Statistics in Psychology

Statistics is the backbone of psychological research. It provides the tools and techniques necessary to collect, organize, analyze, interpret, and present data. Without statistics, psychologists would struggle to draw meaningful conclusions from their experiments, surveys, and observations. Understanding statistical concepts is crucial for both conducting research and critically evaluating the work of others in the field.

In psychology, statistics are broadly categorized into two main types: descriptive statistics and inferential statistics. Descriptive statistics help us summarize and describe the main features of a dataset, while inferential statistics allow us to make generalizations about a larger population based on a sample of data. This section will cover key concepts within both categories, focusing on measures of central tendency, dispersion, the normal curve, and various parametric and non-parametric tests.

1. Measures of Central Tendency

Measures of central tendency are statistical values that represent the center or typical value of a dataset. They give us a single number that summarizes the data. The most common measures of central tendency are the mean, median, and mode.

a. Mean

The mean, often referred to as the average, is calculated by summing all the scores in a dataset and then dividing by the number of scores. It is sensitive to extreme values (outliers).

Formula:

Mean (x̄) = Σx / n

Where: Σx is the sum of all scores, and n is the number of scores.

Example: If a student scores 80, 90, 70, 85, and 95 on five tests, the mean score is (80 + 90 + 70 + 85 + 95) / 5 = 420 / 5 = 84.

b. Median

The median is the middle score in a dataset that has been ordered from least to greatest. If there is an even number of scores, the median is the average of the two middle scores. The median is less affected by outliers than the mean.

Example 1 (Odd number of scores): For the scores 70, 80, 85, 90, 95, the median is 85.

Example 2 (Even number of scores): For the scores 70, 80, 85, 90, 95, 100, the two middle scores are 85 and 90. The median is (85 + 90) / 2 = 87.5.

c. Mode

The mode is the score that appears most frequently in a dataset. A dataset can have one mode (unimodal), two modes (bimodal), or more than two modes (multimodal). If all scores appear with the same frequency, there is no mode. The mode is useful for categorical data.

Example: In the dataset 1, 2, 2, 3, 4, 4, 4, 5, the mode is 4 because it appears most often.

2. Measures of Dispersion

Measures of dispersion, also known as measures of variability, describe how spread out or clustered together the scores in a dataset are. They tell us about the variability of the data. Key measures include the range, variance, and standard deviation.

a. Range

The range is the simplest measure of dispersion. It is the difference between the highest and lowest scores in a dataset. Like the mean, it is sensitive to outliers.

Formula:

Range = Highest Score - Lowest Score

Example: For the scores 70, 80, 85, 90, 95, 100, the range is 100 - 70 = 30.

b. Variance

Variance is the average of the squared differences from the mean. It measures how far each number in the set is from the mean. Squaring the differences ensures that all values are positive and gives more weight to larger deviations.

Formula for Sample Variance (s²):

s² = Σ(x - x̄)² / (n - 1)

Formula for Population Variance (σ²):

σ² = Σ(x - μ)² / N

Where: x is each score, x̄ is the sample mean, μ is the population mean, n is the sample size, N is the population size. We use n-1 in the sample variance formula as a correction (Bessel's correction) to provide a less biased estimate of the population variance.

c. Standard Deviation

The standard deviation is the square root of the variance. It is one of the most widely used measures of dispersion because it is expressed in the same units as the original data, making it easier to interpret than variance. A small standard deviation indicates that the scores are clustered around the mean, while a large standard deviation indicates that the scores are spread out over a wider range.

Formula for Sample Standard Deviation (s):

s = √s² = √[Σ(x - x̄)² / (n - 1)]

Formula for Population Standard Deviation (σ):

σ = √σ² = √[Σ(x - μ)² / N]

Example: If the variance of a dataset is 25, the standard deviation is √25 = 5.

3. The Normal Curve (Normal Distribution)

The normal curve, also known as the Gaussian distribution or bell curve, is a fundamental concept in statistics. It is a symmetrical, bell-shaped probability distribution where most of the data points cluster around the mean, and the frequency gradually decreases as you move further away from the mean in either direction.

Characteristics of the Normal Curve:

• It is symmetrical around the mean.
• The mean, median, and mode are all equal and located at the center of the distribution.
• The total area under the curve is equal to 1 (or 100%).
• The curve is asymptotic to the horizontal axis, meaning it approaches the axis but never touches it.
• Specific percentages of data fall within certain standard deviations from the mean.

