FA1 Flashcards
Correlational, ordinal (qualitative)
Spearman’s correlation coefficient
Correlation, interval or ratio (quantitative), non-parametric
Spearman’s correlation coefficient
Correlation, interval or ratio (quantitative), parametric
Pearson’s correlation coefficient
Experimental, independent groups, ordinal
Mann-Whitney u test
Experimental, independent groups, interval or ratio, non-parametric
Mann-Whitney u test
Experimental, independent groups, interval or ratio, parametric
T-test (unpaired)
Experimental, matched participants or repeated measures, ordinal
Wilcoxon-signed ranks test
Experimental, matched participants or repeated measures, interval or ratio, non-parametric
Wilcoxon-signed ranks test
Experimental, matched participants or repeated measures, interval or ratio, parametric
T-test (paired)
Design
* Independent groups design
* Sample Size: 20 participants
Mann-Whitney U
- Independent groups design
- Not normally distributed data as sample is less an 15/group
Why do you use mean as a central tendency?
Because there are no outliers
Why do you use median as a central tendency?
Because there are outliers
Where you are giving your “opinion/rating” =
Ordinal
Where you are stating a “measured” variable e.g. time =
At least interval
Note:
Independent groups design means
two names = Mann-Whitney
E.g. relationship question + answer
The relationship between age and the identification of faces in ambiguous pictures
The relationship is positive and strong
Contrasting current research and previous research
The Pearson correlation coefficient for the current research (0.98) is larger than the Pearson correlation coefficient for previous (0.70) indicating a stronger relationship
Conclusion =
As … (e.g. age) increases, so does the … (e.g. identification of faces in ambiguous figures) significantly increases
Distinguish standard deviation e.g.
The data for the matching condition (1.17) had a larger standard deviation compared to the data for the mismatching condition (0.79).
Greatest variability =
Largest standard deviation
Strong
0.7
Moderate
0.3
Weak
0.1
p = 0.05 or more
Fail to reject
p = <0.05
Reject
Independent groups design
Groups only try one condition, Mann-Whitney U
Repeated measures design
The groups try both conditions
Parametric for experimental
More than or equal to 15
Parametric for correlational
More than or equal to 25
Standard error of the mean
Precision of estimating the population mean (higher is less precision)
If the tail of graph is on the left it is
negative
If the tail of graph is on the right it is
positive
What does a distinguish question require?
- Definition of 1st concept
- ‘whereas’ (conjunction)
- Definition of 2nd concept
What does a contrast question require?
- Definition of 1st concept (what do they share)
- ‘whereas’ (conjunction)
- Definition of 2nd concept (where do they differ)
(GIVE MORE)
What does a compare question require?
- State a similarity using concept 1 and 2 ‘are both’
- State a difference using concept 1 ‘whereas’ concept 2
- State a significance of either the similarity OR difference ‘this is significant because…’
Confidence interval
range of values
true population mean
The smaller the confidence interval…
the more precise the estimate and the greater the certainty when estimating the population mean
Less than .05 p-value
statistically significant and the null hypothesis can be rejected, the alternative hypothesis is supported
Equal to or greater than .05 p-value
not statistically significant and the null hypothesis cannot be rejected
Error bars are overlapping
no statistical significant difference
Error bars are not overlapping
there is a statistically significant difference between the two conditions
Error bars are substantially overlapping
no statistically significant difference
Type I error
P-value said there is a significant difference but in the population there is actually no significant difference
Also known as a false positive
Type II error
P-value said there is no significant difference but in the population there is actually a significant difference
Also known as a false negative
