STATISTICS FOR BEHAVIORAL SCIENCES Questions & Answers

The book emphasizes that a parameter is a value that describes a population, such as the population mean (μ) or standard deviation (σ). It is typically unknown and estimated. In contrast, a statistic is a value that describes a sample, such as the sample mean (M) or standard deviation (s). Statistics are calculated from observed data and are used to make inferences about unknown population parameters. This distinction is fundamental to understanding the goals of inferential statistics.

Variability is crucial because it measures the spread or dispersion of scores in a distribution. Gravetter and Wallnau explain that high variability makes it difficult to see clear patterns or effects, potentially obscuring a real treatment effect. Conversely, low variability allows for more precise and reliable conclusions. Understanding variability, often quantified by variance or standard deviation, is essential for interpreting descriptive statistics and for the accuracy of inferential tests, as it directly impacts standard error and test statistics.

The main purpose of hypothesis testing, as described by Gravetter and Wallnau, is to use sample data to make decisions or draw conclusions about a population. It's a formal procedure to determine whether a treatment or relationship observed in a sample is statistically significant enough to infer that it exists in the larger population, or if it's likely due to random chance. This involves setting up null and alternative hypotheses and evaluating the probability of the observed data under the null.

Gravetter and Wallnau explain that z-tests are used when the population standard deviation (σ) is known, which is rare in behavioral science research. T-tests, on the other hand, are used when the population standard deviation is unknown and must be estimated from the sample standard deviation (s). This estimation introduces more variability, leading to the use of the t-distribution, which is flatter and more spread out than the normal distribution, especially with smaller sample sizes, to account for the increased uncertainty.

The p-value, or probability value, is highly significant in hypothesis testing as it quantifies the probability of obtaining the observed sample data (or more extreme data) if the null hypothesis were true. Gravetter and Wallnau teach that if the p-value is less than or equal to the predetermined alpha level (e.g., .05), the result is considered statistically significant, leading to the rejection of the null hypothesis. A small p-value suggests that the observed effect is unlikely to be due to chance alone.

The standard error of the mean (σM or sM) quantifies the average distance between a sample mean (M) and the population mean (μ). Gravetter and Wallnau explain that it measures how much variability is expected among sample means if multiple samples were drawn from the same population. A smaller standard error indicates that sample means are more clustered around the population mean, leading to more precise estimates and greater power in hypothesis tests.

Gravetter and Wallnau explain that ANOVA is used when a researcher wants to compare means from three or more groups or conditions. Using multiple t-tests in such a scenario would inflate the Type I error rate (the probability of falsely rejecting a true null hypothesis) across the series of tests. ANOVA provides a single, omnibus test to determine if there are any significant differences among the group means, controlling the overall Type I error rate.

A correlation coefficient (r) tells us two main things about the relationship between two variables: the direction and the strength. Gravetter and Wallnau describe that the sign (+ or -) indicates the direction (positive: variables change in the same direction; negative: variables change in opposite directions). The magnitude (from 0 to 1) indicates the strength, with values closer to 1 (either positive or negative) representing a stronger, more consistent relationship. It does not imply causation.

Gravetter and Wallnau outline several assumptions for the Pearson correlation. The data should consist of pairs of numerical scores (interval or ratio scale). The relationship between the variables should be linear. The scores should be obtained from random sampling. Additionally, for hypothesis testing, the sampling distribution of r is assumed to be normal, which often implies that the underlying population distributions of the variables are normal, especially for smaller sample sizes.

Degrees of freedom (df) are crucial for determining the critical values and shape of sampling distributions (like t, F, or chi-square distributions) for various tests. Gravetter and Wallnau illustrate that df typically relate to the number of scores that are free to vary after certain parameters have been estimated. For example, in a one-sample t-test, df = n-1. In ANOVA, there are separate df for between-groups and within-groups variability, each reflecting the number of independent pieces of information used in their respective calculations.

Gravetter and Wallnau emphasize that statistical significance (indicated by a p-value) tells us whether an observed effect is likely real and not due to chance. Practical significance, often measured by effect size (e.g., Cohen's d, r-squared), tells us about the magnitude or importance of that effect. A statistically significant result might have a very small effect size, meaning it's real but not practically meaningful. Conversely, a large effect size might not be statistically significant if the sample size is too small.

Random sampling is crucial because it helps ensure that the sample is representative of the population, thereby minimizing sampling bias. Gravetter and Wallnau explain that when every individual in the population has an equal chance of being selected, it increases the likelihood that the sample's characteristics will mirror those of the population. This allows researchers to generalize findings from the sample to the population with greater confidence and validity, which is a cornerstone of inferential statistics.

Gravetter and Wallnau explain that when an ANOVA yields a significant F-ratio, it indicates that there is at least one significant difference among the group means, but it doesn't specify which specific groups differ. Post hoc tests (e.g., Tukey's HSD, Scheffé) are then used to conduct follow-up comparisons between all possible pairs of group means. These tests are designed to control the overall Type I error rate across multiple comparisons, preventing an inflated chance of false positives.

Power, as described by Gravetter and Wallnau, is the probability of correctly rejecting a false null hypothesis. In simpler terms, it's the probability of detecting a real effect or difference if one truly exists in the population. Researchers aim for high power (typically .80 or higher) to minimize the risk of a Type II error (failing to detect a real effect). Power is influenced by factors like sample size, effect size, and the alpha level, and is often considered during research design.

Gravetter and Wallnau define the independent variable (IV) as the variable that is manipulated or controlled by the researcher, or that naturally differentiates groups (e.g., gender, treatment condition). The dependent variable (DV) is the variable that is measured or observed to assess the effect of the independent variable. The researcher hypothesizes that changes in the IV will cause changes in the DV. This distinction is fundamental to experimental and quasi-experimental research designs.

The F-ratio in ANOVA is the test statistic used to determine if there are significant differences among two or more group means. Gravetter and Wallnau explain that it is calculated as the ratio of the variance between groups (treatment effect plus error) to the variance within groups (error only). A large F-ratio suggests that the differences between group means are substantially greater than what would be expected by chance, leading to the rejection of the null hypothesis.

Gravetter and Wallnau describe the chi-square test as a non-parametric test used for categorical data. It's primarily used in two main ways: the chi-square test for goodness of fit, which assesses whether observed frequencies for a single categorical variable differ significantly from expected frequencies; and the chi-square test for independence, which examines whether there is a significant relationship or association between two categorical variables in a population.

A confidence interval provides an estimated range of values that is likely to include an unknown population parameter (e.g., the population mean). Gravetter and Wallnau explain that it is constructed around a sample statistic (e.g., sample mean) and indicates the precision of the estimate. A 95% confidence interval, for example, means that if we were to take many samples and construct an interval for each, 95% of those intervals would contain the true population parameter.