The Empirical Rule (68-95-99.7 Rule):

For a normal distribution:

• Approximately 68% of the data falls within one standard deviation (±1σ) of the mean.
• Approximately 95% of the data falls within two standard deviations (±2σ) of the mean.
• Approximately 99.7% of the data falls within three standard deviations (±3σ) of the mean.

Example: If a psychology test has a mean score of 70 and a standard deviation of 10, then about 68% of test-takers scored between 60 (70-10) and 80 (70+10). About 95% scored between 50 (70-2*10) and 90 (70+2*10).

The normal curve is important because many psychological variables (e.g., intelligence, reaction time) tend to follow a normal distribution. It also forms the basis for many inferential statistical tests.

4. Parametric vs. Non-Parametric Tests

Statistical tests are broadly divided into parametric and non-parametric types. The choice between them depends on the nature of the data and the assumptions that can be made about the population from which the data were drawn.

a. Parametric Tests

Parametric tests make specific assumptions about the population distribution, most commonly that the data are normally distributed. They also typically assume that the data are measured on an interval or ratio scale and that the variances of the groups being compared are roughly equal. Parametric tests are generally more powerful than non-parametric tests if their assumptions are met.

Assumptions:

• Data are normally distributed.
• Data are measured on an interval or ratio scale.
• Homogeneity of variance (variances of groups are approximately equal).
• Independence of observations.

b. Non-Parametric Tests

Non-parametric tests, also known as distribution-free tests, do not rely on assumptions about the population distribution. They are often used when the data are nominal or ordinal, or when the assumptions of parametric tests (like normality) are violated. They are generally less powerful than parametric tests.

When to Use:

• Data are nominal or ordinal.
• Sample size is small, and normality cannot be assumed.
• Outliers significantly distort parametric measures.

5. Parametric Tests: t-test

The t-test is a widely used parametric inferential statistical test to determine if there is a significant difference between the means of two groups. It is used when the population standard deviation is unknown and the sample size is relatively small (typically less than 30).

Types of t-tests:

a. One-Sample t-test

This test compares the mean of a single sample to a known population mean or a hypothesized value.

Formula:

t = (x̄ - μ) / (s / √n)

Where: x̄ is the sample mean, μ is the population mean, s is the sample standard deviation, and n is the sample size.

b. Independent Samples t-test (Two-Sample t-test)

This test compares the means of two independent groups to determine if there is a statistically significant difference between them. For example, comparing the test scores of a control group and an experimental group.

Formula (assuming equal variances):

t = (x̄₁ - x̄₂) / s_p √[(1/n₁) + (1/n₂)]

Where: x̄₁ and x̄₂ are the means of the two groups, n₁ and n₂ are the sample sizes, and s_p is the pooled standard deviation.

c. Paired-Samples t-test (Dependent Samples t-test)

This test compares the means of the same group at two different times (e.g., pre-test vs. post-test scores) or under two different conditions. It essentially compares the means of two related samples.

Formula:

t = M_D / (s_D / √n)

Where: M_D is the mean of the differences between paired scores, s_D is the standard deviation of the differences, and n is the number of pairs.

6. Non-Parametric Tests

Non-parametric tests are used when the assumptions for parametric tests are not met. They are often based on ranks rather than the raw scores themselves.

a. Sign Test

The sign test is one of the simplest non-parametric tests. It is used to compare two related samples (like a paired-samples t-test) or to test a hypothesis about a single median. It works by assigning a '+' sign for scores where the first value is greater than the second, a '-' sign for scores where the second is greater than the first, and ignoring any ties. It focuses on the direction of the difference.

Use Case: Comparing pre-test and post-test scores when the assumption of normality is violated.

Example: If we have 10 pairs of scores and 7 show an increase (positive sign), 2 show a decrease (negative sign), and 1 is a tie, we look at the number of positive and negative signs.

b. Wilcoxon Signed-Rank Test

This test is a more powerful alternative to the sign test for related samples. It compares two related samples (like the paired t-test) but considers the magnitude of the differences as well as their direction. It is used when the data are ordinal or interval/ratio but not normally distributed.

Procedure:

1. Calculate the difference between each pair of scores.
2. Ignore pairs with a difference of zero.
3. Rank the absolute values of the non-zero differences.
4. Assign the original signs (positive or negative) back to the ranks.
5. Calculate the sum of the positive ranks (W+) and the sum of the negative ranks (W-).
6. The test statistic is the smaller of W+ or W-.

