Comprehensive Study Notes on Statistical Methods and Analysis

Fundamental Definitions in Statistics

• Skewness: A measure of the asymmetry of the probability distribution of a real-valued random variable about its mean. It indicates whether data is concentrated on one side of the mean more than the other.

• Kurtosis: A statistical measure used to describe the degree to which scores cluster in the tails or the peak of a frequency distribution. It indicates the "peakedness" or flatness of the distribution.

• Normal Curve: Also known as the Gaussian distribution, it is a bell-shaped, symmetric frequency distribution where the mean, median, and mode are all equal and located at the center.

• Inferential Statistics: Techniques that allow us to use study samples to make generalizations that apply to the broader population from which the sample was drawn. It involves hypothesis testing and estimating parameters.

• Descriptive Statistics: Brief descriptive coefficients that summarize a given data set, which can be either a representation of the entire population or a sample of a population. This includes measures of central tendency (mean, median, mode) and measures of variability (range, standard deviation).

• Chance: The possibility of a particular outcome occurring due to random variation rather than a systematic cause.

• Statistic: A characteristic or measure obtained by using data from a sample (e.g., sample mean, denoted as $\bar{x}$).

• Parameter: A characteristic or measure that describes an entire population (e.g., population mean, denoted as $\mu$).

• Standard Error: An estimate of the standard deviation of a sampling distribution. It indicates how much a sample statistic (like the mean) is expected to vary from the actual population parameter ($SE = \frac{\sigma}{\sqrt{n}}$).

• p-value: The probability that the observed results (or more extreme results) occurred by chance, assuming the null hypothesis is true. A lower p-value indicates stronger evidence against the null hypothesis.

• Sample: A subset of individuals, items, or data points selected from a larger population to represent the group as a whole.

• Population: The entire set of individuals or objects of interest that a researcher targets for study.

• Sampling Error: The difference between a sample statistic and the actual population parameter it is intended to estimate, caused by observing a sample instead of the entire population.

• Hypothesis: A specific, testable prediction about the relationship between two or more variables.

• Null Hypothesis ($H_0$): A statement that there is no relationship or no difference between variables. Researchers usually aim to reject this hypothesis.

• Research Hypothesis ($H_1$ or $H_a$): Also called the alternative hypothesis, it is the statement that a relationship or difference exists between variables.

• Statistical Significance: A determination that a relationship between variables or a difference between groups is unlikely to be due to chance. This is typically achieved when the p-value is less than the significance level (e.g., $p < 0.05$).

• Confidence Level: The percentage of times that the confidence interval would include the true population parameter if the study were repeated many times (e.g., $95\%$).

• Confidence Interval: A range of values, derived from sample data, that is likely to contain the value of an unknown population parameter.

• Significance Level (Alpha, $\alpha$): The threshold set by the researcher (e.g., $0.05$ or $0.01$) for deciding whether a result is statistically significant.

• Type I Error: A "false positive" error that occurs when the researcher rejects a true null hypothesis (stating there is an effect when there is not).

• Type II Error: A "false negative" error that occurs when the researcher fails to reject a false null hypothesis (stating there is no effect when there actually is one).

• One-tailed Test: A directional hypothesis test used when the researcher predicts the specific direction of an effect (e.g., "Group A will score higher than Group B").

• Two-tailed Test: A non-directional hypothesis test used when the researcher predicts a difference exists but does not specify the direction (e.g., "There will be a difference between Group A and Group B").

• Statistical Conclusion Validity: The degree to which conclusions about the relationship among variables based on the data are correct and reasonable.

• Correlation: A statistical measure that describes the size and direction of a relationship between two variables.

• Independent Samples: Data collected from two or more groups that are not related to each other (e.g., a group of men vs. a group of women).

• Paired Samples: Also known as dependent samples; data collected from the same subjects at different times (pre-test/post-test) or from closely matched pairs.

• Theoretical Sampling Distribution: A distribution of all possible values of a statistic (e.g., the mean) that would occur if an infinite number of samples of the same size were drawn from the same population.

• Central Limit Theorem: A fundamental principle stating that as the sample size ($n$) increases, the sampling distribution of the mean will approach a normal distribution, regardless of the shape of the population distribution.

Skewness and Kurtosis Analysis

• Determining Kurtosis Visually:

  • A Leptokurtic distribution is very peaked with thick tails.

  • A Platykurtic distribution is relatively flat with thin tails.

  • A Mesokurtic distribution is the normal distribution.

• Relationship Between Mean, Median, and Skewness:

  • Symmetric (Normal) Distribution: $Mean = Median = Mode$.

  • Positive Skew (Right-skewed): The tail extends to the right (higher values). Usually, $Mean > Median$.

  • Negative Skew (Left-skewed): The tail extends to the left (lower values). Usually, $Mean < Median$.

• Impact on the Shape of the Curve:

  • Skewness shifts the peak and stretches one tail of the curve.

  • Kurtosis changes the height of the peak and the thickness of the tails without necessarily shifting the center.

• Thresholds for Normality:

  • A skewness or kurtosis value of $0$ represents perfectly normal data.

  • Values between $-1$ and $+1$ are generally considered "normally distributed" for most social science applications.

  • Values exceeding $\pm 1$ indicate significant non-normality.

• Negative vs. Positive Skewness:

  • Positive Skew: Most data points are clustered at the lower end, with a few extreme high values dragging the mean to the right.

