ACE Model: A twin study model where variance for a certain trait is broken down into three factors: additive genetic factors (A), common environmental factors (C) and unique environmental factors (E) (Neale and Cardon, 1992).
- Additive genetic factors (A) are traits determined entirely by genes. Monozygotic (MZ; identical) twins have identical genes, and thus a correlation of 1.0. Dizygotic (DZ; fraternal) twins have a correlation of 0.5.
- Common environmental factors (C) are influences that both twins have in common, such as family environment. These correlate to 1.0 for both MZ and DZ twins.
- Unique environmental factors (E) are influences experienced by only one twin, such as unique friends or experiences. There is zero correlation in both MZ and DZ twins.
Actor-observer effect: The tendency to attribute our own behavior to situational causes and the behavior of others to personal factors.
Akaike Information Criteria:A measure of model fit that adjusts for the simplicity of the model. The lower the score, the better the model. (Adapted from Kline, 2005: Principles and Practice of Structural Equation Modeling.).
Analysis of variance (ANOVA): ANOVA is a statistical model for analyzing the differences between group means in comparison to differences within group means. ANOVA is helpful for comparing three or more groups (e.g., Treatment Group A, Treatment Group B, Control Group) on a particular dependent measure (e.g., number of days of heavy drinking) and determining whether the differences in group means reaches statistical significance. A mixed model ANOVA is used when the researcher wants to make at least one comparison between different groups and at least one comparison of repeated measures. For instance, a researcher would use a mixed model ANOVA to test differences in Treatment Group A, Treatment Group B, and the Control Condition with measurements taken before treatment and after treatment.
Associative Learning: Associative learning occurs when a person forms a new association. Sometimes we form associations between two pieces of information we take in—for instance, we associate a person’s face with his name. Other times, we form associations between our actions and outcomes. For instance, we might learn to associate eating spoiled food with becoming ill. If we have repeated experiences with an association, that association is strengthened through a process called consolidation. Eventually, the association is stored in long-term memory. Scientists believe that a part of the brain called the medial temporal lobe is essential to associative learning. For more information, see the American Psychological Association’s page on this concept.
Attributions: Inferences people make about the causes of behavior or events; explanations for why behaviors or events occur.
Attribution theory: The study of how people understand and explain the causes of behavior.
Alcohol Use Disorders Identification Test (AUDIT): A standardized measure that ses a 10-point scale to screen and identify people at risk of alcohol problems, Saunders, Aasland, Babor, De La Fuente, & Grant, 1993
Berkson’s bias: One type of selection bias. Research studies sampling people in treatment might find differences between treatment recipients and people not in treatment, and these differences can alter the results. For example, recovering heroin users are dispersed throughout communities and therefore difficult to find to enroll in a research study. Accessing a methadone treatment program, which treats heroin users, would yield more potential study participants in a short time. However, those receiving treatment might have more health problems or psychological disorders than participants not receiving treatment. If the study only samples treatment recipients, its results might only apply to this subset of the recovering heroin user population.
"Big Five" Personality Traits: The "Big Five" is a theory of personality. Many researchers believe that the so-called “Big Five” personality traits together describe the full range of human personality in a largely non-overlapping manner. The “Big Five” factors are extraversion (outgoing/energetic vs. solitary/reserved), conscientiousness (efficient/organized vs. easy-going/careless), agreeableness (friendly/compassionate vs. cold/unkind), neuroticism (sensitive/nervous vs. secure/confident), and openness to experience (inventive/curious vs. consistent/cautious). Twin studies suggest that the Big Five result from a mixture of genetics and environmental factors.
Binge drinking: According to the National Institute of Health, binge drinking is “a pattern of heavy drinking that occurs during an extended period of time set aside for drinking. Has been described as 5/4 binge drinking: five or more drinks in a row on a single occasion for a man or four or more drinks for a woman” (accessed September 12, 2008 from http://science.education.nih.gov/supplements/nih3/alcohol/other/glossary.htm).
Bivariate association: A bivariate assoication occurs when two variables share a relationship such that when one changes, we expect the other to change. (adapted from Healey, Joseph (2012). Statistics: A Tool for Social Research. 9th edition).
Bogus pipeline: An experimental paradigm where researchers claim to have the ability to measure accurately or to confirm hidden information about the participants. This approach is based on the idea that participants who believe researchers will be able to expose any untruthful responses are more likely to be truthful in their answers and open about socially undesirable attitudes.
Bradford Hill’s (1965) Criteria for Causality:These conditions are referred to in order to determine the strength of evidence supporting a causal relationship between two variables. Observations must first show a perceptible association between these variables. If the following criteria are present, it may be possible for researchers to claim causality; however, if they are not present, this does not necessarily rule out causality.
- Strength- The greater the degree of association between one event and another, the greater likelihood that there is a causal relationship.
- Consistency- If the findings are largely consistent with the results of similar research, there is more evidence of causality.
- Specificity- Researchers must be very specific when claiming what conditions produced the effect, or they may otherwise falsely generalize findings to conditions that do not produce the effect.
- Temporality- One event must have taken place before the other to claim that the first was the cause. Longitudinal studies are helpful in establishing temporality because they follow the same participants over time, taking measures of a potential cause and later taking measures of a potential effect.
- Biological gradient- The presumed effect should occur more often when greater exposed to the cause. For example, people who have more exposure to cigarette smoking are more likely to develop lung cancer.
- Plausibility- If a logical and plausible explanation for why the first event caused the second event exists, it provides more support for causality. This is largely dependent on the scientific knowledge available on the subject. For instance, it is plausible that smoking causes lung cancer because cigarettes contain harmful chemicals.
- Coherence- Cause and effect relationships discovered in the laboratory should also naturally occur without scientific interference.
- Experiment- When researchers manipulate one variable and control all of the other variables that could plausibly exert an influence, they have solid evidence of causality.
- Analogy- When one variable appears to be causing the effect, a researcher should investigate and rule out other possible explanations.
Note: None of the preceding criteria can provide certainty of causality, though they serve as a gauge for the likelihood of a causal relationship.
Bradford Hill, A. (1965). The environment and disease: Association or causation, Proceedings of the Royal Society of Medicine, 295-300.
CAGE: The CAGE assesses alcohol problems using four questions. The four questions form an acronym of the assessment's name:
Cut down: Have you ever felt you ought to cut down on your drinking?
Annoyed: Have people annoyed you by criticizing your drinking?
Guilty: Have you ever felt bad or guilty about your drinking?
