Clever Grades

🎧 Read Aloud

Introduction to Statistical Testing and the Sign Test

Inferential testing is a fundamental aspect of psychological research and statistics, used to determine whether observed data provide enough evidence to support a hypothesis about a population. Rather than just describing data (descriptive statistics), inferential statistics allow us to make conclusions, predictions, or generalizations beyond the immediate data we have. This process involves several stages and key concepts, which are essential for understanding how to analyse research results correctly.

Statistical testing begins with a research hypothesis, typically formulated as a null hypothesis (H₀) and an alternative hypothesis (H₁). The null hypothesis states that there is no effect or difference in the population, while the alternative hypothesis suggests the presence of an effect or difference.

To test hypotheses, data from samples are analysed using inferential tests. One such test is the sign test, which is a simple non-parametric test used to evaluate differences between two related samples or conditions when data are nominal or ordinal and assumptions for other parametric tests are not met.

Sign Test Application & Use

1

When to Use the Sign Test

The sign test is appropriate when you have paired or matched data (e.g., before and after measurements on the same individuals) and want to determine if there is a difference in direction.
2

Why it's Robust

It is especially useful when the data do not meet assumptions of normal distribution or when data are at an ordinal level with no precise measurement of magnitude between pairs.

Calculating the Sign Test

✓

Determine Direction

For each pair, decide the direction of difference: positive (+) if the second measure is higher, negative (−) if the first measure is higher. Exclude pairs where there is no difference (ties).
✓

Find Test Statistic

Count the number of positive and negative signs. The test statistic is the smaller of the two counts.
✓

Compare to Critical Value

If the test statistic is less than or equal to the critical value, reject the null hypothesis, indicating a significant difference in direction between conditions.

Core Hypothesis and Error Terms

🚫

Null Hypothesis (H₀)

States that there is no effect or difference in the population.
✅

Alternative Hypothesis (H₁)

Suggests the presence of an effect or difference.
⚖️

Significance Level (α)

The probability threshold (usually 0.05) set before testing.
❓

P-value

The probability of obtaining the observed results, assuming no real effect.

Critical Values and Rejection

Statistical Tables

Hypothesis tests depend on comparing your test statistic against a critical value from statistical tables. The critical value marks the boundary of the rejection region. If your test statistic falls in the rejection region (beyond the critical value), you reject the null hypothesis.

Understanding Error Risks

⚠️
Type I Error (False Positive) Occurs when we incorrectly reject the null hypothesis when it is actually true, concluding there is an effect when there isn't. (Risk controlled by α).
🛑
Type II Error (False Negative) Happens when we fail to reject the null hypothesis even though it is false, missing a real effect. (Probability denoted by β).

The Error Trade-Off

🤔
Is it better to set α very low to minimize the chance of a Type I error?
🦉
Balancing these errors is important. Setting α too low reduces Type I errors but increases Type II errors, and vice versa.

Power of the Test

Power = 1 - β
Power (1 - β) of a test is the likelihood of correctly detecting a true effect, which is the inverse of the probability of making a Type II error (β).

Pro Tip: Matching Test to Data

💡

Level of Measurement: Choosing the correct inferential test depends heavily on whether data are nominal, ordinal, interval, or ratio. This choice determines if a parametric or non-parametric test is appropriate.

Factors Affecting Test Choice

Factor Description Test Requirement
Level of Measurement Nominal, Ordinal, Interval/Ratio
Experimental Design Related or Unrelated samples
Distribution of Data Parametric tests require Normal Distribution
General Rule Correctly identifying these guides the appropriate test.

Inferential Test Selection Guide

SUMMARY OF KEY POINTS FOR TEST SELECTION:

Test Data Type Design Purpose P/NP Distr. Needs
Sign Test Nominal (paired) Related samples Test direction of difference Non-parametric None
Wilcoxon Ordinal Related samples Test median difference Non-parametric None
Related t-test Interval/Ratio Related samples Test mean difference Parametric Normal distribution
Mann-Whitney U Ordinal Independent groups Test rank differences Non-parametric None
Unrelated t-test Interval/Ratio Independent groups Test mean difference Parametric Normal distribution
Spearman’s rho Ordinal/Non-normal I/R Any Correlation Non-parametric None
Pearson’s r Interval/Ratio Any Correlation Parametric Normal distribution
Chi-Squared Nominal Independent groups Association between categories Non-parametric Expected frequencies met
Introduction to Statistical Testing and the Sign Test
Term
Inferential Statistics

What is inferential statistics?

Answer
Definition

Techniques to make conclusions or generalizations beyond the sample data.

Term
Null Hypothesis (H₀)

What is the null hypothesis (H₀)?

Answer
Definition

The hypothesis stating no effect or difference exists in the population.

Term
Alternative Hypothesis (H₁)

What is the alternative hypothesis (H₁)?

Answer
Definition

The hypothesis suggesting there is an effect or difference in the population.

Term
Sign Test Usage

When is the sign test used?

Answer
Explanation

For paired or matched nominal/ordinal data to test difference in direction.

Term
Calculating Sign Test

How do you calculate the sign test?

Answer
Method

Count positive and negative differences, exclude ties, test statistic is smaller count.

Term
Rejecting Null in Sign Test

What does rejecting the null hypothesis in a sign test indicate?

Answer
Interpretation

A significant directional difference exists between paired conditions.

Term
Type I Error

What is a Type I error?

Answer
Definition

Rejecting a true null hypothesis (false positive).

Term
Type II Error

What is a Type II error?

Answer
Definition

Failing to reject a false null hypothesis (false negative).

Term
Significance Level (α)

What does a significance level (α) represent?

Answer
Definition

The probability threshold for rejecting the null hypothesis, commonly 0.05.

Term
Parametric Test Data

What type of data requires parametric tests?

Answer
Explanation

Interval or ratio data with normal distribution assumptions.

Term
Wilcoxon Signed-Rank Test

Which test compares two related samples considering difference magnitude?

Answer
Definition

Wilcoxon Signed-Rank Test.

Term
Difference Between Sign and Wilcoxon Tests

What is the main difference between the sign test and Wilcoxon test?

Answer
Comparison

Sign test only considers direction, Wilcoxon considers magnitude and ranks differences.

Term
Chi-Squared Test Use

What test is suitable for nominal categorical data association?

Answer
Definition

Chi-Squared test.

Term
Test Choice Factors

What factors affect the choice of statistical test?

Answer
Factors

Level of measurement, design (related or independent), and distribution of data.

Tip: Tap a card to flip. Shuffle or Reset anytime.

📊 Introduction to Statistical Testing and the Sign Test

1. What does the null hypothesis (H₀) state in inferential testing?

The null hypothesis assumes no effect or difference; it’s the default to be tested.

2. When is the sign test appropriate to use?

The sign test is used for paired data focusing on difference direction without normality assumptions.

3. What does a p-value ≤ 0.05 mean in hypothesis testing?

p ≤ 0.05 indicates less than 5% chance results are due to randomness; the result is significant.

4. Which error occurs when a true null hypothesis is wrongly rejected?

Type I error is a false positive—rejecting H₀ when it’s actually true.

5. How does the sign test treat the size of differences between pairs?

The sign test only assesses whether the difference is positive or negative regardless of size.

6. Which test should be used with independent groups and ordinal data?

Mann-Whitney U is used for comparing two independent groups with ordinal or non-normal data.

7. What determines the choice between parametric and non-parametric tests?

Parametric tests assume normality and interval/ratio data; non-parametric tests relax these assumptions.

📊 Results

  • Notes and Quiz
  • Psychology
  • AS & A Level
  • Logic and Philosophy
  • AQA