RDA Test 2 Flashcards
What is a relationship design?
Whether variables share a common relationship (i.e. as one thing increases, the other measurement either increase or decreases in response)
What is a difference design?
Whether there are differences between measurements depending on how the participants are measured
What is counterbalancing and why is it used?
Where the order of tasks/conditions for participants are altered for each participant
Used to deal with participants improving or exhibiting demand effects
What is a floor effect?
Where everyone does very poorly, or near the minimum value
What is a ceiling effect?
Where everyone does very well, or near the maximum value
What is the Hawthorne Effect?
Where participants modify their normal behaviour due to them being aware that they are being observed
What is the Clever Hans Effect?
When researchers behave in a way that can influence the participant to act in a way that is desirable for the study
What is social desirability bias?
When participants answer in a way that would be viewed more positively in terms of social norms
What is acquiescence bias?
Where a participant answering questions might agree or disagree with statements/questions without the answer being a true reflection of their feelings or opinion
What causes acquiescence bias?
Participant motivation, the set of options they can choose, all questions being phrased in the same way, etc
What should a sample be representative of?
The population of interest
What is a longitudinal study?
Where we study a behaviour or subject of interest over a longer period to time and we usually make measurements at regular intervals
What is the main reason why longitudinal studies are important?
Allows us to see how things change over time (e.g. an intervention to change a behaviour) within the same group of participants
What is a cross-sectional design study?
Where behaviour is observed over time, but instead of following one group of participants, it uses a cross-section of the population of interest and divides them into time groups
What do descriptive statistics allow us to do?
- Explore and compare our data meaningfully
- Assess any major differences between conditions/variables
- Determine the ‘shape’ of our data distributions
- Check for missing data or unusual data (e.g. an outlier)
- See how much ‘noise’ there is in our data
- Check to see whether the data are ‘fit’ to use for further statistical testing
- Look at measures of central tendency, dispersion and variation
- Organise and aggregate or disaggregate data in a meaningful way
- Get a ‘feel’ for any relevant patterns
- Present data graphically or in a tabular format
What do inferential statistics allow us to do?
- Test whether our data is normally distributed
- Determine whether differences or relationships are statistically meaningful
- Express whether we can retain or reject the null hypothesis
What are the 3 main measures of central tendency?
Mean, median, mode
What is the mean?
The average of a set of numbers, whether integers or decimals
How is the mean worked out?
Add all items in a set up, and then divide by the number or items
What is the most common measure of central tendency?
Mean
What is the median?
The middle of a set of values if you arrange them from smallest to largest
When is the median most useful?
When you have a non-normal distribution or with extreme scores, or when using ordinal data
What is the mode?
The mode is the most commonly occurring number in a set of data
When is the mode most used?
With nominal data
What do measures of variability tell us?
About the spread of data and in some instances, the amount of ‘noise’ in the data set
What are the 5 common measures of dispersion?
- Range and interquartile range
- Mean absolute deviation
- Variance
- Standard deviation
- Standard error of the mean (SE)
What is the range?
The difference between the smallest and largest value is a distribution
What is the biggest limitation regarding the range? What can be used to remedy this?
It is susceptible to an extreme score in a distribution
The interquartile range can be used to remedy this
What is the interquartile range?
When we take the centre 50% of values, between the 25th and 75th percentile
How is the interquartile range usually displayed graphically?
Box plot
What is the mean absolute deviation?
A measure of how much difference or deviation there is from the mean
How is the mean absolute deviation usually worked out?
By working out the difference between each value and the mean (ignoring all the negative signs – otherwise they would sum to zero), adding them up, and then dividing by N
What is variance?
An indication of the overall amount of variability in a set of data, but in a squared format, is used in many statistical formulae, and is denoted as s2
How is variance worked out?
By adding up the squared differences (deviations) between each value in a set of data and the mean, and then dividing by N – 1
What is standard deviation?
The square root of the variance
Measures in the original units of measurement and relates to the standard normal distribution, so we can get a much better sense of the distribution of scores in our data
What does the standard error of the mean measure?
The standard deviation of the population mean
What is kurtosis?
A measure of the tailedness of a distribution (how “pointy” a distribution is)
What is skewness?
Degree of asymmetry, can vary in severity, and can be either positive (a positive skewness value), negative (a negative skewness value) or neither (a skewness value of 0)
What do scatterplots show?
The relationship between variables
What are the features of a scatterplot?
- The plot normally has the predictor variable on the x-axis, and the outcome variable on the y-axis
- We can also include a line of best fit (i.e. the best linear statistical model for the data)
- This can also include some measure of ‘noise’ around the best fitting model (e.g. standard error of the mean, confidence interval, etc.)
What do boxplots look at?
The range of data
What are the features of a boxplot?
