Session 1a Flashcards
The use of statistics
Answer research questions
Population
All entities/individuals of interest. Research questions are often about populations. A parameter is a value that describes the population (like the mean, the variance, etc.)
Smaple
A subset of individuals from the population. This is the data that is examined during a study. An estimate is a value that describes the sample (ex. mean, variance)
Descriptive statistics
Summarize/describe properties of the sample (or the population if we gather data from the entire population)
Inferential statistics
Draw conclusions/inferences regarding the properties of the population, but based only on sample data
How to decide what statistical analyses to do
The appropriate analyses depend on:
1. Study design / research question
2. Type of variables (level of measurement, distribution)
3. Whether assumptions of the analyses are met
Variable
A characteristic that varies across observations (people, location, time, etc.). It’s often a single column in the dataset.
Independent Variable (IV)
- Predictor (or covariate)
- Factors in an experimental design
Dependent Variable (DV)
- Outcome/Response
- Predicted variables
Correlational research
IV is measured by the researcher. It’s food for ecological validity (generalizing research findings to the real world), but not good for inferring causality.
IV and DV may have a relationship due to a 3rd (confounding) variable, or they have a common cause.
Experimental resarch
IV is manipulated by the researcher. It’s good for inferring causality. Manipulating IV in lab settings may sometimes feel detached from the real world.
The statistical methods to analyze data may be the same/similar for correlational and experimental designs.
Between-subjects design
- Each participant in only one experimental condition (e.g., control or treatment)
- If random assignment is used, groups should be approximately equal on any confounding variables
Within-subjects design
- Each participant does more than one experimental conditions (e.g. control and treatment). DV measured multiple times
- Vulnerable to practice effects and fatigue/boredom effects as alternative explanations for differences between conditions. Counterbalancing is used to help rule out these alternative explanations.
Types of variables (levels of measurement)
- Quantitative (high level): ratio, interval
- Categorical (low level): ordinal, nominal
Nominal variables
Classifies objects
- Are two observations the same or different on some attribute?
- Not quantitative, though we can use numbers to index the categories
- When dichotomous: two categories (ex. treatment vs. control)