Eye-opener: Have you ever had a drink first thing in the morning (eye-opener) to steady your hands?
Ewing, J. A. (1984). Detecting alcoholism: The CAGE questionnaire. Journal of the American Medical Association, 252(14), 1905-1907.
CANTAB Spatial Working Memory Test: Participants open boxes on a computer screen to try to locate tokens. Tokens can appear in empty boxes that have been previously opened, but cannot reappear in boxes in which a token has already been found.
Case control study: The epidemiological method of investigation of a condition of interest that compares the history of exposure to specified risks among cases to exposure among persons who resemble the cases in other respects but do not have the condition of interest, the controls. This method is particularly useful in the study of rare conditions.
Causality: Causality when one event causes another event, though it is difficult to prove. Researchers should be cautious making claims about causality and readers should be cautious interpreting claims of causality, as it cannot always be determined with absolute certainty that one event caused an event. If the variables are measured at the same time, as in correlational studies, a third, unmeasured variable could have produced both the “cause” and the “effect;” or, the causal relationship could actually go in reverse—the “effect” might have produced the “cause.” However, when many of Bradford Hill’s (1965) Criteria for Causality are met, this provides greater support for a causal relationship between two events.
Censoring: We are accustomed to use "censored" to indicate that something has been withheld for some reason. The term is used in working with time-related data such as survival analysis (indicate this to be a glossary term) to indicate missing observations rather than withheld information. Although the participant in a study is still available to contribute information. the time when an event occurred might be missing for basically two reasons. First, the data collection ended at a certain point in time. The analysis can estimate what contributed to the event occurring up to that time, but only for those cases in which the event occurred. For the cases in which the event had not occurred by the end of the study period, there is no information about what might contribute to the event at some later date. These data are "censored" or "missing" because the study ended. This type of study is usually referred to as including "suspended" data.
A second common type of censored data is the result of measuring the presence of an event not continually but at some interval. Studies often obtain information over time by repeating measures at intervals. If the event is observed at a scheduled point in time it may not be possible to determine when since the previous measurement period the event occurred. The information about exactly when the event occurred is missing and usually referred to as "interval censored" data.
Guy, W. ECDEU assessment manual for psychopharmacology. US Department of Health, Education and Welfare Publication (ADM) 76-338. Rockville, MD, National Institute of Mental Health, 1976.
Classical Conditioning: Occurs when an individual learns to associate a neutral stimulus, such as the sound of a tone, with a reflexive response, such as an eyeblink, by way of a third stimulus. For example, in the eyeblink classical conditioning task, the researchers repeatedly pair a tone (the Conditioned Stimulus) with a mild puff of air to the eye (the Unconditioned Stimulus). The air puff naturally produces an eyeblink (Unconditioned Response). Over time, participants learn to associate the tone with blinking and will blink after hearing the tone. In other words, they exhibit a Conditioned Response (CR). Researchers often measure CRs in order to understand whether a study participant has demonstrated classical conditioning, a form of associative learning. Classical conditioning is sometimes called Pavlovian conditioning. Pavlov made a famous observation that after repeatedly hearing a bell sound before getting food, dogs will begin to salivate simply upon hearing the bell.
Clinical significance vs. statistical significance: Statistical significance means that there are differences between groups that are large enough to rule out chance occurrence. However, in absolute terms, the basic proportions and differences might be small. In this case, the difference is not clinically significant. For example, people who received a gambling intervention scored significantly lower on a measure of gambling problems than people who did not receive the intervention. However, if the difference in scores was only one point out of 100, then the treatment would not have a clinically significant impact on the participants.
Closed (close-ended) question: A type of question asking the respondent to choose an answer from a specified list.
Cluster analysis: Cluster analysis is an exploratory technique that groups cases (in many instances, participants) together based on similar characteristics. The technique includes information about the optimal number of clusters within a sample and the similarity within and between clusters.
Cluster sampling: The use of random sampling methods to select clusters in two or more stages, for example a sample of individuals from districts, counties, or households, from a UNIVERSE of all districts, counties, or households that are available for study.
Cognitive Behavioral Therapy (CBT): A type of psychotherapeutic approach addressing maladaptive emotions, behaviors, and cognitive processes. The goal is to systematically challenge the dysfunctional emotions, behaviors, and cognitions while learning coping strategies which will be used later in life even following the conclusion of therapy. This technique is used for a number of mental health disorders such as depression, anxiety, substance abuse.
Cognitive disinhibition: A personality trait characterized by impulsive behavior, sensation seeking, and disregard for consequences. Cognitive disinhibition is independent of behavioral disinhibition – the ability to inhibit responses.
Cognitive epidemiology: A field of research that connects people’s cognitive functioning to their morbidity and mortality.
Deary, Batty. (2007). Cognitive Epidemiology. Journal of Epidemiological Community Health, 61(5), 378-384
Cognitive restructuring: A method of treatment in which participants are taught to recognize and correct distorted cognitions that contribute to their problem.
Cohen's kappa coefficient: A statistical measure of agreement between raters or on a measure conducted at different points in time. The values indicate < 0 no agreement; 0–.20 - slight, .21–.40 - fair, .41–.60 - moderate, .61–.80 - substantial, and .81–1 - almost perfect agreement.
Confidence interval: A confidence interval is a numerical range that the researcher believes covers the true value of the population parameter . Most commonly, researchers report a 95% confidence interval (95% CI). This means that if we were to repeatedly make new point estimates using the same procedure (i.e., by drawing new samples, conducting surveys, and calculating point estimates and confidence intervals), the confidence intervals would contain the true value within the population 95% of the time. The CI is often reported in square brackets following the point estimate. There are some short-hands for interpreting CIs. For logistic regression, confidence intervals that do not cross 1 (e.g., 1.50 to 5.53) imply that there is a statistically significant difference in the odds of a given event, compared to a reference group. In addition, the larger the confidence interval for a particular estimate, the more caution is required in concluding that the point estimate represents the true population parameter.
Confirmation bias: The tendency to seek, interpret, and create information that verifies existing beliefs.
Confirmatory factor analysis:A type of factor analysis where the researcher begins with a preconceived notion or hypothesis of which factors influence which variables and then measures how well the results and data fit the hypothesis.
Confounding: This occurs when the apparent relationship between a predictor and outcome is influenced by other factors, some of which might be unmeasured or unrealized. Scientists can use study designs and analytic strategies to control for confounding in their research.
