Data Analysis: Statistics for Item Interactions
Purpose To provide a broad overview of statistical analyses appropriate for exploring interactions and relationships between items measured in survey research To examine conditions when each is appropriate Qualifier: Generally, such analyses will call for software programs more sophisticated than Survey Monkey
Outline Cross Tabulations Analysis of Variance Paired t-Tests Correlation Analysis Regression Analysis
Note All analytical methods are based on some criterion for evaluating statistical significance Statistical significance implies that the chance that the observed event would really occur in the population is good
Cross-Tabulations Examines relationships between two categorical variables No causality needs to exist
Cross-Tabulations Advantages: Flexible Robust Easy Limitations: needs adequately-sized cells to be effective; not as sensitive as other measures of association
Cross-Tabulations Statistical significance is measured via a chi-square statistic which evaluates the difference between the ‘expected’ value and ‘actual’ value of research outcomes
Analysis of Variance (ANOVA) Compares variable means between distinct groups and determines if the difference between the group means are significantly different Calculated F-ratio If so, inferences can be made about group differences that may exist in the population
ANOVA Useful when examining the relationship between a categorical variable (independent variable) and a continuous variable (dependent variable) ANOVA is appropriate when make comparisons between two or more independent groups Example: Does gender affect attitudes toward immigration? Categorical: Men v. Women Continuous: Attitude toward immigration
Paired t-Tests Paired t-test analysis is used when means are compared on two variables from the same respondent t-value focuses on the difference between scores Example: Do women’s attitudes toward immigration influence their political party affiliation?
Correlation Analysis Tests the degree and significance of the relationship between two continuous variables from interval or ratio scales “shows how much the two variables move together” (p. 324) Not necessary to establish causality
Correlation Analysis Question Is one’s “intent to change their consumer behavior” related to the extent to which they think “climate change is an important issue”?
Correlation Analysis Statistical significance determined by The sign (+/-) of the correlation The absolute value of the correlation Pearson Product Moment and Spearman rank correlation statistics are most commonly used
Regression Analysis Used to measure the relationship between an independent and dependent variable in terms of The degree and direction of influence The significance of the relationship The predictive formula that establishes the nature of the relationship (see Berry, et al, p.9)