Mean:

Commonly known as "average," a measure of central tendency determined by adding up the quantities of each unit in a distribution and then dividing by the number of units.

Median:

A measure of central tendency that represents the midpoint in a distribution of ordered data.

Mode:

A measure of central tendency that represents the most frequent value in a distribution.

Correlation coefficient (r):

A measure that indicates the strength of the linear relationship between two numerical variables. As r tends to +1, the two variables are more strongly positively associated; as r tends to -1, the two variables are more strongly negatively associated.

Statistical significance:

An empirical standard of confidence that the measured relationship between variables is unlikely to have occurred by chance alone. Although there are different formulae and standards for determining statistical significance, most require that the probability of the measured relationship occuring because of chance be less than 5 percent to be statistically significant.

Crosstabulation:

A table format for reporting measurements of relationships between variables. The number of columns correspond to the number of categorical attributes in the independent variable (x); the number of rows correspond to the number of categorical attributes in the dependent variable (y); and each cell reports the proportion of the sample for each type of relationship between variables.

Causality:

A relationships between variables in which at least three criteria are empirically established:

1. X and Y are correlated. 2. X precedes Y in time. 3. The X-Y relationship is nonspurious; no other competing variables account for the observed X-Y relationship. Carter (2001: 19) offers a fourth and final criteria: 4. The X-Y relationship is intuitively pleasing; it fits with our current understanding of how the world functions.

Spuriousness:

A relationship between independent and dependent variables in which the hypothesized relationship is in fact caused by the influence of a third variable that is independent and antecedent to the other two. In crosstabulations, spuriousness is revealed when the direction and strength of the originally hypothesized relationship (i.e., the difference of percentages in cells of each row) diminishes or disappears to zero after a third variable is controlled for with the use of partial tables.

Multivariate model:

A hypothesis in which the effect of more than one independent variable is studied. In crosstabulations, a multivariate model is revealed by changed but consistent relationships (i.e., strengthened in one, weakened in the other) in the originally hypothesized relationship after a third variable is controlled for with the use of partial tables.

Intervening variable:

A third variable that logically falls in a time sequence, and systematically explains the hypothesized relationship, between the independent and dependent variables. In crosstabulations, an intervening variable is revealed when the direction and strength of the originally hypothesized relationship (i.e., the difference of percentages in cells of each row) diminishes or disappears to zero after a third variable is controlled for by constructing partial tables. Note that this is the same effect revealed by spuriousness; only logical analysis about the sequence of the third variable can determine whether it is an intervening variable or hides a spurious relationship.

Interaction effect:

The tendency for a third variable to interact with the independent variable, thereby altering the relationship between the independent and dependent variables. Thus, the originally hypothesized relationship between the independent and dependent variables will vary under different conditions of the third variable. In crosstabulations, an interaction effect is revealed when the originally hypothesized relationship maintains itself in one partial table but diminishes or disappears to zero in the other after a third variable is controlled for.

 

Survey research:

Quantitative social research in which one systematically asks many people the same questions, then records and analyzes their answers.

Questionnaire:

A measurement instrument that provides written instructions and questions which respondents self-administer in order to provide data for analysis.

Interview schedule:

A measurement instrument that provides instructions and questions which the researcher verbally administers to informants in order to gather data for analysis.

Secondary data analysis:

Research in which one does not gather data oneself but reexamines data previously gathered by someone else to test original hypotheses.

Closed-ended questions:

Survey research questions in which respondents choose from a fixed set of answers.

Open-ended questions:

Survey research questions which respondents answer in their own words.

Unit of analysis:

The kind of empirical case or unit that a researcher observes, measures, and analyzes in a study.

Response rate:

The percentage of respondents who complete a questionnaire or interview. Although researchers disagree about what constitutes an acceptable response rate, most consider anything below 50% to be poor and over 90% to be excellent.

Index:

A measure that sums or combines many separate measures of a variable.

Scale:

An index that measures the intensity, direction, level, or potency of a variable constructed along a continuum. Most scales measure ordinal variables.

 

Regression analysis:

A form of statistical analysis that is especially useful in explaining multivariate models. A regression equation expresses the relationship between two or more variables algebraically, estimating the average change in a dependent variable given a change in the independent variable(s). In its simplest (linear) form, a regression equation is usually written: Y = a + b1X1 + b2X2 + ... + bkXk + e where... Y is the dependent variable a (alpha) is the constant (a.k.a. intercept) b (beta) is the regression coefficient (a.k.a. slope) X is the independent variable e is the error term

Dummy variable:

An operational construct that represents a categorical variable as a two-category numerical variable for statistical analysis. E.g., for gender: female = 1, male = 0.

 

Sample:

A group of subjects that are selected for study in order to make generalizations about a broader population.