Convenience sampling: A study design where those administering the survey passively advertise it and depend on participants to come to them. The primary issue with convenience sampling is that the results tend to be biased towards motivated respondents and strong opinions.
Control: Researchers design experiments to eliminate (control for) the effect of non-experimental variables on the results. For example, a study may compare the outcomes of two groups that differ only because one received the experimental condition and the other did not (the control group). Medication trials are often “placebo controlled” by using a control group that is treated similarly in all ways except they receive a non-active medication (placebo) instead of the active medication being studied.
Counterbalanced design:Counterbalancing is a method for controlling order effects in a repeated measures (or within-subjects) design (i.e., a design in which the same participants are exposed to multiple experimental conditions). In a counterbalanced design, different participants receive are exposed to the same experimental conditions but in different orders. For example, one group of participants might complete a task under a placebo condition and then under an alcohol condition. Another group might complete the task first under an alcohol condition and then under a placebo condition.
DeFries-Fulker regression model: Used to estimate genetic heritability of and shared environmental influences on a behavior. The model uses one sibling’s behavior (S1) and the degree of genetic relation between siblings to predict the other sibling’s behavior (S2). The unstandardized coefficient associated with S1 provides the amount of variance in S2 accounted for by shared environment, and the unstandardized coefficient associated with the interaction between S1 and sibling relationship provides the amount of variance in S2 accounted for by heritability.
Delphi method: A systematic process by which a panel of experts answer questionnaires about a topic, the responses are analyzed by an independent source and the results given back to panelists. The panelists answer the questionnaire again taking into consideration the responses of the panel. The process repeats until the panel comes to consensus on the topic.
Diverting: People who divert medication take pharmaceuticals without a documented prescription. These people commonly receive medication from friends and family members with a prescription, purchase it on the street, or find some other outlet.
Dissociation: Dissociation refers to a state of feeling detached or removed from one’s surroundings. Dissociation can take many different forms, from daydreaming to believing the world is unreal to severe psychological disorders.
Doctor shopping: People who doctor shop have multiple clinicians who prescribe controlled substances for them. Often each clinician is unaware that their patient receives additional prescriptions from other clinicians.
Dorsolateral: Of, relating to, or involving both the back and the sides.
Ecological momentary assessment: A technique used to gather data on momentary states of individuals who are in their “real-world” environments at the time of assessment, often using easily transportable tools such as handheld computers or mobile phones. Because assessments are completed as psychological states or behaviors occur, rather than at some later point, there is less potential for recall biases.
Stone, A. A., Shiffman, S. S., & DeVries, M. W. (1999). Ecological momentary assessment. In D. Kahneman, E. Diener, & N. Schwartz (Eds.), Well-being: The foundations of hedonic psychology (pp. 26–39). New York, NY: Russell Sage Foundation.
Effect size: The magnitude of an experimental effect, often calculated as the mean difference between two groups divided by the standard deviation of the difference. Unlike statistical significance, effect size is not dependent on sample size. Effect size is important to consider because statistically significant effects might not be practically or clinically meaningful (for example, because they were derived from very large samples), and effects that are not statically significant might be still be strong enough to be practically or clinically meaningful.
Ego-defensive bias: The tendency to take more credit for success or good behavior than for failure or bad behavior.
Enzyme: Any of numerous complex proteins that are produced by living cells and catalyze specific biochemical reactions.
Excitatory: Promoting the action of (an)other neuron(s); increasing the probability that a postsynaptic neuron will fire.
External validity:The degree of generalizability of the obtained results beyond the given setting and a sample group of participants. Experiments are said to possess external validity if their results may be used to make predictions about entire population in different circumstances. Researchers can increase external validity by selecting a representable sample of participants and creating conditions that are similar to real life. Frequently, strict control for confounding variables (i.e., high internal validity) prevents achieving high external validity due to creating too artificial a lab setting.
fMRI: (Functional Magnetic Resonance Imaging) An imaging of the brain using MRI technology, fMRI scan uses radio waves and a strong magnetic field to measure the levels of blood in specific parts of the brain.
Factor Analysis: Data reduction technique that analyzes the relationships among a set of variables in order to identify clusters that vary together as a group. Each cluster consists of several measures of the same underlying dimension (or factor) and the common information can be largely or entirely explained by a single composite of the several variables in the cluster.
False consensus effect: The tendency to overestimate the consensus of our own opinions, attributes, and behaviors.
Fisher’s exact test: This statistical test measures the likelihood that the proportion of cases distributed across two cells (e.g., Yes or No) in two independent groups (A and B) could have happened by chance if the two groups were sampled from the same population. Fisher's exact test computes the probability (p) of obtaining the observed arrangement of “yeses” as compared to obtaining each of all the possible arrangements of “yeses” in the two groups. In the case below (see figure 1), the probability of observing a difference in proportions as great or greater than 2 of 8 versus 3 of 7 is .427. Fisher’s test is an “exact” test because the actual number of ways in which the “yeses” and “nos” can be arranged is an exact calculation (particularly with small samples, the most common use of the test).
Figure 1 - Fisher’s Exact Test
Fundamental attribution error: The tendency to focus on the role of personal causes and underestimate the impact of situations on other people’s behavior.
Gambling levels: Public health systems often assign levels to phenomena of interest; For example first and second degree burns. The gambling level system is a public health tool that classifies people along a gambling involvement continuum. This continuum often begins with healthy gambling behavior that does not produce any adverse reactions. On the other end of the continuum is gambling behavior with the most serious adverse consequences.
Level 0: People who do not gamble
Level 1: People who gamble with no adverse consequences
Level 2: People who gamble with any of a wide-range of adverse reactions or consequences that do not qualify the gambler for the most serious form of gambling disorder; level 2 gamblers represent people who may be moving in either of two directions: toward an increasingly disordered state (level 3) or toward a healthier level of gambling (level 1)
Level 3: People who gamble with adverse consequences that are sufficiently serious and co-occurring so as to meet diagnostic criteria
Level 4: People who seek help for gambling problems regardless of the extent of their problems or distress
Generalized estimating equations: Generalized estimating equations are statistical tools that allow for the prediction of an outcome based on both fixed factors and an estimation of random factors. Generalized estimating equations differ from regression in that they account partially for random variables that can affect the outcome, for example, changes in weather or mood.
Gold standard: An independent criterion that can validate the presence or absence of a disordered state.
