Variables, Design and Hypothesis Flashcards
What is an experimental scientific method?
Vary/manipulate IV whilst holding everything else constant
Measure changes in a chosen DV
Changes in IV should cause changes in DV - can infer causality
What is a quasi-experimental scientific method?
Similar to experimental but IV cannot be manipulated
Can be trickier to eliminate all confounding variables
What designs can a quasi-experimental method be used with?
Non-equivalent groups
Pretest-post-test design
What is a correlational scientific method?
No manipulations made
Measure two or more variables and determine extent to which they are related to each other (co-related)
Cannot infer causality
How many IVs can an experiment have?
One or more
How many levels should an IV have?
Two or more
Why does an IV need levels?
To have comparison to see efficiency
How many DVs should an experiment have?
One or more
What is the operationalisation of DV?
Specifying how we should measure it as precisely as possible
What are the four measurement scales for DVs?
Nominal
Ordinal
Interval
Ratio
What is nominal data?
Non-numerical categories
Very distinct
What are examples of nominal data?
Preferred travel method (car, bus, train, air)
Favourite colour
What is ordinal data?
Discrete numbers in a certain order
What are examples of ordinal data?
Socioeconomic status
Education level
Happiness levels
What is interval data?
Values that have a meaningful difference between them
Continuous data
What are examples of interval data?
Temperature
Year
IQ
What is ratio data?
Values that have an absolute zero
What are examples of ratio data?
Height
Weight
Income
What are confounding variables?
Things that can interfere with results that we’re not controlling
Not manipulated but could have influence on results of an experiment
Want to eliminate/minimise these as much as possible
When can confounding variables occur?
When some aspect of experimental situation varies systematically with IV
What is a between-subjects design?
Participants only take part in one level of IV
How can individual differences be accounted for in a between-subjects design?
Randomly assign participants to groups
How powerful is a between-subjects design?
Less powerful as need more participants to detect a genuine effect
What is a within-subjects design?
Same participant performs all levels of IV
Also repeated measures design
How powerful is a within-subjects design?
More powerful as fewer participants needed to detect a genuine effect
What are order effects?
When order of conditions are same could argue results due to order
Due to practice or boredom
What is best to use in a within-subjects design?
Randomisation of trials
Counterbalancing
What is a matched subjects design?
Participants matched with someone else with regards to demographic characteristics
This “pair” tested as one individual over two levels of IV
What is a correlational design?
Not manipulating variables
Look at variables that already exist and see to what extent they co-vary
Doesn’t imply causation
What are hypotheses?
Theory-driven idea as to why narrow set of phenomenon occur
What are the two types of hypotheses?
Experimental
Statistical
What is an experimental/research hypothesis?
Conceptual idea that tries to explain an observation
Based on our original theories
What are statistical hypotheses?
Specific statement that we can use to collect data and test our hypothesis with
Prediction
What are the two types of statistical hypothesis?
Null (H0)
Alternative (H1)
What is a null hypothesis?
Our observations from our samples imply they come from same population
For parametric statistics, what does a null hypothesis say?
All means equal
For non-parametric statistics, what does a null hypothesis state?
All distributions equal
What is an alternative hypothesis?
Logical alternative to null hypothesis
Predict significant different/relationship between variables
What are the two types of alternative hypothesis?
Directional
Non-directional
What is a directional alternative hypothesis?
Will be a higher/lower difference
What is a non-directional alternative hypothesis?
There will be a difference
What are the properties of H0 and H1?
Mutually exclusive - only 1 statement true
Exhaustive - cover all possible outcomes in experiment
How do you conduct null hypothesis significance testing?
Only reject H0 when probability of it being true (p) lower than specific criterion (alpha)
Using inferential testing
How is an inferential test conducted?
Generate test statistic
Set specific alpha criterion (usually 0.05)
Using these values to help determine our probability value (p-value)
What does p < 0.05 mean?
Less than 5% probability results happened by chance
Suggest something unique happening between population
Significant result
Reject H0 and support H1
What does it mean if p > 0.05?
5% or more probability events happened by chance
Suggest nothing unique happening between populations
Non-significant result
Failed to reject H0 and only have support for H0
