Research in which investigators examine naturally existing public records to test a theory or hypothesis.
The most likely outcome (or the entire distribution of all possible outcomes) in a probabilistic situation that is repeated a great number of times under ideal conditions. For example, the expected value of an 80 percent chance of winning $100 is $80 because if you take this gamble a great number of times, the average amount of money you will gain will be $80 per trial of the gamble.
extreme groups approach
An approach to selecting people for inclusion in a laboratory study. In the case of this approach, experimenters recruit people to take part in a study only if such people receive extreme (i.e., very high or very low) scores on an individual difference measure of interest. See prescreening and median split.
Reasoning from the specific to the general (e.g., drawing a conclusion about human memory by noticing that a certain pattern of findings shows up in many different studies of memory). See also deduction, problem of induction.
A specific form of patching in quasi-experiments in which a researcher breaks a group of participants into one or more subgroups to test for subgroup differences that are consistent with the focal theory or with competing theories. As an example, a researcher interested in whether the trauma of being exposed to a tornado leads to anxiety disorders might identify participants who were and who were not physically injured during a tornado to see if those who were injured showed a greater subsequent increase in their level of anxiety disorders.
Questions that are worded in ways suggesting that some kinds of answers are more appropriate than others (e.g., "What do you like most about this wonderful textbook?").
An approach to selecting people for inclusion in a laboratory study. In the case of this approach, experimenters recruit people who score in either the top half or the bottom half (i.e., people above or below the median) on an individual difference measure of interest. This approach is usually inferior to the extreme groups approach. See prescreening and extreme groups approach.
method of deduction
A method of theory testing in which a specific hypothesis is derived from a broader theory and then tested empirically. This is contrasted with the method of induction because researchers first begin with a theory and then make observations. Logically, this method can be used only to falsify or qualify a theory because support for a theory in one instance never proves that the theory is true in all other instances (see the problem of induction).
method of induction
A method of scientific reasoning popularized by Francis Bacon. It refers to making many specific observations in an effort to draw general conclusions (to formulate theories or laws) about the nature of things. This is contrasted with the method of deduction because researchers first begin with observations and then try to formulate theories on the basis of these observations (but see the problem of induction).
The finding that people like the letters in their own names (1) more than they like other letters of the alphabet and (2) more than other people like these same letters (when these letters do not occur in their names).
A kind of quasi-experiment in which the researcher makes use of archival data documenting the consequences of a natural manipulation such as a natural disaster or a change in traffic laws in a particular state. The best natural experiments typically involve arbitrary or near-chance events that affect a large group of people.
A very simple research design in which all the participants are in a single group that received a natural or experimental manipulation. This is the rough equivalent of a pseudo-experiment. It is usually very difficult to draw any clear conclusions from a study making use of a one-group design.
one-group, pretest-posttest design
A quasi-experimental research design in which measures are taken from a single group of research participants both prior to and after the participants receive a natural manipulation. Because this design does not include a control group, it usually yields findings that are open to many alternative explanations.
The term Campbell and Stanley (1966) used to refer to a set of quasi-experimental findings in which a researcher has made repeated use of patching (the addition of control groups) to refine and strengthen the design. See patching.
A research method in which a researcher adds new conditions to a quasi-experiment to help establish the size of an effect, to test for the influence of conceivable confounds, or both.
A research design in which the researcher measures at least one independent variable and manipulates at least one other independent variable. Person-by-treatment designs are always factorial designs.
posttest-only design with nonequivalent group
A quasi-experimental design in which a researcher compares two similar but nonidentical groups after one and only one of the groups experiences a treatment condition.
A procedure for identifying the participants who will take part in a laboratory study (usually a person-by-treatment quasi-experiment). Researchers who use this technique give an individual difference measure to a large group of people prior to the time that they run their laboratory studies. People who receive specific scores (usually extreme scores) on the individual difference measure are then recruited to take part in the laboratory study.
pretest-posttest design with nonequivalent groups
A quasi-experimental research design in which data are collected from two presumably comparable groups of research participants both prior to and after one of the groups receives a natural manipulation. To the degree that one can safely assume that the two groups were truly similar prior to the time that they experienced different levels of the manipulation, this is a relatively strong quasi-experimental design.
problem of induction
A shortcoming of the method of induction. The problem, pointed out by David Hume, is that one can never safely infer anything with absolute certainty on the basis of induction (reasoning from specific observations to general principles). The problem exists because a single disconfirming instance can invalidate a principle that has received consistent support in hundreds of previous observations.
A research design in which the researcher has only partial control over his or her independent variables. Quasi-experiments include both natural experiments that are typically based on archival data and person-by-treatment quasi-experiments that are typically conducted in the laboratory.
A technique for assigning participants to different conditions in an experiment. The use of random assignment means that every person in the study has an equal chance of being assigned to any of the conditions of the study. The use of random assignment makes it highly likely that the groups of participants in different experimental conditions are highly similar to one another.
A quasi-experimental research design in which researchers collect multiple waves of data from two presumably comparable groups of research participants. If the two groups of participants receive highly similar scores on a dependent measure prior to a natural manipulation but begin to receive different scores immediately after the natural manipulation, this strongly suggests that the manipulation may be responsible for the observed post-manipulation differences.