Growth mixture modeling: A statistical technique used to model individuals' growth over time on a given variable or set of variables classify individuals according to those trajectories. Growth mixture modeling is one of several person-centered approaches used to understand individual trajectories. Person-centered approaches account for individual variation within samples and groups. This is in contrast to variable-centered approaches, such as regression, which highlight averages and relationships between variables. For more information, see Jung & Wickrama (2008). An introduction to latent class growth analysis and growth mixture modeling, Social and Personality Psychology Compass, 2/1, 302-317. http://www.statmodel.com/download/JungWickramaLCGALGMM.pdf.
Hazard rate: Formally speaking, hazard rate is the ratio of the probability density function to the survival function. This site provides a concrete example: When you are born, you have a certain probability of dying at any age; that’s the probability density. For example, a girl born today has, say, a 1% chance of dying at 80 years. However, as she survives for a while, her probabilities of dying on any given day keep changing. For example, when she is 79 years old, she might have a 5% chance of dying at 80 years.
Hazard ratio: Hazard ratio is an important outcome measure, often used in survival (or time-to-event) analysis. In this kind of analysis, the researcher studies the time that it takes for a particular event to occur. The researcher calculates two "hazards." A hazard is the probability that an individual at any given time is going to experience an event (e.g., getting sick, attempting suicide, dying). A high hazard suggests a high probability that the event in question will happen within a given timeframe. The researcher calculates hazards for different kinds of situations -- for example, in a treatment condition versus a control condition--and calculates the ratio. The resulting ratio describes the relative likelihood that an outcome will occur within a given group, within a given timeframe.
Hookah: A hookah (also known as shisha, or narghila) is a waterpipe used to smoke primarily flavored tobacco, although it can be used for marijuana and other substances. A typical hookah consists of a 1-3 foot base containing the tobacco, coals, and water, as well as several hoses to inhale the smoke through. Originating from the Middle East, the hookah is becoming increasingly popular as a social smoking behavior in the United States. A typical hookah session might last an hour or even longer.
Hole cards: In poker, cards that are dealt to a player face down. Contrast with board cards in stud poker, which are dealt face up, or community cards in hold’em poker, which are placed in the middle of the table and shared by all the players.
Humanities: Those branches of knowledge, such as philosophy, literature, and art, that are concerned with human thought and culture; the liberal arts. (Source: The American Heritage® Dictionary of the English Language, Fourth Edition. Copyright © 2000 by Houghton Mifflin Company.)
Idiographic Approach: A method of inquiry that focuses on the individual rather than the group. It is used to gather information about the subjective experience of an individual in order to determine what makes the individual unique. By contrast, a nomothetic research approach attempts to establish general laws that are applicable across individuals. Both approaches have both strengths and weaknesses.
Implicit Association Test (IAT): A computerized reaction-time task designed to measure the strength of association between target concepts (e.g., flowers and insects) and attribute concepts (e.g., pleasant and unpleasant). Participants categorize stimuli as quickly as possible by pressing a computer key. In critical blocks, participants use the same computer key to categorize both a target concept and an attribute concept. The IAT measures the milliseconds required to sort stimuli. The strength of the association between targets and attributes , operationalized as the quickness with which participants respond, is referred to as an “implicit association.” You can take a demo test online.
Incidence Incidence is an important concept in public health research. It refers to the proportion of people in the population who have newly developed a given medical condition. It differs from prevalence, which refers to the total proportion of people who have a given medical condition.
Incident rate ratio: The ratio of two incidence rates. The incidence rate among the exposed proportion of the population, divided by the incidence rate in the unexposed portion of the population, gives a relative measure of the effect of a given exposure. Incident rate ratios that are close to 1.0 indicates little effect of a given exposure on health outcomes. http://www.hsph.harvard.edu/thegeocodingproject/webpage/monograph/glossary.htm
Indirect effect:A measure of the extent to which a given variable mediates the relationship between an initial variable and an outcome variable. Statistically, it is defined as the reduction of the effect of the initial variable on the outcome variable.
Information Sampling Test: Paticipants open as many boxes as they choose in a 25 box grid. Each box reveals one of two colors. Participants must decide which color is in the majority; they can make this decision at any point and receive points for correct answers. In one condition, they receive the same points regardless of how many boxes they open; in the other there points decrease the more boxes they open.
Inhibitory: Modifying, inhibiting, or suppressing the action of (an)other neuron(s); reducing the probability that a post-synaptic neuron will fire.
In situ hybridization: In situ hybridization is a method of locating a specific mRNA sequence by introducing its compliment into the tissue. The compliment is usually attached to a label of some sort, fluorescent or otherwise, so that when these two complimentary halves find one another and bind together, the researcher can easily locate the mRNA he or she is interested in.
Intent to treat analysis: A procedure in the conduct and analysis of randomized trials. All patients allocated to each arm of the treatment regimen are analyzed together as representing that treatment arm, whether or not they received or completed the prescribed regimen. Failure to follow this step defeats the main purpose of random allocation and can invalidate the results.
Internal validity: The degree of validity of causal inferences in scientific studies. Experiments are said to possess internal validity if they demonstrate a causal relation between two variables (a cause and an effect). Researchers can increase internal validity by carefully controlling for confounding variables that could provide alternative explanations for observed effects.
Intraclass correlation: A type of correlation used when members of a pair can not be assigned to distinct variables. For example, data for wives and husbands could be organized by the two variables “wife data” and “husband data,” but data for two same sex siblings could not be assigned to variables in any non-arbitrary way. If a regular Pearson correlation were used in the latter case, it would change depending on which sibling was labeled sibling 1 or sibling 2. The intraclass correlation does not rely on the order of the variables, and is thus more appropriate in the latter case. Also see: http://www.uvm.edu/~dhowell/StatPages/More_Stuff/icc/icc.html for a good reference.
Intrinsic and Extrinsic Motivation: Ryan and Deci's Self-Determination Theory defines intrinsic motivation as motivation that is generated internally, such as interest or curiosity; and extrinsic motivation as motivation that is generated externally by an outcome outside of the activity itself, such as a money reward or an evaluation.
Intrinsic and Extrinsic Goals: Intrinsic goals are goals that are self-satisfying independent of external factors. Extrinsic goals are goals that focus on external consequences outside of the task alone.
Deci, E. L. (1971). Effects of externally mediated rewards on intrinsic motivation. Journal of Personality and Social Psychology, 18(1), 105-115.
Vansteenkiste, M., Lens, W., & Deci, E. L. (2006). Intrinsic versus extrinsic goal contents in Self-Determination Theory: Another look at the quality of academic motivation. Educational Psychologist, 41(1), 19-31.
