Intro to Research design and data Flashcards
Independent variable
what changes
Dependent Variable
What you measure
control variable
Initially keep the same
control condition
Helps understand role of l V and rule out alternative explanations for results
Extraneous variable
Not controlled in exp, can effect but is unlikely to change direction of effect
confounding variable
EV that varies systematically with l V to influence DV
e.g. every Monday class very noisy due to a football match outside
likely to influence + change direction of results
Experimental Design - Between subject
- Either in one condition or the other
- Each ppt contributes to one data point
- Independent Measures design
Between subject Design Benefit
- Avoids ppts / experimenter effects
- Avoids order + fatigue effects
Between Subject Design Disadvantage
_ Takes longer
- Is less powerful
intro variation due to indiv differences
Within Subject Design
Take part in both conditions
- repeated measures design
within subsect Design Benefit
- Accounts for indiv differences
- cost and time effective
within subject Design Disadvantages
- order effects + fatigue effects
- ppts more likely to guess nature of exp
- Can’t be used with quanti-experimental designs
Matched Pairs Design
Different ppt in all conditions, ppts matched
Matched Pairs Design Benefit
- Accounts for indiv differences
Matched Pairs Design Disadvantage
-Difficult to match people accurately (so match on what’s relevant to exp)
categorical Data
- sometimes called discrete data
- Nominal } No hierarchical order, can be presented as frequencies
- ordinal } Have hierarchy e.g First year, second, third
Numerical / continuous data
- presented as means + standard deviation
- Interval } Scalar, no meaningful zero, temperature
- Ratio } scalar with an absolute data, time, heart rate
APA formatting of graphs
- No title
- Axis Title
- No gridlines
- figure legend underneath graph
standard Error
Standard deviation divided by square root of N
Z scores
-Standardised score
- Rep datapoints relationship to mean
- useful for comparing ppts conditions
-useful when units differ
(score-mean)/ standard deviation
Distribution of data
- Normally distributed = Bell-shaped curve
- most data is around avg point
-test for normality using Shapiro-Wilk
confidence Intervals
- used as a measure of spread of data
- Estimate population with 9 5 % confidence mean lies within this range
- 1.96 standard deviation either side of the mean
- 1.96 is a Z score
= Mean +/- ( 1.96 * Standard Error )
Non-normal distributions
Neg skewed} mean Pulled down
pos skewed}mean pulled up
- In this case median is a better representation
- use non-parametric tests
Experimental designs
- manipulate 1 variable systematically + see its affect on the other variables for us to establish a causal relationship
- random allocation of ppts to a condition
Correlational designs
- Have no IV or DV
- Look at relationship between variables
- Can’t infer causation from correlations
pos skew
- tail on the right is longer than the left
- peak of graph is to the left
neg skew
- peak of graph is to the right
- left tail of graph is much longer
Normal distribution properties
- symmetrical about the mean
- Tail should meet the x axis at infinity
- bell -shapes
- equal mean, median + mode