Use Case: Comparing two related measurements where the differences are not normally distributed.

c. Mann-Whitney U Test (Wilcoxon Rank-Sum Test)

This is the non-parametric equivalent of the independent samples t-test. It is used to compare two independent groups when the dependent variable is measured on an ordinal scale or when the interval/ratio data are not normally distributed.

Procedure:

1. Combine all scores from both groups and rank them from lowest to highest.
2. Sum the ranks for each group separately (R1 and R2).
3. Calculate the U statistic for each group using the sums of ranks and sample sizes.
4. Compare the calculated U value to a critical U value from a table to determine significance.

Use Case: Comparing the effectiveness of two different teaching methods on student performance when scores are not normally distributed.

d. Kruskal-Wallis H Test

This is the non-parametric equivalent of a one-way ANOVA (Analysis of Variance). It is used to compare three or more independent groups when the dependent variable is measured on an ordinal scale or when the interval/ratio data are not normally distributed.

Procedure:

1. Combine all scores from all groups and rank them from lowest to highest.
2. Sum the ranks for each group separately.
3. Calculate the H statistic using the sums of ranks and sample sizes for each group.
4. Compare the H statistic to a chi-square distribution to determine significance.

Use Case: Comparing the levels of job satisfaction across three different departments in a company, where satisfaction is measured on a Likert scale.

e. Friedman Test

This is the non-parametric equivalent of a repeated-measures ANOVA. It is used to compare three or more related samples (e.g., measurements taken at three or more time points on the same individuals) when the dependent variable is measured on an ordinal scale or interval/ratio data are not normally distributed.

Procedure:

1. Rank the scores within each subject (block) across the different conditions or time points.
2. Sum the ranks for each condition across all subjects.
3. Calculate the Friedman test statistic (often denoted as χ²_r or Q).
4. Compare the test statistic to a chi-square distribution to determine significance.

Use Case: Comparing the effectiveness of three different therapeutic interventions administered to the same group of patients over time.

7. Power Analysis

Statistical power is the probability of correctly rejecting a false null hypothesis. In simpler terms, it is the probability of finding a statistically significant effect when one truly exists in the population. A power analysis is conducted to determine the sample size needed to detect an effect of a certain size with a given level of confidence (typically 80% or 90% power). It can also be used to determine the power of a study given a specific sample size and effect size.

Key Components of Power Analysis:

• Alpha (α) level: The probability of a Type I error (rejecting a true null hypothesis). Typically set at 0.05.
• Statistical Power (1-β): The probability of correctly rejecting a false null hypothesis. Typically set at 0.80 or 0.90.
• Effect Size: The magnitude of the difference or relationship in the population. This is often the most challenging component to estimate.
• Sample Size (N): The number of participants in the study.

Power analysis is crucial for research design to ensure that a study has a sufficient chance of detecting a meaningful effect if it exists, thus avoiding wasted resources and inconclusive results.

8. Effect Size

Effect size is a quantitative measure of the magnitude of a phenomenon. While statistical significance (p-value) tells us whether an effect is likely due to chance, effect size tells us how large or important that effect is. It is independent of sample size.

Common Measures of Effect Size:

• Cohen's d: Used for differences between two means. It represents the difference between the means in standard deviation units.
  • Small effect: d = 0.2
  • Medium effect: d = 0.5
  • Large effect: d = 0.8
• Pearson's r (correlation coefficient): Measures the strength and direction of the linear relationship between two continuous variables.
  • Small effect: r = 0.1
  • Medium effect: r = 0.3
  • Large effect: r = 0.5
• Omega-squared (ω²) or Eta-squared (η²): Used in ANOVA to represent the proportion of variance in the dependent variable that is accounted for by the independent variable(s).

Interpreting effect size requires context. A "large" effect in one field might be considered "medium" or "small" in another. However, it provides a standardized way to compare the magnitude of findings across different studies.

Summary Table of Non-Parametric Tests

Test Name	Parametric Equivalent	Type of Data	Number of Groups	Related/Independent
Sign Test	Paired t-test	Ordinal	2	Related
Wilcoxon Signed-Rank Test	Paired t-test	Ordinal/Interval	2	Related
Mann-Whitney U Test	Independent t-test	Ordinal/Interval	2	Independent
Kruskal-Wallis H Test	One-way ANOVA	Ordinal/Interval	3+	Independent
Friedman Test	Repeated Measures ANOVA	Ordinal/Interval	3+	Related