UNITS OF ANALYSIS
The level of social life on which social scientists focus. Sometimes referred to as: individuals, cases, observations, etc.
Examples of units of analysis include: people (most common), organizations, schools, industries, families, states, etc…
Can anyone think of others?
VARIABLES
A property of a unit of analysis that can have more than one value and/or the sets of attributes which describe characteristics of a unit/units of analysis.
Independent and/or Dependent Variables
An independent variable is a "causal" or "predictor" variable in a hypothesis (see below).
A dependent variable is an "outcome" variable or the variable one is trying to predict and/or explain in a hypothesis (see below).
being able to determine the difference between independent and dependent variables is very, very important for this class
MECE
The categories of variables must be both: “mutually exclusive” and "collectively exhaustive."
"Mutually exclusive" means the variables cannot fit into more than one category
“Collectively Exhaustive” means there should be a category for each case, even if it is “none of the above” or “other”
HYPOTHESIS
A hypothesis is a testable statement about the relationship between two (or more) variables. Many hypotheses are bivariate, meaning they involve 2 variables only.
IF X, THEN Y
In this example, the unit of analysis is people. The independent variable is amount of time spent sleeping the night before a test and the dependent variable is the test score.
One hypothesis might be: If a person sleeps only 3 hours the night before a test (X), then their test score (Y) will decrease.
The independent variable is the "causal" or "predictor" variable. In this class, the symbol of the independent variable is X.
The dependent variable is the "outcome" variable or the variable one is trying to predict and/or explain. In this class, the symbol of the dependent variable is Y.
CAUSALITY
There are three criteria needed to make an argument for causality between two variables.
- Correlation or Covariation. Changes in the independent variable must be associated with changes in the dependent variable.
- Time order: Cause must precede the effect.
- Non-spuriousness: Relationship cannot be explained by other factors.
What are some examples of a hypothesis predicting a causal relationship with sociological variables?
“I expect a _________ relationship between education level and annual income.”
“I expect a positive relationship between education level and annual income.”
In this case, education (x), is predicted to causally effect income (y).
Is it possible to expect a positive relationship between annual income and education?
"I expect a _________ relationship between the annual income of one's parents and the child's education."
"I expect a positive relationship between the annual income of one's parents and the child's education."
Time order can be hard to determine. In cases where either variable can be the independent or dependent variable, the decision of which variable is which is determined by what the research is trying to study. If the researcher is trying to predict education (y), then income may be an independent variable (x). If the researcher is trying to explain income (y), then education will be an independent variable (x).
Nonspuriousness can also be hard to determine. Spuriousness happens when two variables are correlated, meaning they trend together, but neither variable causes the other. Instead the correlation between the two variables that is due to variation, or cause of, a third variable. When this third variable, an extraneous variable, causes variation in 2 variables, it is said to have created a spurious relationship.
Both ice cream sales and sunburns increase in the summer. Because of this correlation (both variables, ice cream sales and sunburns increasing), one may conclude that increased ice cream sales cause sunburns or vice versa that sunburns cause increased ice cream sales. However, one needs to consider the third, intervening variable, of the hot, dry and sunny weather in the summer. This third variable "explains away" the causal relationship between ice cream sales and sun burns; shark attacks and forest fires are other variables that increase in the summer and are not in a causal relationship with ice cream sales or sunburn. However, all of these variables are correlated...in that they all increase in the summer...and the relationships are also spurious, i.e. not causal.
Credits:
Created with images by Tumisu - "graph chart investment" • mpewny - "new york skyline new york city" • pasja1000 - "senior elderly people" • Aaron Burden - "untitled image" • Gaelle Marcel - "untitled image" • Tierra Mallorca - "Shooting in my office"