Iowa Gambling Task (IGT): A task designed to simulate decision making, particularly the kind of decision-making often involved in gambling. In the task, participants are presented with four decks of cards. (These might be actual decks of cards, or virtual cards on a computer screen.) The participant is told that the game requires them to choose cards, one at a time, from any of the four decks. The participant receives a monetary reward after turning over each card. The reward amount varies from card to card. Some cards, however, come with both a reward and a penalty. The decks differ in how much profit they ultimately provide. Some decks (i.e., "disadvantageous decks") are programmed such that they pay larger rewards throughout the entire task and, over time, produce larger and larger penalties. Other decks (i.e., "advantageous decks") pay smaller rewards throughout the task but also produce smaller penalties. Over time, it becomes more profitable to select from the latter decks. Researchers typically measure how long it takes participants to learn to choose from the advantageous decks. People with certain types of mental health conditions or brain abnormalities tend to show poor performance on this task, indicating a reduced sensitivity to future consequences.
Reference: Bechara, A., Damásio, A. R., Damásio, H., Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 50 (1-3): 7–15.
Item Response Theory: The study of test and item scores based on assumptions concerning the mathematical relationship between abilities (or other hypothesized traits) and item responses. Other names and subsets include Item Characteristic Curve Theory, Latent Trait Theory, Rasch Model, 2PL Model, 3PL model and the Birnbaum model.
Baker, Frank (2001). The Basics of Item Response Theory. ERIC Clearinghouse on Assessment and Evaluation, University of Maryland, College Park, MD.
Learned helplessness: The phenomenon in which experience with an uncontrollable event creates passive behavior toward a subsequent threat to well-being.
Likert scale: A survey item where the participant is shown a statement and then asked about the extent to which the he or she disagrees or agrees with the statement. There is a short, ordered list of between 3 and 7 possible responses, ranging from strong disagreement to strong agreement. For example, there may be five possible answers: strongly disapprove, disapprove, neutral, approve, strongly approve. Researchers adapt this style for questions about a range of concepts.
Linear regression: A type of regression analysis whereby the researcher attempts to model the relationship between a dependent variable (y) and independent variable(s) (x) fitting a linear equation to the data. Linear Regression estimates the coefficients of the linear equation that best predict the value of the dependent variable. For example, a researcher might try to predict a salesperson's total yearly sales (the dependent variable) from independent variables such as age, education, and years of experience.
Logistic regression: A method used to predict the presence or absence of a characteristic or outcome based on values of a set of predictor variables. It is similar to a linear regression model but is suited to models where the dependent variable is dichotomous. Logistic regression coefficients can be used to estimate odds ratios for each of the independent variables in the model. - Cited from the SPSS Help Library
Long Term Potentiation (LTP): LTP is a commonly used measure of addiction in neuroscience. Neurons in the nucleus accumbens (part of the reward center) of the brain are stimulated with an electrode and the responses are recorded. Substances like cocaine can cause neurons to respond more strongly to an identical stimulus than untreated neurons (potentiation). This response can be maintained for relatively long periods, even after the withdrawal of the substance (long term).
Matched Pairs t-test: Procedure for testing whether one population or experimental group is significantly different from another. In this experimental design, members of one group are paired with similar members from another. For example, to test whether drinking a sports drink helps athletes recover faster than drinking water, a researcher might match each participant with another with the same gender and similar height and weight, and then give one of them the sports drink and the other water.
Mean: One of several statistics that describe the central tendency in a group. Other measures of central tendency include the median and the mode. The mean is simply the average of all the measures; the average is the sum of all measures divided by the number of these measures. The presence of a few extreme values can result in a mean that is not a good description of the central tendency of the group as a whole.
Median: One of several statistics that describe the central tendency in a group. The median is the middle value of all measures in a group. Researchers can create a median split by dividing participants into two groups: one group with higher values will be above the middle value, and one group with lower values will be below the middle value. This procedure turns a continuous variable into a categorical variable, resulting in the loss of information about graduations between participants.
Mediator: A mediator is a variable that helps account for or explain a relationship between two other variables. For example, the effect of a treatment (variable A) on alcohol consumption (variable B) might be because of improved stress-management skills (moderator variable). To qualify as a mediator, the variable must replace (or strongly diminish) the direct relationship between variable A and B. It is different from a moderator in that moderators help change a relationship, while mediators explain a relationship
Middle frontal, inferior frontal, orbital gyrus: Areas of the frontal lobe of the brain where decision making is thought to be localized.
Mode: One of several statistics that describe the central tendency in a group. Other measures of central tendency include the mean and the median. The mode is the most frequent value observed in the group. Generally, the mode is used to describe the group tendency when there are only a few possible values.
Moderator: A moderator is a variable that changes the strength or direction of a relationship between two other variables. For example, the relationship between age (variable A) and height (variable B) might be different for men and women (moderator). It is different from a mediator in that mediators help explain a relationship, while moderators change a relationship
Motivational Interviewing (MI): MI underscores personal motivations and benefits for change. It helps to establish a commitment to reduce or stop problem behavior and promotes the participant’s self-efficacy.
Multi-level Models: Multi-level models, also known as mixed-effects models, are mathematical/statistical models used to analyze nested or hierarchical data.
Multistage sampling: A form of cluster sampling that selects among clusters defined at each stage. Rather than sampling from within all members making up a cluster, only some members of the cluster are selected at each stage. For example, a cluster sampling of households in a city is facilitated by sampling among households only in selected city blocks that are in selected postal areas that are in selected voting districts.
Nested Survey Design: In a nested survey design, levels of one factor are completely subsumed or nested inside the levels of another. For example, consider a survey of elementary school students in a district with multiple elementary schools. If we record the students’ schools and homerooms, then the variable “homeroom” will be nested within the variable “school”.”
Neurotransmitter: A chemical in the body which moves between neurons and communicates chemical messages such as pain, pleasure, emotion, and touch sensation. Some common neurotransmitters are serotonin, dopamine, and norepinephrine.
Nonlinear regression: A method of finding a nonlinear model of the relationship between the dependent variable and a set of independent variables. Unlike traditional linear regression, which is restricted to estimating linear models, nonlinear regression can estimate models with arbitrary relationships between independent and dependent variables. This is accomplished using iterative estimation algorithms
Normal distribution: A continuous probability distribution, which appears bell shaped, is symmetrical about the mean and has the most probable scores concentrated around the mean. Often times when we speak of the normal distribution we use as an example the Standard Normal Distribution, which has a mean of 0 and variance of 1. Also called Gaussian distribution.
