IV and DVs / Hypotheses Flashcards
What is an IV and a DV
IV: the variable you manipulate
DV: the variable you measure
What is a one tailed hypothesis with example
Operationalise variable. Say if more or less. Predict direction.
A dose of SSRIs one month later will lead to a reduction in the Beck’s Depression Inventory
What is a two tailed hypothesis with example
Non directional. There will be a significant difference but do not say direction.
There will be a difference in Beck’s Depression Inventory Scores when people have SSRIs compared to those who do not take SSRIs.
What is a null hypothesis with two examples
The null is where there it is everything the hypothesis is not
So on the SSRI one the null will be there are no difference, or it increases the Beck’s depression inventory. So there are two null hypotheses / two ways of formulating it.
What must variables be?
What is a variable?
Something that varies or can be varied!
* Must have at least two possible values.
* Must be observable and recordable.
* Research is all about operationalized variables
They must be:
All variables need to be valid: they need to ,measure what is supposed to be measured. Validity is a continuum and not an absolute. IE is it measuring cognitive function, or just recall or numbers?
Variables must also be reliable: produces consistent measurements in the same situations. Can assess via test retest, inter-rater reliability (both rate a task and see if they correlate), correlation between various items.
When is a result significant? What happens if a result is significant?
We want the p value to be really small. P equal or less than 0.05 then we accept the hypothesis. If it is not significant then we cannot reject the two nulls we have put forward.
If you get non significant result you NEVER accept the null, it means you CANNOT REJECT the null. You either reject the null or do not reject the null hypothesis.
How can we improve reliability?
- Replication (more a result is repeated, the more reliable it is)
- Pre-registration (The Registered Replication Report (RRR)
- Registered reports
- RRR: Registered Replication Reports - The Registered Replication Report (RRR) is a new type of article introduced by Perspectives on Psychological Science. It consists of a collection of independently conducted, direct replications of an original study, all of which follow a shared, predetermined protocol
- Bayesian analysis
- Meta-analysis (a systematic review of all studies in a field, published or not, see if overall significant effect or correlation seen)
How can we assess reliability?
Can assess via test retest method ( It measures test consistency over time by giving the same test twice to the same people at different times to see if the scores are the same) Test-retest reliability is calculated using the Pearson Correlation Coefficient between the results of the test taken at two different times, and a correlation of at least 0.80 or higher typically indicates good reliability.
Inter-rater reliability (both rate a task and see if they correlate) - extent to which different raters or evaluators agree in assessing a particular phenomenon, behavior, or characteristic. It’s a measure of consistency and agreement between individuals scoring or evaluating the same items or behaviors.
Correlation between various items.
What is a good score for a correlation? ie a coefficient that would indicate good reliability or validity
0.7. 1 is perfect.
What is an issue with the peer review process at the moment? What is being done about it?
Publication bias – Significant results get published, non significant do not. File-drawer effect. Michal gave example of study that was found as significant but never repeated
What is operationalisation?
A description of a construct such that another researcher can produce or measure the same thing