  • Negative Skew: Most data points are clustered at the higher end, with a few extreme low values dragging the mean to the left.

Probability

• Probability of Independent Events: Two events are independent if the occurrence of one does not affect the probability of the other. The probability of both occurring is calculated by multiplying their individual probabilities: $P(A \text{ and } B) = P(A) \times P(B)$.

• Probability of Dependent Events: If the occurrence of event A changes the probability of event B, the events are dependent. It is calculated as: $P(A \text{ and } B) = P(A) \times P(B|A)$, where $P(B|A)$ is the probability of B given that A has occurred.

• Probability of Two Outcomes Both Occurring: This is the "joint probability." For independent events, multiply the probabilities. For mutually exclusive events (where only one can happen), the joint probability of both happening simultaneously is $0$.

Theoretical Sampling Distribution and the Central Limit Theorem

• Relationship Between Sample Statistics and Population Parameters: Sample statistics (e.g., $\bar{x}$) are used as "point estimates" to guess the unknown population parameters (e.g., $\mu$). Due to sampling error, these are rarely identical.

• Central Limit Theorem (CLT): This theorem justifies using the normal distribution for inferential statistics because it guarantees that as sample size ($n$) becomes large (typically $n \ge 30$), the distribution of sample means will be normal even if the underlying population is skewed.

• Sample Error and Sample Size: There is an inverse relationship. As the sample size ($n$) increases, the sampling error decreases. Larger samples provide a more precise estimate of the population.

• Standard Error (SE):

  • Definition: It is essentially the standard deviation of the sampling distribution.

  • Calculation: $SE = \frac{s}{\sqrt{n}}$, where $s$ is the sample standard deviation and $n$ is the sample size.

• Utility of the Sampling Distribution: It allows us to determine the probability that a specific sample mean would be obtained from a population. This is the foundation for calculating p-values and confidence intervals.

Hypothesis Testing Logic and Errors

• Null ($H_0$) vs. Research ($H_1$) Hypothesis:

  • $H_0$: "The new therapy has no effect on depression scores."

  • $H_1$: "The new therapy reduces depression scores."

• Decision Rules:

  • Reject the Null: When the calculated p-value is less than or equal to the significance level ($p \le \alpha$). This means the result is statistically significant.

  • Fail to Reject the Null: When the p-value is greater than the significance level ($p > \alpha$). There is not enough evidence to support the research hypothesis.

• Type I and Type II Errors:

  • Type I (Alpha Error): Rejecting $H_0$ when it is true. This is often considered more serious for social workers because it may lead to the adoption of an ineffective (and potentially costly or harmful) intervention.

  • Type II (Beta Error): Failing to reject $H_0$ when it is false. This means an effective intervention is overlooked.

• One vs. Two-Tailed Tests:

  • One-Tailed (Directional): Used when the theory specifically predicts an increase or a decrease. More powerful but riskier if the effect is in the opposite direction.

  • Two-Tailed (Non-directional): Used when the researcher tests for any difference (increase or decrease). It is the standard conservative approach in research.

Analysis of t-Tests

• Purpose: The t-test allows researchers to compare the means of two groups (or a sample mean to a population mean) to determine if the difference is statistically significant.

• Types of t-Tests:

  • One-sample t-test: Compares the mean of a single sample to a known or hypothesized population mean (e.g., comparing a school's average GPA to the national average).

  • Two-sample (Independent) t-test: Compares the means of two unrelated groups (e.g., comparing the test scores of students who used a tutor vs. those who did not).

  • Paired Samples (Dependent) t-test: Compares the means of two related groups or the same group at two different times (e.g., measuring patient anxiety levels before and after a clinical session).

Reading and Selecting Statistical Tests

• Reading Research Articles:

  • Abstracts provide the research question and context for hypotheses.

  • Complex tables require identifying the Mean ($\bar{x}$), Standard Deviation ($SD$), t-score ($t$), and significance ($p$) to determine if findings support the author's claims.

• Test Selection Based on Variables:

  • Chi-square: Used when both the independent and dependent variables are categorical/nominal (e.g., Gender and Vote Choice).

  • Pearson Correlation: Used when measuring the linear relationship between two continuous/interval-ratio variables (e.g., Age and Annual Income).

  • Independent Samples t-test: Used when the independent variable is categorical with exactly two groups and the dependent variable is continuous (e.g., Treatment vs. Control group compared on a 100-point health scale).

  • One-sample t-test: Used when comparing a continuous sample variable to a set constant population value.

  • Paired Samples t-test: Used for pre-test/post-test designs or matched pairs with continuous data.

SPSS Procedures and Interpretation

• Skewness and Kurtosis: Found in the Descriptives or Frequencies menu. Interpret values to check for normality thresholds ($\pm 1$).

• Histograms: Used to visually inspect the distribution of data and the presence of outliers or skew.

• Standard Error Calculation: Often listed in the Descriptives output as "S.E. Mean."

• Confidence Intervals: Can be calculated during t-test procedures or through the "Explore" function in SPSS. Always interpret as the range where the true population mean likely falls.

• Pearson Correlation: Conducted via Analyze -> Correlate -> Bivariate. Interpretation includes looking at the Pearson Correlation Coefficient ($r$) for strength and the p-value for significance.

• Conducting t-tests:

  • Choose based on design (One-sample, Independent, or Paired).

  • Interpret the t-statistic ($t$), degrees of freedom ($df$), and the significance ($Sig. 2-tailed$) value.