1-DV Data Type Cheat Sheet

This cheat sheet outlines statistical tests for analyzing one dependent variable (1-DV) based on the measurement scale of the DV and the number of independent variable (IV) groups. It distinguishes between tests for differences and associations.

Core Principles

• Understand your data type: Nominal, Ordinal, or Scale (Interval/Ratio).
• Identify your research question: Are you looking for differences between groups or associations between variables?
• Determine the number of groups/samples for your independent variable.
• Parametric tests are used for Scale data (Interval/Ratio) and assume normal distribution.
• Nonparametric tests are used for Ordinal or Nominal data, or when parametric assumptions are violated.
• Independent samples involve separate groups; related samples involve the same group measured twice or matched pairs.

Action Steps

• 1. Identify the measurement scale of your Dependent Variable (DV).
• 2. Determine if you are comparing means/distributions (differences) or examining relationships (associations).
• 3. Count the number of groups or samples involved in your Independent Variable (IV).
• 4. Select the appropriate statistical test based on steps 1-3 and the test's assumptions.

Key Terms

• 1-DV: One Dependent Variable; the outcome variable being measured.
• IV: Independent Variable; the variable manipulated or used to group participants.
• Parametric Test: Statistical test assuming data follows a specific distribution (e.g., normal distribution).
• Nonparametric Test: Statistical test that does not assume a specific data distribution.
• Independent Samples: Data from separate, unrelated groups of participants.
• Related Samples: Data from the same group measured multiple times or from matched pairs.
• Scale Data: Interval or Ratio data, allowing for meaningful arithmetic operations.
• Ordinal Data: Data ranked in order, but intervals between ranks are not necessarily equal.
• Nominal Data: Categorical data with no inherent order (e.g., gender, color).

Pro Tips

• When in doubt with ordinal data, use nonparametric tests.
• Partial correlation is useful for controlling confounding variables.
• Chi-square is specifically for nominal data associations.
• Mann-Whitney U and Wilcoxon Signed Rank are nonparametric alternatives to t-tests.

Pitfalls to Avoid

• Using parametric tests on ordinal or nominal data without justification.
• Confusing independent and related samples designs.
• Applying tests for differences when the research question is about association.
• Ignoring the assumptions of parametric tests (e.g., normality, homogeneity of variance).

Myth vs Reality

• Nonparametric tests are always less powerful than parametric tests.: Nonparametric tests can be more powerful when the assumptions of parametric tests are severely violated, especially with small sample sizes or skewed data.
• You can only use t-tests for comparing two groups.: T-tests are specifically for comparing the means of *two* groups. For more than two groups, ANOVA is typically used.

Real World Examples

• Comparing test scores between students who used study app A vs. study app B.: Independent-samples t-test (if scores are Scale data).
• Assessing if a new teaching method improves student performance over the old method.: Paired-samples t-test (if same students tested before/after) or Independent-samples t-test (if different groups).
• Investigating the relationship between hours studied and exam grade.: Pearson's bivariate correlation (if hours and grade are Scale data).
• Determining if there's a link between political party affiliation (Nominal) and voting preference (Nominal).: Chi-square test of contingencies.