Full book glossary Flashcards
Interval variable
A variable where the values are numerical and where differences between values are consistent across the range of values
Axes
The horizontal and vertical scales of a graph (see x-axis and y-axis)
Unique effect size
The result of ANOVA analysis. A measure of the effect of each IV on the DV which discards all overlapping covariation. The unique effect size is the contribution made by an IV to a DV that is not made by any other measured IV
Confidence interval
A range of values that a particular statistical value is believed to fall within at a specified level of confidence e.g. 95%
Inter-quartile range
Measure of dispersion used typically for ordinal data; the range of values in the middle two quadrants of a data set
Maximum likelihood
A value of an estimated parameter (mean, coefficient etc.) that has the highest likelihood making it the most likely value
Association
Neutral word to describe a relationship
Direct effect size
The effect size in a linear model between an IV and a DV. It shows how much the DV is changed by a change in the IV
Extraneous variable
An unwanted variable that may be influencing a DV
Model effect size
Effect size of the whole model on the DV
Confirmatory research
Research designed to test a specific hypothesis
ANOVA
Analysis of Variance, type of statistical test that splits variance of DV into its various sources
Covariance
A measure of the joint variability between two variables
Scale
A common way of using numbers to describe quantities. The use of cm to describe the length of things is an example. Other scales can be used for the same purpose (inches, cubits). Scales specify how much the quantity involved must be changed to increase its value by a certain amount
Nominal variable
Another term used for Categorical variables
Outlier
- Authors of new, alternative statistics textbooks
- Also used to describe data points that appear not to belong to the population being studied, either randomly occurring or due to research error
Categorical variable
Variable where data is divided into labelled groups (categories); there is no order to the groups
Likert scale
Ordinal self-report scale used for responses to questions, often with agree – neutral – disagree options
Linear
- Used to describe relationships where both variables plotted together form a straight line
- Also used to describe combinations of variables by addition
Mediation
The possibility that the effect of an IV on a DV occurs through an intermediary (mediating) variable
Variable
Any way in which people/animals/situations differ
2-Tailed Test
Statistical test with no predicted effect size direction
Experimental study
Study which deliberately creates an independent variable, such as dividing participants into active and control groups
Variance
Measure of variability: spread of values, the square of the standard deviation
Order effects
Within-participants design issue where participants are influenced by experiencing one condition before another: can influence the IV and the DV
Idea
Any proposed relationship between variables, specific or otherwise
Uncertainty
In psychological research, uncertainty is how well (or not well) a sample parameter matches a population parameter
Discrete
A scale or a set of values are discrete if there are breaks or jumps in between them. Having group members labelled “person 1” and “person 2” uses discrete values as there is no “person 1½”
Contingency table
A table comparing expected and observed frequencies for a chi-square test
Type II error
False negative outcome; when the null hypothesis is not rejected, despite the population having an effect
Null hypothesis
The hypothesis that there is no effect in the population
Correlation coefficient
Effect size that measures the strength of a linear relationship on a normalized scale
Median
Measure of central tendency for Ordinal data, the middle value of a data set or halfway point between two central values
Non-parametric test
Statistical tests conducted on Ordinal or atypical data sets which don’t meet the requirements of parametric tests
Standard error
The standard deviation of the Sampling Distribution
Null hypothesis testing
Test used when looking for evidence to reject the null hypothesis
Likelihood
The relative chance that an event (like a sample) was caused by another event (like a specific population)
Variability
A term to describe the mount of differences between different people/animals/situations
Scatter plot
Type of graph with two continuous numeric variables plotted. Each data point is a single unconnected point in the graph
Opportunity sampling
Sampling strategy that uses any available group; no attempts to match a population or collect a random set of participants
Categories
The values of a categorical variable; also referred to as groups. Traits (e.g. blue eyes/brown eyes) or situations (before/after)
Path model
A model where variables are not as a simple split into predictors (IVs) and a response (DV) but are instead allowed to connect more freely
Mixed measures
Term used to describe an ANOVA where 1 or more variables are between-participants and 1 or more are within-participants
Likelihood function
A graph that shows the relative likelihood that a particular sample effect size came from a range of different population effect sizes
Residual
The difference between the expected value for an Interval DV and the actual value for one participant
AIC
Akaike Information Criterion. Value used to assess the fit of a model, typically by comparing models. Smaller is better
Natural effect size
An effect size measured in the same units as the DV
Type I error
False positive; the incorrect rejection of the null hypothesis
X-axis
Horizontal scale of a graph
Deviation
The difference between the value of a data point and the mean for that data set
Local mean
A theoretical concept where a set of values are divided into infinite small groups, and each group has its own mean, to generate a line of best fit
Logarithmic scale
A transformation that expands the amount of space given to small values and contracts the amount of space given to large values
Value
A specific description of a variable for one person or situation: e.g. ‘green’ for the variable eye colour, or ‘112’ for IQ score
Endogenous variable
A response variable, or dependent variable in a model
Interaction
A mechanism by which the effect of one IV on a DV depends on the value of another IV
Ordinal variable
A variable with ordered values, but no meaningful difference between values
Bar chart
Chart typically used to present frequencies of categorical data. Data illustrated using separated columns
Standard deviation
Measure of dispersion: based on the range of squared deviations from the mean. A small standard deviation indicates a small spread of values in a group
Parametric test
Standard null hypothesis statistical testing, typically between Interval and Categorical variables
Dispersion
The spread of a set of a values, often defined using the standard deviation or inter-quartile range
Patchy coverage
A situation where the range of values obtained for a variable (usually an IV) does not cover the full possible range but instead has gaps
Hypothesis
A specific and testable prediction including a description of expected effect
Sample size
Number of participants recruited, commonly denoted with ‘n=’
Observational study
Study that makes use of variables that already exist, as distinct from an experimental study where an experimental variable is created
Marginal distribution
The distribution of values for a single variable – usually used in the context of a scatter plot of two variables
Exogenous variable
A predictor or independent variable in a model that is not proposed to have any causes amongst the other variables
Sum of squared deviations
A useful intermediate step in calculating means, standard deviations, regression and ANOVA
Standardized effect size
An effect size system that uses the range zero to infinity. Standardized effect sizes do not depend on the scales used to measure variables: they use a standard scale
Estimate
Logical guess based on known information
Repeated measures
Term used to describe an ANOVA where both IVs use a within-participants design
r-statistic
Normalised effect size ranging from -1 to +1, where 0 is no effect
Mean
Measure of central tendency for Interval data, the value where the sum of squared deviations is smallest
Statistically significant
Term given to results typically where p<0.05 (or other alpha)
Contrast test
A modified version of a 1-tailed test where categorical data is converted to numeric data to allow a correlation
Distribution
The whole set of values for a variable in a population or sample. Distributions are usually illustrated by showing the frequency of different values
Inferential statistics
Statistics which use sample data to produce knowledge about a population and determine the uncertainty of that knowledge
Confidence limits
The end values of the range of a confidence interval
Between-participants
Design choice for categorical data, where participants each only experience one group of a categorical variable
Replication
The process of repeating a piece of research, copying the original design as accurately as is possible. It is normal to increase the sample size to have more statistical power
1-way ANOVA
Statistical test for Categorical IV and Interval DV
Central tendency
The typical value used to summarise a set of values
1-Tailed Test
Statistical test with a predicted effect size direction
Count
See frequency
General linear model
A statistical model where there is a set of predictors (IVs) whose added effects are thought to relate to the response variable (DV)
Bivariate
Meaning two-variable
Regression line
The straight line of best fit between two Interval variables. Best fit is usually assessed by minimizing the sum of squared deviations, and the regression line is then sometimes called the Least Squares fit
Counterbalancing
A within-participants design measure which allows some participants to experience one situation first then the other, and the other participants to experience the situations in the opposite order. Controls for order effects
Measurement error
The difference between the real value of some variable in a participant and the measured value
Dependent variable (DV)
A variable that is treated as if its value may have been affected by the values of any other variables in a hypothesis. It is usually the variable we are trying to explain
Publication bias
The tendency for journals and other publications to only report research which delivers statistically significant findings
Sampling distribution
The distribution of expected sample effect sizes given a particular population and sampling design
BESD
Binomial effect size display. Effect size that measures effects as comparison between expected and observed values
Population
Everyone of interest in a piece of research: as big as everyone in the whole world, as small as any limited group chosen
Skew
Asymmetric distribution with one long tail, either positive or negative
Placebo
The non-specific effect of a treatment that arises purely because of participant expectations (e.g. sugar pills which cure headaches)
Line graph
A graph where a sequence of values are shown as (usually but not necessarily) connected points
