Experimental design and t-test Flashcards
What is a Quasi experiment ?
What is the advantages / disadvantages ?
When analysing two naturally occurring variables.
E.g. People’s income and their happiness
Advantage:
This has high ecological validity: represents data from the real world (in contrast to manipulative experiments)
Disadvantage:
Can’t explain causality, there could be a lot of underlying hidden variables (cf. third variable problem)
What is a Full experiment ?
What is the advantages / disadvantages ?
In controlled experimental settings, experimenters can manipulate the independent/predictor variable to observe if the outcome/dependent variable changes.
Advantages: This give us possibility to infer causality
Disadvantages: Because of the experimental setting it doesn’t represent the real world as well (low ecological value)
What is Independent measures design / Between-participant design ?
What is the advantages / disadvantages ?
Two different groups of participants (most commonly a control group and a test group)
Advantages: The participants doesn’t encounter the order effect
Disadvantages: hard to balance groups perfectly (which is why we pseudorandomize; making a good distribution of gender, age etc.)
What is Repeated measures design / Within-participant design ?
What is the advantages / disadvantages ?
The same people participate in two different conditions in the experiment
Advantages: There are fewer individual differences
Disadvantages: The order effect (the fact that they’ve done condition 1 will affect their behaviour in condition 2)
What is the t-test ?
+ use the “population explanation”
The variance between groups compared to within groups
It is under inferential statistic and is used to determine if there is a statistically significant difference between the means of two groups
We take the difference between the means (effect) and divide by standard deviation og the two groups (error)
Population formulation:
- Do the two groups come from two different populations? (→ true effect)
- Or from the same underlying population? (→ sampling noise)
in other words:
Do the differences in our samples reflect differences in the population?
What is the formula for the t test ?
t = (x1 − x2) / sqrt(s1/N1 + s2/N2)
The variance between groups compared to within groups
What is the p-value?
How do you interpret the p-value in the t-test?
So, you can’t really infer any usable data from the t-distribution, therefore we make it a probability distribution where we then get some p-values.
When we perform a t-test, we test the hypothesis that the two samples have the same mean.
The smaller the p-value, the more likely it is that you should reject H0
Normally the threshold is 0.05 (5%), so if the p-value is smaller than 0.05 the H0 should be rejected meaning that the likelihood of H1 is supported, the chance of real statistical significance between the two means is supported.
What is hypothesis testing all about?
Finding out whether you should reject your null-hypothesis….