- The middle 50% of data that is not disturbed by the outliers (extreme scores) (indicated by dots outside the whiskers)
- Any potential outliers that might be causing skew
- The median value in the distribution (the black bar in the box)
- The whiskers represent the top 25% and bottom 25% of values
How do we tell if a distribution is roughly normal with boxplots?
If the box is proportional to the whiskers
What do histograms look at?
The distribution of data
What are quantile-quantile plots (Q-Q plots) used for? And how do they do this?
To assess whether the data is normally distributed
They do this by comparing a theoretical distribution of values (for the mean and standard deviation of our data) on the x-axis, with our observed values plotted on the y-axis
If the data is normally distributed, then we expect the data points to form a straight line across the graph
What does it mean if something is statistically significant?
The probability of an observation occurring is remote enough
What is threshold for statistical significance?
p ≤ .05
Is p = 0.68 significant?
No
Where is the critical region on a distribution curve for a one-tailed prediction?
The last 5% of the tail (the rejected region)
What are non-parametric tests?
Tests used if the distribution is not normal (e.g. chi-square)
What are parametric tests?
Tests used if the distribution is normal (e.g. t-test)
What does the Shapiro-Wilk test tell us?
Whether our distribution of values is significantly different to normal. That is, whether it’s too asymmetrical.
We use this to see if our distribution is normal or not.
What does bivariate correlation focus on?
The relationships between samples
What does correlational analysis allow us to do?
- Explore whether there is a real relationship between variables which is unlikely to have occurred due to external factors (i.e. unlikely to have occurred due to chance)
- Also allows us to determine the:
- Direction of the relationship
- Strength of this relationship
How is correlational analysis usually represented visually?
Scatterplot
What statistical test should be used for this study:
- Relationship study
- Categorical data
- Parametric
Pearson’s r
What statistical test should be used for this study:
- Relationship study
- Categorical data
- Non-parametric
Spearman’s rho
What statistical test should be used for this study:
- Difference study
- Within participant design
- Parametric
Paired samples t-test
What statistical test should be used for this study:
- Difference study
- Within participant design
- Non-parametric
Wilcoxon
What statistical test should be used for this study:
- Difference study
- Between participant design
- 2 groups
- Parametric
Independent samples t-test
What statistical test should be used for this study:
- Difference study
- Between participant design
- 2 groups
- Non-parametric
Mann-Whitney
Correlation does not allow us to say anything about _________.
Why?
Causality
There may be a 3rd variable that we are unaware of
On a plot, what axis does the predictor variable go on and which axis does the outcome variable go on?
Predictor - X-axis
Outcome - Y-axis
What is the correlation coefficient calculated using?
Covariance
What is covariance?
The degree to which scores on two variables deviate from their sample means
Why do we need to use standardized units?
Because two variables will most likely be two different types of measurement (e.g. number of ice-creams sold vs. temperature) and so we need to convert into a common metric
When we standardize our measures we end up with a value that ranges from -1 > + 1
How do we work out the z-score?
We use the coefficient (r) and divide by the SE (standard error) of r
What does the z-score tell us?
Whether our r value deviates from zero enough to be in a critical area of the normal distribution
What does a value of 1 mean for correlation coefficient?
Perfect positive correlation
What does a value of 0 mean for correlation coefficient?
No correlation
What does a value of -1 mean for correlation coefficient?
Perfect negative correlation
What does a positive correlation mean?
As one variable increases, the other variable increases in response
What does a negative correlation mean?
As one variable increases, the other variable decreases in response
On a scatterplot, how can we tell how strong a relationship is?
The more the points adhere to a straight line, the stronger the relationship is, the more scattered the points are, the weaker the relationship is
What does high correlation mean for shared variance?
Lots of shared variance
What does low correlation mean for shared variance?
Little shared variance
What would we expect when one variable deviates from the mean if the 2 variables are related?
Similar changes for the other variable
How do you calculate shared variance?
r x r (x 100 [for %])
What does the spearman’s rho test do?
Ranks the data from 2 variables
Each value is assigned a rank (e.g. 1st, 2nd 3rd, etc., plus ranks that are tied in rank)
Pearson’s formula for r is then applied to the ranks to calculate the correlation statistic (rs)
Why is it better to use pearson’s r in comparison to spearman’s rho?
Spearman’s rho is less likely to find significance due to it not being as robust
Describe what descriptive statistics can tell us about data
Allows us to look at measures of central tendency, and get a feel for any relevant patterns
Describe what inferential statistics can tell us about data
Whether our data is normally distributed and determines whether difference or relationships are statistically meaningful
What does a positive skew look like?
Scores fall to the top end of the distribution
What does a negative skew look like?
Scores fall the bottom end of the distribution
