SOCIAL STATISTICS
Managing Data, Conducting Analyses, Presenting Results Fifth Edition
Numbers used in a dataset to represent categories
Codes
The process of taking a variable with a larger number of categories and recoding it into a new variable with a smaller number of categories.
Collapsing
An increasingly popular means of collecting data in which the researchers " survey" artifacts, media, or documents, asking each piece of data a set of questions
Content analysis
The most common measure of reliability of indexes, ranging from 0 (low reliability, these variables don’t fit together at all) to 1 (high reliability, these variables fit together well)
Cronbach’s alpha
In an explanatory relationship, it is the variable we are trying to explain something about
Dependent variable
The set of statistical procedures that allows us to describe for any variable where the center of the data is and how closely the cases surround this center
Descriptive statistics; also called summary statistics
A variable that has only two categories
Dichotomy
The set of statistical procedures that allows us to examine relationships among variables
Explanatory statistics
A method of dealing with missing data in which you take an educated guess regarding what the data point would be
Imputation
In an explanatory relationship, the variable we use to try to explain why another variable varies
Independent variable
The process of taking multiple variables and combining them into a single new variable
Indexing
A claim about a population based on sample data
Inference
The set of statistical procedures that allows us to speak about a population based on statistical results from a sample from that population
Inferential statistics
A level of measurement between ordinal and ratio, such as temperature, in which distance between units makes sense, but ratios do not make sense
Interval level
A basic characteristic of a variable that determines what statistical procedures can be used with that variable
Level of measurement
A type of survey question in which the respondent is given a statement and asked to place herself on an ordinary scale, typically strongly agree, agree, neither agree nor disagree, disagree, strongly disagree
Likert scale
A set of questions in a large dataset, usually about a similar topic
Module
The lowest and most restrictive level of measurement that simply puts cases into categories that cannot be ranked
Nominal level
A level of measurement that allows you to order the categories in a meaningful way, but doesn’t allow you to do anything else (such as multiply or divide)
Ordinal level
A synonym for the term dependent variable; a variable whose variation we are attempting to explain
Outcome variable
A synonym for the term independent variable; a variable that we are using to predict changes in the dependent variable
Predictor variable
The highest level of measurement that, because of a meaningful zero, allows us to multiply and divide values of the variable
Ratio level
The process of taking a variable’s original codes and giving each code a new code
Recoding
With regard to indexing, how well the variables in your index fit together
Reliability
A form of bias in which the respondent, knowing that he is being studied, will give the interviewer the answer that the respondent thinks the interviewer wants to hear
Social desirability bias
What a case is your study is, whether this be an individual (this is the case for most survey research), an organization, a city, or a country
Unit of analysis