Nucleus accumbens: Area of the brain’s reward pathway associated with feelings of pleasure.
H0: Null hypothesis: The purpose of most statistical tests is to determine the likelihood that the observed results could have happened by chance given the hypothesis to be tested. The “null” version of a hypothesis (usually a difference in outcome due to difference in conditions) expects no difference. If the statistical test indicates that, given the hypothesis, the observed difference was not likely to have happened by chance, we can reject the hypothesis. Rejecting the null hypothesis rejects the expectation of no effect and encourages acceptance of an effect.
Observational studies: Studies and surveys where data is collected from participants without the researchers affecting the variables and factors directly. Observational studies are usually contrasted with experiments, where researchers assign factor levels (e.g., drug dosages) to participants.
Observer bias: Systematic difference between a true value and that actually reported by the observer, due to a failure of the observer to measure or identify a phenomenon accurately.
Odds: The ratio of the number of specific events to the number of other events (i.e., odds = # specific / # other). The odds of an event differ from the probability of an event in that the denominator for calculating odds is the non-events and the denominator for probability is the total of all events.
Odds ratio: An odds ratio is a measure of effect size. It indicates how likely a group is to report a specific outcome, compared to a control group. An odds ratio greater than 1 indicates the group is more likely to report that outcome, while an odds ratio less than 1 indicates the group is less likely to report the outcome. Odds ratios are typically reported with the 95% confidence interval.
For example, if the results for a study show that Group A has an odds ratio of 2.5, and Group B has an odds ratio of 0.75, then people in group A were two and a half times more likely to report the outcome, compared to the control group. Participants in group B reported the outcome 3/4 as often as those in the control group.
Open Label Investigation: An open-label trial is a type of clinical trial in which both the researchers and participants know what is being administered (e.g., treatment A versus treatment B; treatment A versus placebo). This contrasts with trials in which participants (and potentially the researchers as well) are not aware of which formulation patients are receiving.
Open (open-ended) question: This type of question asks the respondent to answer in sentence or phrase form.
Optical density: Optical density is a measure of light transmittance through a given area of a developed photo image, and can be expressed as the number of dark spots in that area. This number is usually given relative to light transmittance through other areas of the image.
Orbital: Pertaining to the region above the eyes.
P-value: A term for the probability that an event will occur. In common research terminology, the p-value of an inferential statistic is the probability that an observed outcome—and more extreme outcomes—could have happened by chance under the null hypothesis that there was no effect. Small p-values such as five times in a hundred (p = .05) are used benchmarks to distinguish effects that could have happened by chance from effects unlikely to happen by chance.
PGSI: See Problem Gambling Severity Index.
Phase II: Repeating and continuing safety assessments started in Phase I and testing how well the drug works on large groups of individuals. Phase II is further separated into a and b.
Phase IIa: Specifically tests how large or small drug dosages should be.
Phase IIb: Tests the efficacy of specific doses.
Phase III: Randomized controlled testing of a drug on a larger, hundreds to thousands, number of individuals (the number depends on the disease being researched) to test a drug’s safety, efficacy, and possible additional uses on a larger scale. Drugs and treatments in the Phase III stage can be sold on the market with FDA approval unless harmful effects are discovered or reported.
Phase IV: Drugs in this phase have qualified for sale on the market after all appropriate forms of testing. However, drugs in Phase IV are still tested periodically to locate any long-terms adverse effects in patients, to remain competitive with newer drugs, and to make sure that the drug does not negatively interact with other drugs, foods, or conditions.
Population: The entire collection of observations to which the study results generalize. The study sampling procedures will define the accessible study population.
Population parameter and point estimate: Population parameters are characteristics of populations of interest. Researchers often want to estimate population parameters. For example, a researcher studying American teenagers might want to know the percentage of teenagers who have used marijuana in the past month. Usually, obtaining the value of the population parameter is impractical—in this case, because it would require asking every American teenager about marijuana use. Instead, researchers draw a sample from the population. They calculate a point estimate, a sample statistic calculated solely from the sample data. However, a sample almost never exactly mirrors the original population, and a point estimate is almost never the exact value of the unknown parameter. Researchers use confidence intervals to indicate how much caution is needed before concluding that the point estimate is a true reflection of the population parameter.
Post hoc tests: Post-hoc tests are usually focused comparisons (i.e., comparing one group or condition against another group or condition) that researchers conduct after they have already determined that there are at least some differences among a larger set of groups or conditions.
Primary prevention strategy: Primary prevention aims to reduce the incidence of disease.
µ (Population mean): By convention, lower case Greek letters are often used to indicate population parameters (the letter “mu” for mean) and upper case English letters indicate the sample parameters (capital M for mean).
Prefrontal cortex: The area of the cortex most advanced in humans compared with other animals. This area primarily provides executive functions over other areas of the brain.
Probability: The ratio of the number of specific events to the number of all possible events (i.e., p = # specific / # specific + # other). Probabilities can range from zero (i.e., impossible) to one (i.e., every time). The probability of an event differs from the odds of an event in that the denominator for calculating odds is the non-events and the denominator for probability is the total of all events.
Probit regression models: Similar to logistic models but use a log-normal transformation (the probit transformation) of the dependent variable. Where logit and logistic regression are appropriate when the categories of the dependent are equal or well dispersed, probit may be recommended when the middle categories have greater frequencies than the high and low tail categories, or with binomial dependents when an underlying normal distribution is assumed. As a practical matter, probit and logistic models yield the same substantive conclusions for the same data the great majority of the time.
Q statistic: Tests for the homogeneity of effects in a set of several studies. A significant Q statistic indicates significant heterogeneity among studies beyond the variability within studies. Q is not a measure of the degree of heterogeneity. (Source: Huedo-Medina, T.B., Sanchez-Meca, J., Marın-Martınez, F., and Botella, J. (2006) Assessing Heterogeneity in Meta-Analysis: Q Statistic or I2 Index?, Psychological Methods, 11(2) 193–206.)
Proxy: A variable that stands in for another unmeasured variable. Often, the association between a proxy and an outcome variable can appear causal when actually the proxy variable is associated with a third unmeasured variable that is the true causal factor.
