Correlation Flashcards
What do correlation studies investigate?
Whether there are relationships between different co-variables.
What does positive correlation mean?
As one co-variable increases, the other co-variable increases.
What does negative correlation mean?
As one co-variable increases the other co-variable decreases.
What does no correlation mean?
There is no definite trend and that the two co-variables do not appear to be related to each other.
What are the variables under analysis called in correlation research?
Co-variables
What is a research question?
- A broad question about the concept being investigated.
- They must end with a ‘?’.
What is the research aim?
- Identifies a more specific concept within the research question which the research wants to investigate.
- The researcher’s aim will have an impact in the research method they go on to use (e.g. a correlation study or an experiment).
What does it mean to operationalise variables? CHECK
Making variables measurable.
What should you keep in mind when operationalising variables in a correlation study?
- Any co-variable needs to be expressed as a number
- Each participant must have two separate (ordinal level) numbers that relate to them - one for each co-variable.
.
What does a two-tailed correlation hypothesis predict?
A significant relationship is being predicted but no expectation is being conveyed as to the direction that the relationship might go in.
What form does a two-tailed correlation hypothesis take?
“There will be a significant correlation between co-variables y and z.”
What does a one-tailed correlation hypothesis predict?
A significant relationship is being predicted indicating the researcher’s expectation that the results will go in a particular direction.
What form does a one-tailed correlation hypothesis take?
“There will be a significant positive/negative correlation between co-variables y and z.”
What does a null correlation hypothesis predict? CHECK
Any apparent correlation between the two co-variables will not be sufficiently large enough to be beyond what might be expected by chance.
What form does a null correlation hypothesis take?
“There will not be a significant correlation between co-variables y and z; any relationship will be due to chance factors.”
Define primary data.
Data gathered directly from the participants by the researcher.
Define secondary data.
Data that has already been gathered by someone other than the researcher.
What form of data can psychologists carrying out correlation studies use?
- Primary data
- Secondary data
How would the form of data used affect the sampling method used? check if relevant
The researcher will need to decide what the target population is that they want to be able to make generalisations about and then settle on a way of obtaining a sample of either participants (primary data) or data that will be representative of this (secondary data).
How would the form of data used affect the ethics? check if relevant
- If collecting primary data, the researcher will clearly need to obtain consent from participants, avoid deceiving them, give them a chance to withdraw their data etc…
- If using secondary data, then it may be that this is already in the public domain. However, the researcher would still need to ensure that the people whose data is being analysed are not identifiable in any way so as to respect their right to privacy.
What is the appropriate form of descriptive statistic to use to visually display the results of a correlation study?
A scatter graph
What does each cross on a scatter graph represent?
A different pair of values
What does the total number of crosses on a scatter graph indicate?
The size of the data set the researcher has been working with.
Distribution of pairs of values ? CHECK IF NEEDED
.
.
Broadly speaking what are the findings from a scatter graph?
Relate to raw data:
- Modal
- Median
- Mean response on a particular co-variable
- Range of responses for a particular co-variable
- Whether a pair of values constitutes as an outlier
Broadly speaking what are the conclusions from a scatter graph?
Relate to broad inferences that you can make from that raw data:
- Direction of any correlation
- Approximate strength of an correlation
What should you consider when drawing a scatter graph?
- Precise title making it clear that it related to a relationship between two co-variables.
- Each axis should start from zero (can have a zigzag to show a broken scale after the zero).
- Correctly labelled axes (it doesn’t matter which co-variable goes on either axis).
What is an inferential statistics test? CHECK
What is the range that the correlation co-efficient can sit within?
-1 and +1
What does a correlation that has a plus sign as part of its correlation co-efficient show?
A positive correlation
What does a correlation that has a minus sign as part of its correlation co-efficient show?
A negative correlation
What does a correlation with a co-efficient around zero show?
No correkation
What does it mean the further away the correlation co-efficient from zero?
The stronger the correlation
What correlation co-efficient represents a perfect negative correlation?
-1
What correlation co-efficient represents a perfect positive correlation?
+1
What are the advantages of correlation studies?
- It can tell not only tell whether two co-variables are related but also the direction and strength of the relationship.
- It is a useful technique when either practical or ethical reasons mean that variables can’t be manipulated.
- Act as a good starting point for research.
What are the disadvantages of correlation studies?
- They do not tell us anything about cause and effect.
- The inferential statistical tests will not always pick up on a relationship between two co-variables.
- They do not require the collection of any qualitative data which limits our ability to know what sits behind (and might explain) any relationship that might be found.