Population:

The group of all subjects, either known or unknown, from which a sample is selected.

Elements:

The individual units, often individual persons, that comprise a sample.

Probability sampling methods:

Methods for drawing a sample in which the probability of selecting population elements is known. With probability sampling methods, the researcher uses random sampling so that the representativeness of the sample characteristics to the (known) population characteristics can be statistically calculated.

Nonprobability sampling methods:

Methods for drawing a sample in which the probability of selecting population elements is not known. With nonprobability sampling methods, the researcher must explicitly explain how the sample represents the population from which it was drawn.

Sampling frame:

The list from which the elements of the population are selected. Sampling methods are only as sound as the sampling frame operationalizes the population.

Availability sample:

A nonprobability sample in which elements are drawn based on their availability to the researcher. Also known as a convenience sample.

Snowball sample:

A nonprobability sample in which the researcher asks the initial elements, usually people, to refer other potential elements for inclusion in the sample. The process is repeated until the sample grows (or snowballs) to the size desired by the researcher.

Purposive sample:

A nonprobability sample in which the researcher selects elements for a specific purpose, usually because of the unique position of the elements.

Stratified random sample:

A probability sample that is organized to capture known group differences among the population. In the first stage, elements are sorted into separate groups (or strata) according to the selected group characteristics, e.g., men and women, different racial and ethnic groups. In the second stage, elements are randomly sampled from within the strata.

 

Field research:

Research about social groups and behavior based on careful observation in a natural social environment. Field research can entail any research methods, not just participant observation.

Ethnography:

Literally, "people" + "writing." The description of a group, culture, or social practice as its members understand it.

Inductive research:

Research that begins with the observation stage. Typically, inductive researchers let their research questions emerge from their initial observations and analysis.

Participant observation:

A method of research involving participating and observing first hand in the social behavior and groups you are studying.

 

Reactivity:

A problem of validity that occurs when subjects are aware they are being studied and alter their behavior from its "natural" patterns.

 

Field notes:

Writings in which a field researcher records his or her personal observations (the indicators of participant observation). Field notes tend to include at least (1) thick descriptions, (2) running hypotheses, and (3) notes for further investigation.

 

Triangulation:

Drawing together multiple types of evidence gathered from different sources using different methods of data collection.

Control group:

In an experiment, the comparison group who is not exposed to the experimental treatment (i.e., the independent variable).

 

Covert research:

Research conducted without the knowledge or consent of those being studied or by misrepresenting the role of the researcher.

 

Open setting:

A setting for social behavior where the field researcher can enter without permission.

 

Closed setting:

A setting for social behavior where the field researcher must obtain permission from members before entering.

 

Experiment:

A research method that seeks to isolate the effects of an independent variable on a dependent variable under strictly controlled conditions.

 

Treatment:

The independent variable that is administered to subjects in an experiment.

 

Experimental group:

In an experiment, the group who is exposed to the treatment (i.e., the independent variable).

Pretest:

In an experiment, the baseline measure that is compared with the measures taken after the treatment (i.e., posttest).

 

Posttest:

In an experiment, the measure of the effect of the experimental treatment that is often compared to a pretest.

 

Internal validity:

The extent to which an experiment has actually caused what it appears to cause.

 

 

External validity:

The extent to which an experiment can be generalized to other settings, other treatments, and other subjects.

 

Posttest distortion:

In an experiment, the degree to which the observed effect of the treatment is actually due to the effect of measuring the dependent variable twice (in pretest and posttest measures).

Hawthorne effect:

In an experiment, when subjects demonstrate the expected effect without being exposed to the treatment, solely because they are participating in the experiment.

 

 

Classical experiment:

The most fundament experimental design: the experimental and control group are given the pretest, an experimental group is given the treatment, and the experimental and control group are given the posttest.

 

Solomon four-group design:

An experimental design with two experimental and control group pairs, one of which is not exposed to pretest measures.

 

Posttest-only control group design:

An experimental design where neither the experimental nor the control group is exposed to pretest measures.

 

Risk-benefit ratio:

A measure of a study's ethics: are the risks to subjects and science greater than the benefits to science and society?

Content analysis:

An unobtrusive method for analyzing texts that involves four stages.

1. Variable construction: deciding which characteristics or themes of the texts are to be analyzed and how they are to be observed 2. Sampling: selecting the texts to be analyzed 3. Observation: code (or measure) each text for the characteristics or themes 4. Analysis: aggregate the measurements and make numerical descriptions of the texts.

 

Manifest content:

Elements of a text that are already observable and readily measurable without (much) interpretation by the content analyst.

 

Latent content:

Elements of a text that require interpretation by the content analyst before they can be observed and measured.

 

Intercoder reliability:

In content analysis, the need to confirm the accuracy of a coder's measurements by checking (some sample of) them against a second coder's measurements.