Week 8-Correlational analysis

Question 1

Define Correlational analysis

Answer 1

Assesses the relationships (associations) between
variables.

Question 2

Define Correlation coefficients

Answer 2

The strength of the association. ranging from -1 to +1.
*Negative values suggest a negative correlation. Positive values suggest a positive correlation. (these are effect sizes).

Question 3

Define Positive correlation

Answer 3

As one variable increases so
does the other.

Question 4

Define Negative correlation

Answer 4

As one variable increases the
other decreases.

Question 5

Define No correlation

Answer 5

No relationship between the
variables

Question 6

What could be possible explanations for correlations?

Answer 6

–A causes B/B causes A:
 Time-order relationship?
–C causes A and B (3rd variable?)
– Coincidence/chance.
 Not likely – a logical relationship.

Question 7

What are the 2 main inferential tests in analysing correlation data?

Answer 7

1.Spearman Rank Correlation
2.Pearson Product-Moment Correlation

Question 8

Explain Spearman Rank Correlation

Answer 8

Ordinal level data
OR interval/ratio level data which is not normally distributed (skewed).

Question 9

Explain Pearson Product-Moment Correlation

Answer 9

Interval/ratio level data which is normally distributed.

Question 10

How do you report correlational results?

Answer 10

1.Type of correlation
2.Was it significant?
3.+/-?
4.Variables
5.Inferential statistics

Question 11

Depending on the direction of your hypotheses, what else can we do with the p-value?

Answer 11

■If we have a one-tailed hypothesis we can halve the two-tailed p value because our alpha (α) value changes (α is the level at which you will say an effect is
significant.)
■As discussed before α is typically .05, so p values below .05 are significant.
■In one-tailed tests alpha is .025.
■So our p value is .008
■This would become .004
■Sometimes SPSS gives you one-tailed and two-tailed p values.

Question 12

Define degrees of freedom

Answer 12

the number of observations (pieces of information) in the data that are free to vary when estimating statistical parameters.

Question 13

What’s the best way to present correlations and why?

Answer 13

Using scatterplots.
*Scatterplots make it very easy to
see the strength of a correlation also showing a ‘line of best fit’.
*This is a line which best represents the pattern of our data.
*Correlational analysis essentially
tests how well this line fits the data.

Question 14

What are correlation matrix useful for?

Answer 14

If you have lots of variables and
you correlate them you could
present this information in a table
known as a correlation matrix. (i.e., good for visualising)