $H_0$ ).
Interpretation of Significant Findings:
- A statistically significant ANOVA indicates that at least one population mean differs from the others.
- An ANOVA test alone does not identify which specific pairs of group means differ; further post-hoc analysis is required.

Correlation Coefficient ( $r$ ):
- A value of positive one ( $+1$ ) indicates a strong positive linear relationship.
- A value of negative one ( $-1$ ) indicates a strong negative linear relationship.
- Primary causal caveat: Correlation does not prove causation.
Regression Slope:
- The slope defines the predicted change in the dependent variable ( $y$ ) for every one-unit ( $1$ ) increase in the independent variable ( $x$ ).
- Positive slope: As $x$ increases, $y$ increases.
- Negative slope: As $x$ increases, $y$ decreases.

Chi-Square Goodness-of-Fit Test: Compares one single categorical variable against expected counts or theoretical distributions.
- Degrees of Freedom ( $\text{df}$ ): Calculated as the number of categories minus one: $\text{df} = k - 1$
Chi-Square Test of Homogeneity: Used to compare the distributions of a categorical variable among two or more distinct populations.
Chi-Square Test of Independence: Used to test for an association or relationship between two categorical variables collected from a single population.
Decision Rules:
- If the p-value is less than alpha ( $p < \alpha$ ), reject the null hypothesis ( $H_0$ ).
- If the p-value is greater than alpha ( $p > \alpha$ ), fail to reject the null hypothesis ( $H_0$ ).

Four-Step Analytical Sequence:
1. Identify the test: Determine the appropriate statistical procedure based on variables, sample count, and research design.
2. Find the hook word: Locate key directional or relational indicator terms in the problem statement.
3. Compare the numbers: Compare the p-value directly to alpha ( $\alpha$ ), or compare the test statistic ( $F$ ) to the critical value ( $F_{\text{critical}}$ ).
4. Choose the conclusion: State the formal statistical decision (reject or fail to reject the null hypothesis).
Directional Keyword Warning: Slow down and pay close attention to the specific phrases "less than", "greater than", and "not equal to", as these specific words dictate whether the test is left-tailed, right-tailed, or two-tailed.