STATS 250 Chapter 6

Case-Control Study (Observational)

In a case-control study, "cases" who have a particular attribute or condition are com-pared to "controls" who do not. A distinguishing feature of a case-control study is that individuals are sampled within categories of the response variable (cases and controls).

Randomized Experiments

Randomized experiments are experiments in which the participants are randomly assigned to participate in one condition or another. The different "conditions" are called treatments.

Replication

Replication in an experiment refers to the idea that more than one experimental unit is assigned to each treatment condition. The number of experimental units per treatment should be large enough to provide suitably accurate estimates of response differences among the treatments.

Rule for Concluding Cause and Effect

The Rule for Concluding Cause and Effect is that cause-and-effect relationships can be inferred from randomized experiments but not from observational studies.

Subjects

When experimental units are people, they often are called subjects.

Confounding Variable

A confounding variable is a variable that both affects the response variable and also is related to the explanatory variable. The effect of a confounding variable on the response variable cannot be separated from the effect of the explanatory variable.

Double-Blind Experiment

A double-blind experiment is one in which neither the participant nor the researcher taking the measurements knows who had which treatment.

Matched-Pair Design

A matched-pair design is an experimental design in which individuals are first matched on important characteristics and then each member of a pair receives a different treatment.

Placebo

A placebo looks like the real drug but has no active ingredients. Placebos can be amazingly effective; studies have shown that they can help up to 62% of headache sufferers, 58% of those suffering from seasickness, and 39% of those with postoperative wound pain.

Single-Blind Experiment

A single-blind experiment is one in which the participants do not know which treatment they have been assigned. An experiment would also be called single-blind if the participants knew the treatments but the re-searcher was kept blind.

Completely Randomized Design

An experiment is called a completely randomized design when treatments are randomly assigned to experimental units without using matched pairs or blocks.

Explanatory vs. Response Variable

An explanatory variable is one that may explain or may cause differences in a response variable (sometimes called an outcome variable).

Interacting Variable

An interacting variable is an explanatory variable that affects the magnitude of the relationship between another explanatory variable and a response variable. The interaction is said to be between the two explanatory variables.

Control Groups

Control groups are treated identically in all respects except that they don't receive the active treatment.

Independent Variable

Explanatory variables are sometimes called independent variables.

Prospective Study (Observational)

In a prospective study, researchers follow participants into the future and record relevant events and variables. The prospective approach generally is a better procedure because people often do not remember past events accurately. A possible difficulty with a prospective study, however, is that participants may change their behavior because they know that it is being recorded.

Repeated Measures Design

In a repeated measures design each experimental unit receives all treatments, ideally in a random order. This is an easy and efficient way to control for variation among individuals.

Retrospective Study (Observational)

In a retrospective study, the data are from the past. Participants may be asked to recall past events, or the researchers may use information that already has been recorded about the participants (medical records, for instance).

Double Dummy Experiment

In an experiment to compare two treatments, it sometimes is impossible to mask which treatment is which, so each group receives one active treatment and one placebo treatment. For instance, suppose researchers wanted to compare the effects of nicotine patches to nicotine gum for helping people quit smoking. Obviously, participants would know if they were wearing a patch or chewing gum. So partici-pants are asked to both wear a patch and to chew gum. One of the two will have the active ingredient (nicotine), but the other will be a placebo.

Experiment

In an experiment, researchers manipulate something and measure the effect of the manipulation on some outcome of interest.

Observational Study

In an observational study, the researchers simply observe or question the participants about opinions, behaviors, or outcomes. Participants are not asked to do anything differently.

Participants

In both experiments and observational studies, the subjects may also be called participants.

Experimental Unit

In experiments, the most basic entity (person, plant, and so on) to which different treatments can be assigned is called an experimental unit.

Replication in Science

More generally, replication in science refers to the principle that a single experiment rarely provides sufficient evidence for anything, so it is important to have independent researchers try to reproduce findings.

Unit

The generic term unit is used to indicate a single individual or object being measured.

Block Design

The matched-pair design is a special case of a block design. In a block design, experimental units are first divided into homogeneous groups called blocks, and each treatment is randomly assigned to one or more units within each block.

Dependent Variable

The term dependent variable is sometimes used for the response variable because the values may be thought to depend on the values of the explanatory variable(s).

Lurking Variable

The term lurking variable is sometimes used to describe a potential confounding variable that is not measured and is not considered in the interpretation of a study.

Experimenter Effects

There are numerous ways in which the experimenter can bias the results. These experimenter effects include recording the data erroneously to match the desired outcome, treating subjects differently on the basis of which condition they are receiving, and subtly making the subjects aware of the desired outcome. Most of these problems can be overcome by using double-blind designs and by including a placebo group or a control group that gets identical handling except for the active part of the treatment.

Hawthorne Effect

When participants in an experiment respond differently than they otherwise would, just because they are in the experiment. The Hawthorne effect is a common problem in medical research. Many treatments have been observed to have a higher success rate in clinical trials than they do in actual practice. This may occur because patients and researchers are highly motivated to correctly carry out a treatment protocol in a clinical trial.

Lack of Ecological Validity

When variables have been removed from their natural setting and are measured in the laboratory or in some other artificial setting. Thus, the results do not accurately reflect the impact of the variables in the real world or in everyday life.