QSU Brief: Represents Factors 1 and 2 from the full QSU, but is a 10-item questionnaire that takes less than two minutes to complete. Factor 1 is the intention and desire to smoke that is anticipated as a pleasant, enjoyable and satisfying behavior and Factor 2 is anticipation of relief from negative affect and nicotine withdrawal through smoking with an urgent desire to smoke.
Cox, L. S., Tiffany, S. T., & Christen, A. G. (2001). Evaluation of the brief questionnaire of smoking urges (QSU-brief) in laboratory and clinical settings. Nicotine & Tobacco Research, 3(1), 7-16.
Tiffany, S. T., & Drobes, D. J. (1991). The development and initial validation of a questionnaire on smoking urges. British Journal of Addiction, 86(11), 1467-1476.
Quartile: The segment of a sample representing a sequential quarter (25%) of a frequency distribution. The boundaries are at the 25th, 50th, and 75th percentiles.
Random error: A wrong result due to chance. Unknown sources of variation are equally likely to distort the result in any direction.
Random-effects Meta-Analysis:A model used when the results to be included in a meta-analysis are considered heterogeneous. Put simply, it is a process of adjusting the standard errors of the study-specific estimates to include the variation among the effects observed in the various studies. The simplest and most common form is the DerSimonian and Laird model.
Randomized controlled trial: A randomized controlled trial is a rigorous scientific experiment where study subjects are randomly allocated into groups to receive or not to receive the medication/intervention under investigation.
Regression Analysis: A statistical process used by researchers to estimate the relationship between a dependent variable (Y) and one or more independent variables (X). It can describe how changes in the level of Y relate to changes in the level of X. Although researchers often use causal terminology (i.e., "the independent variable, parental drinking, predicted the dependent variable, adolescent drinking") such causal relationships can only be inferred in specific circumstances, and not when researchers take measures of X and Y at the same time.
Regression to the mean: The tendency of extreme values to become less extreme (closer to the average) when they are reassessed.
Relative risk: Relative risk, or risk ratio, is the ratio of the probability of an event under one condition (or group) to the probability under another condition. Risk ratio is interpretable as a comparative quantity; the RR comparing Group A with an event probability of .6 to Group B with a probability of .3 is 2.0. The same risk ratio would result if the probabilities were ten times smaller (i.e., .06 to .03). In both comparisons, the risk ratio indicates that the event is twice more likely to occur in Group A than Group B. Depending on the choice of the event, risk ratio could be a “risk” if the event is untoward, cancer say, or it could be a “benefit” if the event was desired, not getting cancer.
Resilience: A person's ability to adapt positively (i.e. not engaging in risky behavior) in the face of adversity (i.e. prior exposure to risky behavior).
Response rate: The proportion, percentage, or fraction of individuals who reply to inquiries that have been directed to all individuals in a population, or a representative sample of them. The validity of findings or results and conclusions derived from them is dependent on the response rate. If the response rate is low, i.e., if there are large proportions of nonresponders, the validity of the results is dubious, even if impeccable sampling methods have been used, because of the magnitude of random error.
Reward pathway: The dopaminergic pathway in the brain that prompts positive feelings in response to certain behaviors or events. Areas of the brain found along the reward pathway include, among others, the ventral tegmentum, the nucleus accumbens, and the prefrontal cortex.
Risk factor: A behavior or characteristic that is shown to be associated with health conditions.
Sample: A group of observations selected from the study population. The sample may not be representative of a defined study population if the sampling strategy or recruiting procedures are flawed.
Secondary prevention strategy: Secondary prevention aims to reduce the prevalence of disease by shortening its duration.
Selection bias: Error due to systematic differences in characteristics between those who take part in a study and those who do not. Selection bias affects conclusions and generalizations that are made regarding the study sample.
Self-fulfilling prophecy: The process by which one’s expectations about a person eventually lead that person to behave in ways that confirm those expectations.
Self-limit programs: Programs offered by casinos or Internet gambling sites that impose wagering or deposit limits.
Self-report: A report about one’s thoughts or behavior provided by the subject of the research. Self-report measures have the advantage of being easily administered. Also, they allow the participant to provide his or her own perspective. On the other hand, they can be biased by the tendency to present oneself in a favorable light, or by problems with memory and introspection.
Self-selection: When eligibility for membership in sample or a study group is based on an individual characteristic, the sampling procedure is usually termed “self-selection.” For example, participants in a study comparing smokers to non-smokers would be assigned to one group or the other depending on their own decision to smoke or not. Self-selection generally precludes being able to control for the effect of other variables and observed group differences may not be due to the characteristic that determined selection.
Selection and assignment: Selection refers to the process by which a sample is recruited from a population. A random selection of observations from the population will generally yield a sample that is representative of the entire population. Self-selection and other non-random recruitments will constrain the population and therefore, the ability to generalize the study findings. Assignment refers to the process by which the sample is further divided into experimental and control groups. A random assignment of the sample will generally yield groups that are similar to each other and the difference among them will be attributable to the experimental conditions. Non-random assignment procedures limit a study’s ability to control for non-experimental characteristics.
Sensitivity: An instrument’s ability to identify people who have the characteristic under investigation;(# of people with a characteristic who are correctly identified by instrument)/(# of people actually with characteristic).
Skewed: An asymmetrical (non-normal) distribution in which values tail off more in one direction than the other.
Snowball sampling: A method of selecting for study the members of a “hidden” population (e.g., illicit drug users). Those participants initially identified are asked to name acquaintances who are added to the sample; these, in turn, are asked to name further acquaintances until enough participants are added for sufficient power for the proposed study.
Somatic marker hypothesis: Affective states associated with former experience (reward or punishment) are unconsciously encoded as somatic markers. The somatic markers are activated when we face a similar situation and help us guide decisions without any conscious effort.
Somatic states: The physiological changes that occur in the body in response to an emotionally charged stimulus that is present in the environment, or to a thought (memory) that has some emotional significance.
Lesieur HR, Blume SB (1987). The South Oaks Gambling Screen (SOGS): a new instrument for the identification of pathological gamblers. The American Journal of Psychiatry 144, 1184–8.
Specificity: An instrument’s ability to rule out people who do not have the characteristic under investigation;(# of people without characteristic who are correctly identified by instrument)/(# of people actually without characteristic).
Spurious: A situation in which an apparent relationship between two or more variables is lacking validity.
Standard error: Statistics from a sample of the population will vary from sample to sample. The standard error is a measure of the size of the variation in the sample statistic over all samples of the same size as the study sample. The greater the study sample size, the smaller the variation across samples. The size of the standard error determines the range of values that will include the population parameter. Smaller standard errors lead to more precise estimates of the characteristic in the population.