Significance testing
See null hypothesis testing
APA format
American Psychological Association format for presenting statistical results
Sample effect size
Measured effect size of a sample: always certain, if the calculations are done correctly. But an uncertain estimate of the population effect size
Sample
Set of participants recruited for a piece of research
Stratified sampling
Recruitment strategy that attempts to match the frequency distribution of the population to create a representative sample
Sampling error
The (usually unknown) difference between a sample estimate and the population value. It is usually unknown because we don’t usually have the population value
Y-axis
Vertical scale of a graph
Data analysis
Typically used to refer to the inferential stage of statistics
Exploratory research
Research intended to investigate phenomena without prior expectations that might be confirmed or not
Correlation
The strength of any linear relationship between two variables
Dummy variable
Extra variables created to replace categorical variables. Dummy variables can be treated as numerical with values 0 and 1
Omnibus test
A test that there is some effect in a set of data, without being more specific about where
Logistic regression
A form of regression where the DV is Categorical and the regression line (which is usually S-shaped) gives the probability of the DV being one or other category as a function of the IV
Moderator
Another term for interaction. One IV is said to moderate (change) the effect of another IV on a DV
Regression
A process of measuring the quantitative effect of one or more Interval IVs on a DV
Data
Information deliberately collected together to answer a question: quantitative, in the context of this book
Effect size
A value used to quantify the strength of a relationship between variables
Bias
A systematic (i.e. not random) difference between a sample and the population
Pearson correlation
Test used to assess linear relationships between numeric variables
Average
Also called a ‘typical value’; measure of central tendency. Single value which is used to summarise a set of values. Usually refers to the mean
Non independence
Used to describe participants who are related in some way
Mode
The most common value. A measure of central tendency useful especially for Categorical data (typically the biggest group)
Constant
Any number that doesn’t change
Square root
Any number to the power of ½. The square root of 9 is 3
Statistical test
General term typically used to describe null hypothesis statistical testing
Causation
The assumption that one variable affects another
Alternative hypothesis
The hypothesis that makes the prediction that researchers are interested in. Not statistically tested
Power analysis
Calculation based on predicted effect size to determine sample size that has 80% chance of producing a significant result (referred to as power)
Normal distribution
Typically described as a ‘bell shaped curve’; a symmetrical smooth distribution where the mean, mode, and median are equal and at the peak of the curve
Post-hoc
A specific test that is applied after an initial omnibus analysis has shown that there is an effect within the data. The post-hoc test is designed to establish where the effect lies
Inference
Using statistics to produce knowledge about a population based on data from a sample; a conclusion that is always uncertain
Alpha
Accepted criterion value of p for statistical significance, also chance of a Type I error. Most commonly used value is 0.05
P-value
Probability value: the probability of the null hypothesis producing the same or a more extreme result. Used to determine statistical significance
Descriptive statistics
Statistics used to describe a sample, to summarise and indicate patterns
Research design
Decisions about sample, measurement, and data collection for research
Random sampling
Sampling strategy where all members of a population have equal chance of being selected for a sample
Covariation
Possible relationship between two different IVs
t-statistic
Test statistic for the t-test. The t statistic is found by dividing an estimated value by its standard error
Degrees of freedom
A measure of how much raw information remains in a sample after statistical calculations. Initially, the degrees of freedom equals the number of data points. When a mean is calculated, the degrees of freedom is reduced by one because the mean replaces that much raw information
Model
A description of a pattern in some data. The model is then thought to capture important variability in the data
Test statistic
Output of any statistical test: used traditionally as a halfway step for calculations of p and then significance when tests are done by hand
Normalized effect size
An effect size measured on a scale from -1 to +1 (see r-statistic)
Main effect
The independent effect of one IV on one DV
Independent variable (IV)
Variable that is predicted to influence the DV. The IV is treated as independent of the DV
Frequency
Relative commonness of a value of a variable: e.g. proportion of participants in one group
Pooled standard deviation
Typical standard deviation of multiple groups, weighted by group size
Probability
The chance that an event in the future will happen
Within-participants
Research design for a Categorical IV where participants experience all groups of the IV
Total effect size
The effect of one IV on a DV, disregarding any other variables
Cronbach’s alpha
A measure of the consistency of responses to a set of questions about a common state or trait
Coefficient
A ratio: how much change in a DV for a given change in an IV
Latent variable
Any unmeasured variable: internal state or otherwise