Statistical power: The probability that an experiment will find an effect if that effect exists in the total population. The number of observations in the study (i.e., sample size) sample greatly influences statistical power.
Statistical significance: Researchers often engage in significance testing in order to estimate the probability of making a Type I error—that is, claiming that a relationship exits when, in reality, no such relationship exists. For example, a researcher might make a Type I error by claiming that women underperform men on a test of negotiation skills, when no such difference really exists. Researchers engage in significance testing first by computing a statistical test (such as a t-test or ANOVA). The values of statistical tests are associated with p-values—the larger the test value, the smaller the p-value, or probability of making a Type I error. Sample size also plays an important role: a statistical test result might have a p-value of 0.05 with a sample size of 5, or a p-value of 0.025 with a sample size of 45. The tradition is to conclude that a finding is “statistically significant” if the p-value is less than 0.05. However, it is important not to define the results of a statistical test purely in terms of statistical significance. A result that is statistically significant is not necessarily practically or clinically significant. Other indicators, especially effect size, can tell us a lot about the true strength of any effect uncovered.
Stroop Effect: The Stroop effect (also, Stroop test) is based on a phenomenon that most people read a printed word more quickly and automatically than they are able to name the color in which that word is printed. When presented with a list of written color names that differ from the color ink in which they are printed (e.g. the word “blue” written in red ink), a person will be prone to produce the color name rather than the ink color; overriding this impulse requires response inhibition. Researchers use responses to the Stroop task, particularly the reaction time and error rates, to measure response inhibition in studies of various neuropsychological conditions, impulsivity, attention and other cognitive phenomena. The Stroop test is named after its developer, John Stroop.
Grant JE, Steinberg MA, Kim SW, Rounsaville BJ, Potenza MN (2004). Preliminary validity and reliability testing of a structured clinical interview for pathological gambling. Psychiatric Research 128, 79–88.
Subclinical: A condition where the disease is detectable by certain tests but signs or symptoms are not revealed.
Survival analysis: Allows us to estimate the hazard (or risk) of death, or other event of interest, past a certain time for individuals with certain characteristics. For example, researchers can ask whether men and women have different risks for developing lung cancer based on their cigarette smoking? Survival models, in which one of the variables is time, are similar to ordinary regression models. The model identifies the probability that the time of death is later than some specified time. It provides coefficients for each of the predicting variables. A positive regression coefficient for a variable means that the hazard is greater, and thus the prognosis worse. Conversely, a negative regression coefficient implies a better prognosis for patients with higher values of that variable. The choice of a specific model (e.g., the Kaplan-Meier (product-limit) Estimator, the Cox Proportional Hazards Model, the Weibull Fit.) mostly depends on the underlying assumptions about the data distribution and the extent of data censoring problems.
Systematic error: A wrong result due to bias. Sources of variation will distort the result in one direction.
Tell: Behavior or change in behavior that may give astute opponents clues about a poker player’s hole cards or future plans for the hand. Examples include facial tells (e.g., looking away when holding a strong hand), arm/hand tells (e.g., slamming a bet down to intimidate when bluffing), and betting tells (e.g., only betting big with extremely strong hands).
Tertiary prevention strategy: Aims to reduce the number and/or impact of disease complications.
Test-retest reliability: An index of the extent to which the result of a psychological test changes over time. If the construct being measured is not expected to change over time, the researcher should expect to observe high test-rest reliability. For example, a person's level of introversion should not change dramatically from one day to the next. If that person's score on a measure of introversion changes dramatically from day-to-day -- if it has low test-retest reliability--this suggests that there is something wrong with the measure.
Tetrachoric correlation: Assesses the relationship between dichotomous variables (i.e. variables with only two possible responses, such as yes or no). These correlations assume an underlying continuous distribution of dichotomous variables.
Transtheoretical Model (TTM): Assesses a person's readiness to engage in healthier behaviors, based on where they fit in a series of "stages of change." The five stages of change are "precontemplation", "contemplation", "preparation", "action", and "maintenance." It is used commonly in research pertaining to smoking cessation, weight loss, and adherence to medication. James Prochaska and colleagues developed the TTM in the early 1980s.
Tukey's Honest Significant Difference (HSD) Test: A procedure for testing whether the means of different groups are significantly different. Tukey’s HSD compares the mean of each group to the means of each of the others.
Twelve-step Program: Twelve-step programs are addiction treatment programs where people with similar substance abuse problems meet regularly to assist each other with the recovery process. Ideas behind the twelve-step process include (1) acceptance that willpower alone is insufficient to overcome the problem, (2) willingness to use the support structure of other recovering individuals, and (3) active involvement in 12-step meetings.
Two-stage cluster sample: Consists of selecting a, usually random, sample of natural clusters of people such as households or schools and then selecting a random, representative sample of one or more people within these clusters.
Validity: A psychological instrument's validity is the extent to which it accurately measures what it is designed to measure. Said another way, a test and its scoring system are considered valid if the theory and evidence support the interpretations of the scores implied by the proposed uses of the test.
Value bet: A bet where the gambler’s expected value is greater than $0. Some gamblers will analyze different betting options or combinations looking for value bets. The idea is that a gambler who makes only value bets will make money in the long run, provided that he is not ruined by any short-term losing streaks.
Ventromedial: Of, relating to, or involving both the front and the middle.
Weighted least squares regression: In standard least squares regression, the researchers assumes that the response variable has the same measurement error or variability, no matter what values the factors or independent variables take. In some applications, however, the measurement error varies with the independent variables. In these cases, weighted least squares regression accounts for this variability.
Pallanti S, DeCaria CM, Grant JE, Urpe M, Hollander E (2005). Reliability and validity of the pathological gambling adaptation of the yale-brown obsessive–compulsive scale (PG-YBOCS). Journal of Gambling Studies 21, 431–43.
Z score (http://www.musc.edu/dc/icrebm/glossary.html): Identifies the distance that an original (raw) score is from the mean. The distance is measure in standard deviation units, e.g., a Z score of 1.0 indicates that the raw score was one standard deviation greater than the mean.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Earlbaum Associates.
Last, J. (2001) A Dictionary of Epidemiology (fourth edition). Oxford: Oxford University Press.
Last, J. (2007). A dictionary of public health. Oxford: Oxford University Press.
Evidence Based Medicine Glossary (2002). Medical University of South Carolina. Retrieved November 27, 2002, from the World Wide Web: