Glossary

Abductive inference

This is a form of inference in which an explanation of a set of observations is proposed on the grounds that it is the best explanation available. It is sometimes referred to as inference to the best explanation. Abductive inference was developed by the philosopher Charles Peirce in the late 19th century.

Action research

Action research is research conducted with the aim of bringing about change to improve some aspect of social conditions. It is both participative and collaborative: participative in the sense that stakeholders are involved in all stages of the research; and collaborative in that participants are invited to contribute to the design and implementation of the study. Reflection on the outcomes is a key ingredient of action research in considering the impact of the research and working out ways to improve the outcome.

Analysis of variance

This is an inferential statistical technique for testing for differences among means of any number of groups where the dependent variable is measured on an interval or a ratio scale. The method only tests whether all the group means in the population are equal. If the test leads to the rejection of this hypothesis, the researcher can conclude that there are differences but not state where those differences lie.

Applied research

This is research conducted to address specific issues or problems considered to be of practical significance. The focus of applied research is on problem solving rather than on expanding the knowledge base in a particular field.

Basic research

This is research conducted with the aim of extending knowledge in the field in which the research is conducted. It is curiosity driven, whereby researchers seek to fill gaps in knowledge as revealed in existing research.

Case study method

A case study is a research method that investigates a situation as it occurs in its real-life context. It can involve directly observing events as they occur, interviewing people or groups involved in the event, or obtaining secondary data describing the event.

Causality

A causal connection is one in which one event (the cause) brings about the occurrence of another event (the effect). For example, high blood alcohol level in drivers can be seen as a cause of motor vehicle accidents (although not the only cause). There are three requirements for establishing causality. For an event to be a cause of another, the following requirements must be met:

• 1.
  The two events must occur together (constant conjunction).
• 2.
  The cause must occur prior in time to the effect.
• 3.
  Other possible causes must be ruled out.

Chi-square test (of association)

The chi-square test is a statistical inference procedure used to test for association between two variables when they are measured on ordinal or nominal scales.

Coding strategies

Coding strategies are methods used to construct codes to analyse qualitative data. Three strategies have been identified:

• 1.
  Open coding: a strategy for constructing codes in the analysis of qualitative data where codes are generated from the data.
• 2.
  Axial coding: a strategy for constructing codes in the analysis of qualitative data which involves establishing relationships among codes formed during the open coding process. It identifies what open codes have in common and links them together by setting up higher level codes.
• 3.
  Selective coding: this is the final stage of coding where codes are integrated into a theoretical framework aimed at providing an explanatory framework for the coding scheme.

Complex causation

Complex causation is characterized by multiple causes determining an effect as distinct from single cause–effect relationships. Such causes may operate as either–or causes or as conjunctural causes where the simultaneous occurrence of two or more causes produces the effect. A situation in which both either–or and conjunctural causation operates is referred to as multiple conjunctural causation. In this case, two or more different combinations of conditions may produce the same effect.

Conceptual framework

The conceptual framework for a research project is the set of concepts used in the research project, including their definitions and type. It also includes the paradigm within which the researcher is operating, although this is not always made explicit.

Conflict theory

Conflict theory sees social structures as products of conflict between social classes where the conflict arises from economic inequalities between these classes. It is seen to originate in the work of Karl Marx but has been developed by later theorists such as C. Wright Mills.

Constructivism

Constructivism is a paradigm in which reality is seen to be socially constructed. It is adopted by some qualitative researchers who see social constructions as the subject matter of social science.

Content analysis

Content analysis is a method of classifying qualitative data by classifying text into categories, where the categories are usually predetermined. It has been applied to the analysis of media stereotypes, such as the portrayal of gender in magazines, newspapers, television and films.

Conversation analysis

Conversation analysis focuses on human interaction as it occurs naturally through talk. It involves a detailed analysis of conversations to provide an account of what happens in social interactions. It is based on the assumption that conversation is central to human interaction and is the primary means by which we relate to one another.

Correlation coefficient

A correlation coefficient is a measure of association between two variables. Where the variables are measured on interval/ratio scales, the measure of association used is normally the Pearson product moment correlation. For other measures, Spearman’s rho can be used.

Critical realism

Critical realism is a paradigm that shares many features with post-positivism, in particular the existence of an external reality which can be accessed by the use of scientific method. It was formulated by the philosopher Roy Bhaskar in the 1970s and then termed transcendental realism. It differs from post-positivism in that it postulates unobservable events that can act as causes of observable ones.

Cross-sectional survey

This is a survey that is administered at a single point in time. A cross-sectional survey is usually designed to compare subsets of the population, such as males and females, on opinions or practices, such as alcohol consumption.

Deductive inference

Deductive inference involves using logical principles to draw conclusions from a set of premises. If the premises are true, and the logical principles are followed, the conclusion is necessarily true. This form of inference is used in deriving predictions from a theory rather than drawing inferences from research findings.

Dependent variable

A dependent variable is one which the researcher is seeking to explain or understand. It is seen as a consequence of the independent variables, such as performance on a simulated driving test where the independent variables might be number of hours of sleep deprivation (manipulated by the researcher) and age and gender of the participant (background variables).

Discourse analysis

Discourse analysis is an approach to the analysis of text that situates the text within the context in which it was produced. It seeks to interpret the text by examining the assumptions underlying its production. These assumptions are often hidden and require a critical analysis of the ways in which meaning is constructed by the linguistic formations that give rise to them. Social realities are seen as being constructed by language in the form of discourses.

Ethnographic writing

Three approaches to the writing of ethnographic reports are:

• 1.
  Realist tales: in this approach, the report is written in the third person and the researcher is absent from the content.
• 2.
  Confessional tales: here the account incorporates the experiences of the researcher and is written from the viewpoint of the researcher
• 3.
  Impressionist tales: these are designed to achieve dramatic effect by being written as a story to convey the experience of the researcher.

Ethnography

Ethnography is a research method involving the study of social groups or cultures through collecting information about the group or culture over time, both by direct observation and interviews with members.

Evidence-based practice (EBP)

Evidence-based practice (EBP) is the process in which decision-making is based on the best available evidence derived from research findings. Although originating in clinical practice, it has been extended to fields beyond medicine, including allied health professions, education, psychology, social work and many others.

External validity

External validity refers to the extent to which findings from a research study can be generalized to other populations or settings.

Feminist theory

Feminist theory focuses on gender inequalities in societies where women have experienced discrimination in social, political and economic spheres. As a theoretical framework, feminist social theory was developed in the late 1960s by writers Shulamith Firestone, Betty Friedan and others.

Field study

A field study is a data collection procedure involving observations of people or groups engaged in their normal everyday activities. Field studies often form part of the data collection in a case study.

Focus group interview

A focus group interview is a form of group interview in which the interviewer, often referred to as the facilitator, directs questions to the group on a specific topic. Members of the group are usually chosen because of their involvement with the topic being discussed.

Historical-comparative research

Historical-comparative research examines society and culture either within nations or across nations and either at a single point in time or across different time periods.

Independent variable

An independent variable is one which is either manipulated by the researcher such as time spent awake in a sleep deprivation study, or a background variable, such as the age of participants, which is seen to be relevant to the research outcomes.

Index

An index combines a number of distinct measures of a concept into a single score. Indexes are usually constructed from secondary data such as country statistical information. An example of an index is the Human Development Index (HDI), which combines life expectancy at birth, education level and income of a country to provide a measure of well-being in a country.

Individualistic explanation

An individualistic explanation of a social phenomenon locates the explanation within the motives, values or intentions of individuals. Proponents of this approach argue that all social processes can be explained by recourse to individual actions and that there is no need to invoke higher-level structures in the explanation.

Inductive inference

Inductive inference involves generalizing from a set of specific cases to a general principle. If the specific cases all exhibit a particular characteristic, then it can be inductively inferred that all cases of this kind will exhibit this characteristic. Such an inference is not necessarily true since it is always possible to find cases that do not exhibit the characteristic.

Inferential statistics

Inferential statistics are the procedures used to generalize from a sample to a population. For these procedures to be applicable, samples need to be randomly chosen from a population. The logic of inferential statistics is that it enables a generalization to be made to a population with a known level of confidence.

Internal validity

Internal validity refers to the extent to which a research study enables cause–effect relationships to be established from the findings of the study.

Interpretive rigour

Interpretive rigour refers to the extent to which interpretations made on the basis of research findings are credible.

Interpretive understanding

Interpretive understanding as a goal of theorizing is supported by those who claim that explanation is not appropriate for the social sciences. Instead, social research should be aimed at gaining an understanding of the meaning people attribute to their social experience. What is now called interpretive sociology was developed in the early 20th century by the German sociologists Max Weber and George Simmel. It was a reaction against positivism, which was seen by interpretive theorists as neglecting the real subject matter of social science, namely, the ways in which people make sense of their situations by attributing meaning to them.

Interpretivism

Interpretivism is a paradigm which focuses on the meaning and understanding of social reality as experienced by individuals. It is based on the premise that there is no single external reality but rather multiple realities that are constructed by individuals as they interact with their environment. The goal of social research is to uncover these realities.

Intervening variable

An intervening variable is one that mediates in some way the relationship between the independent and dependent variables. Intervening variables can be either mediator or moderator variables. A mediator variable is one that is affected by the independent variable and in turn affects the dependent variable. A moderator variable is one that affects the strength of the relationship between the independent and the dependent variables.

Kruskal-Wallis test

The Kruskal-Wallis test is a non-parametric method that tests whether two or more independent samples originate from the same distribution. It is used when requirements for the use of analysis of variance are not satisfied.

Likert scales

This technique for constructing a scale from multiple items was originally developed by Rensis Likert in 1932 and has survived to still be widely used in developing attitude scales. Items are chosen to represent levels of agreement and are rated on a scale from strongly agree to strongly disagree. Scores are assigned to each category and a total score is obtained by adding the score on each item in the scale. Items that discriminate between high and low scorers on the scale are retained in the final version and items that don’t discriminate are deleted, leaving a scale consisting of items that are considered to measure the attitude well.

Longitudinal survey

This is a survey that is administered at more than one point in time. Longitudinal surveys are designed to track changes in attitude or behaviour over time.

Macro-level theory

Macro-level theory is at a high level of abstraction encompassing whole societies, social institutions or sectors of society, or historical periods. Researchers may generate their research topics from such theories but the concepts are too general to guide researchers in formulating specific research projects. Some macro-level theories are structural functionalism, conflict theory and social positions theory.

Measurement scales

Measurement scales used in the social sciences are as follows:

• i.
  Nominal scale: a nominal scale is one in which the concept is divided into categories that are mutually exclusive and exhaustive, such as gender, marital status or nationality. Nominal scales are the weakest form of measurement.
• ii.
  Ordinal scale: an ordinal scale is one in which an order relationship exists among the values assigned to the concept. Level of education is an example of an ordinal scale.
• iii.
  Interval scale: in an interval scale, distance between levels of the measurement are preserved. The classic example of an interval scale is temperature, as measured by the Fahrenheit or Celsius scales. In an interval scale, there is no absolute zero.
• iv.
  Ratio scale: a ratio scale is one that preserves intervals, as in an interval scale, but in addition has an absolute zero. Height and weight are examples of ratio scales. The only arbitrary component in a ratio scale is the unit of measurement, such as centimetres or inches.

Micro-level theory

Micro-level theory focuses on the more specific aspects of social life and uses concepts that can more directly relate to the research project. An example of a micro-level theory is self-efficacy theory, derived from Bandura’s social cognitive theory.

Mid-range theory

Mid-range theory deals with social processes at a societal level. The concepts involved are less general than those of macro theory but are still quite removed from the specific situations that researchers study.

Multiple-item scale

A multiple-item scale combines a number of separate questions into a single scale. Examples in common use include Likert scales and Rasch scales.

Narrative analysis

Narrative analysis is a method for interpreting stories contributed by individuals in either written or oral forms. It seeks to situate the stories in the wider culture within which they are created and to identify the ideologies embedded in them. Approaches to narrative analysis include:

• i.
  The thematic approach: this approach focuses on the identification of plots and subplots in the stories and integrating them into themes.
• ii.
  The structural approach: this approach focuses on the linguistic structure of the stories. It takes each clause and examines how it connects to the next and subsequent clauses so as to build a sequence that provides a cohesive unit that reveals the meaning of the narrative.
• iii.
  The dialogical/performative approach: this approach focuses on the larger questions that situate the story or stories being told in their context. The approach seeks to answer questions such as what is being accomplished by the story and why is it being told.

Non-parametric methods

These are inferential statistical methods used to test for differences among groups when analysis of variance is not applicable.

Non-probability sampling

Non-probability sampling refers to a variety of methods where sampled elements are chosen in a non-random manner, usually based on criteria set by the researcher. Such sampling procedures do not use random procedures for choosing sample elements, so that no generalization to a population is warranted. Methods of non-probability sampling include:

• i.
  Purposive sampling: purposive sampling involves choosing sample elements on the basis of their relevance to the research questions. These elements may be research sites in a multiple case study or particular individuals to be interviewed. The basis of choice is how well the sampled unit is likely to provide information that answers the research questions.
• ii.
  Theoretical sampling: theoretical sampling involves choosing sampling units that can be used to test a theory. In this form of sampling, the researcher derives predictions from a theory and then chooses a sample unit that satisfies the conditions for a test of the theory.
• iii.
  Convenience sampling: a convenience sample is one chosen by a researcher because of its ready accessibility. Convenience sampling involves including sample elements available to the researcher, either due to their being located in the proximity of the researcher or their being easily contacted, such as through Facebook.
• iv.
  Snowball sampling: a snowball sample is one where subsequent sampled elements depend on previously chosen elements, often through a referral process. The analogy is to a snowball gathering more snow and getting larger as it rolls downhill. Snowball sampling is used when participants are difficult to find because of the particular characteristics they possess that are either rare or hidden in a population.

Observational protocol

An observational protocol is a guide to observers, detailing what it is they should be observing and how to record their observations. It is used in field research where researchers observe events as they occur in ‘real world’ situations.

Paradigm

A paradigm is a broad concept about the nature of research and its philosophical underpinnings. In the social sciences, a paradigm has been defined variously as a world view, as an epistemological stance, as shared beliefs among a community of researchers and as a model example of quality research.

Parallel mixed-methods designs

These are mixed-methods designs in which the qualitative and quantitative methods are implemented concurrently. Variants of this design are:

• i.
  Convergent parallel design: in this design, the qualitative and quantitative methods are of equal status and are implemented independently, with integration occurring at the interpretation phase of the research.
• ii.
  Embedded parallel design: in this design, one method (e.g. the qualitative) is embedded in the other (e.g. the quantitative). Usually, the embedded component plays a minor role in the design.

Participant observation

This is a method of collecting information about the activities of groups or cultures where the researcher becomes a member of the group. The researcher may or may not reveal their research role. In the former case, the researcher is referred to as a complete participant, and referred to in the latter case as a participant observer.

Positivism

Positivism is a paradigm in which the aim of research is seen to involve objective, value-free observations of an external reality, in accordance with scientific method with a view to establishing cause–effect relationships.

Post-hoc comparison

This is an inferential statistical technique used when an analysis of variance leads to a rejection of the null hypothesis that all group means are not equal. A post-hoc test such as the Scheffe technique is used to locate the differences among the means.

Post-positivism

Post-positivism is a paradigm in which the aim of research is to attain both explanation and understanding of social reality through research employing a wide range of methods. It is a reformulation of positivism to modify the requirements of objectivity and value freedom to recognize the influence of values and subjectivity in observations.

Post-structuralism

Post-structuralism was a reaction against what was perceived to be a deterministic approach of structuralism. Social processes cannot be understood by locating them in a larger system but are better seen in relation to the discourses that construct knowledge about these processes. The development of post-structuralism is usually attributed to the work of Michel Foucault.

Primary and secondary sources

A primary source is a first-hand account of an event or situation as it occurred, whereas a secondary source is an analysis or reconstruction of that situation. Primary sources include diaries, autobiographies, legislation, government publications and original documents such as birth or death certificates. Secondary sources include newspaper reports of events, books reporting on events, and biographies. Both types of source are, however, secondary data in that they have been produced by others and not by the researcher.

Probability sampling

Probability sampling is a method where all members of a defined population have a known probability of selection in the sample. In most cases, the sampling is conducted using a random procedure. Probability sampling methods include the following:

• i.
  Simple random sampling: a simple random sample is one in which all members of the population from which the sample is chosen have a known, usually equal, probability of selection. This requires all elements of the population to be identified in some way so that they can be eligible for choice in a sample.
• ii.
  Cluster sampling: in cluster sampling, the population is divided into clusters. Such clusters are convenient ways of subdividing the population for easier access. Samples are chosen randomly from within the clusters.
• iii.
  Systematic sampling: a variation on simple random sampling is systematic sampling where one element of the population is chosen at random and then every kth member, where k is a whole number, is chosen thereafter until the sample size is reached. For this to be achieved, the population needs to be ordered in some way. The number k is referred to as the sampling fraction.

Qualitative comparison analysis (QCA)

Qualitative comparison analysis is an approach to the analysis of qualitative data in which data consists of events in which outcomes can be determined and the analysis consists of identifying conditions relevant to the explanation of the outcome. The aim of QCA is to establish a set of conditions that can provide a causal explanation of the outcome.

Qualitizing

Qualitizing refers to the process of converting quantitative into qualitative data. This is done in order to identify categories from the quantitative data to facilitate analysis of the qualitative data within those categories.

Quantitizing

Quantitizing refers to the process of converting qualitative into quantitative data. This is done to facilitate comparisons of the quantitative and qualitative data by converting them into a common format. To do this, qualitative data is coded and the codes converted into numerical scales.

Quasi-experiment

This is an experimental method that shares all the conditions of the randomized control trial except that of random allocation. Quasi-experiments are conducted when randomization is not possible or is impractical. In a case control design, the groups are matched on some factor or factors, such as age, thought to be capable of influencing outcomes.

Randomized control trial

This is an experimental method in which participants are assigned randomly to one or more treatment conditions or to a control condition. The treatment conditions are those in which participants receive some form of intervention such as a programme or other treatment, whereas the control group receives no, or the existing, treatment. The effects of the intervention are ascertained by comparing the experimental or treatment conditions with the control condition.

Rasch scales

This technique was introduced by Rasch in 1960 in which he proposed a model for measuring both items and individuals on the one measurement scale. The technique is widely used in educational measurement.

Regression analysis

Regression analysis is a statistical inference procedure used to predict a dependent variable from two or more independent variables, where all variables are normally measured on interval/ratio scales.

Reliability of measures

Measurements need to meet the criteria of reliability and validity to be useful in research. There are several notions of both reliability and validity that apply to the measurement of social concepts:

• i.
  Stability: stability refers to the extent to which the measure does not change over time. It is measured by the test–retest method. This involves taking the measure at two time periods and calculating the correlation between the two measures. A high correlation means that the test–retest reliability is high.
• ii.
  Equivalence: equivalence refers to the extent to which the measure yields similar results to alternative measures of the same concept. It is also used to refer to the extent to which multiple items in a scale correlate with one another. For a multiple-item scale to measure a single concept such as an attitude, items need to correlate highly with one another and with the total score on the scale. The measure of equivalence used in such cases is Cronbach’s alpha.

Research question

Research questions are those formulated by the researcher that the research project aims to answer. They are the main focus of the research. Research questions should be distinguished from aims and hypotheses. Aims are the broad statements of the purpose of the research and are more general than research questions, while hypotheses are more specific predictions about the outcomes of research.

Research utilization

Research utilization refers to the implementation of research findings to improve policy or practice in human services. It can take the following forms:

• 1.
  Instrumental use: research findings are used to influence a policy or practice.
• 2.
  Tactical use: research findings are marshalled selectively by policy makers to justify a decision already made.
• 3.
  Conceptual use: research influences how policy makers approach issues and problems.
• 4.
  Process use: being involved in research influences how practitioners understand how research is produced.

Sampling frame

A sampling frame is the list of all possible participants eligible to be included in a research project, referred to as the population. The sampling frame includes all elements of the population and identifies the means for accessing them.

Secondary data

Secondary data is data that exists prior to the research being undertaken. It can be data from previous studies, data collected by government agencies such as statistical information, or information held by organizations such as employment records, meeting records, and so on. It can either be publicly available, as is the case for most government agency data, or private, being owned by individuals or organizations. It can also be either quantitative, such as government statistical data, or qualitative, such as minutes of meetings or diary entries.

Semi-structured interview

Semi-structured interviews include predetermined questions, as in structured interviews, but permit the interviewer to ask follow-up questions on some topics for which additional information is considered necessary.

Sequential mixed-methods designs

These are designs in which the qualitative and quantitative methods are implemented sequentially in time. Variants of this design are as follows:

• 1.
  The elaboration sequential design: in this design, the quantitative method precedes the qualitative method. Usually, the qualitative method is designed to follow up on some of the findings in the quantitative method to provide further understanding or explanation.
• 2.
  The developmental sequential design: in this design, the qualitative method precedes the quantitative method. Usually, the quantitative method is designed to examine the generalizability of the findings of the qualitative method or to develop concepts identified in the qualitative stage.
• 3.
  The iterative sequential design: in this design, more than two methods are implemented sequentially with subsequent methods designed to explore findings in preceding methods.

Single-item scale

A single-item scale is one constructed from a rating scale or a single question providing a range of options from one extreme to the other. An example is a rating of job satisfaction.

Social desirability bias

Social desirability bias refers to a tendency for people to present a more socially desirable image of themselves to the researcher, rather than report their actual beliefs or behaviours.

Social positions theory

Blau (1977) sees social structure as a multidimensional space of social positions in which individuals are located and these positions are instrumental in determining social relations. Structural parameters such as age, race, education and socioeconomic status are the axes in the space that underlie the distinctions individuals make in their social relations.

Statistical conclusion validity

Statistical conclusion validity refers to the extent to which statistical inference procedures used in a study are adequate to detect relationships or differences among variables measured in the study.

Strategic research

Research conducted to expand knowledge in areas considered to be of national significance. It tends to be focused on policy areas considered to be important by governments such as health and education.

Structural explanation

A structural explanation is one in which an event is located within a larger system which imposes constraints on its occurrence. These constraints set limits on what actions are possible in a particular situation.

Structural functionalism

Functionalism sees society as being composed of a system of interrelated components which work together to bring about stability. These components satisfy the needs of the system as a whole and act to shape society. Functionalism as a theoretical methodology was developed by social theorists such as Emile Durkheim, Robert Merton and Talcott Parsons, among many others, in the early part of the 20th century.

Structuralism

Structuralism refers to the view that social processes can only be understood by their relation to overarching social systems or structures. This view was prominent in anthropology due to the work of Claude Lévi-Strauss, of Jacques Lacan in psychoanalytic theory and of Louis Althusser in social science.

Structured interview

A structured interview is one in which the questions to be asked of the interviewee are predetermined, with little scope for variation by the interviewer. Structured interviews are frequently used in case studies where the researcher usually has a clear idea of what questions need to be asked.

Survey method

A survey is a research method designed to collect information from a sample of individuals or groups, usually selected randomly from a defined population. The information collected is often attitudes, opinions, practices, knowledge or other matters, with a view to generalizing findings to the population from which the sample was drawn.

Symbolic interactionism

Symbolic interactionism is a framework that focuses on how individuals construct meaning through their communications with others. It is attributed to George Herbert Mead, but was developed by Herbert Blumer after Mead’s death in 1931.

Theoretical framework

The theoretical framework is the connections that are proposed to exist among the concepts used in the study. It provides a framework for specific theories that share concepts and theoretical mechanisms. A theory is a specific version of a theoretical framework.

Triangulation

Triangulation is the use of multiple research methods to provide a more comprehensive understanding and/or explanation of the phenomenon being investigated than could be obtained by use of a single method. It is widely used as a justification for the use of mixed methods.

Unstructured interview

Unstructured interviews are those in which a general topic is specified but the questioning is left to the discretion of the interviewer. In such cases, the interview schedule would consist of an outline of the information needed from the interview and some suggested lines of questioning but would not include specific questions.

Validity of measures

Validity of measures refers to the extent to which measures actually measure what they are purported to measure. There are a number of alternative meanings of measurement validity:

• Face validity: face validity refers to the judgement by users of the measure that it is a valid measure. That is, a measure has face validity if there is consensus among users that it is a valid measure of the concept. It is not a particularly useful notion of validity as there are many situations where wide agreement over a concept has turned out to be false.
• Content validity: a measure has content validity if it includes all aspects of the concept being measured. For some concepts, areas in which the concept applies are not always obvious so that care needs to be taken in defining the concept to ensure such areas are included.
• Construct validity: construct validity refers to the extent to which a measure of a concept does in fact measure the concept accurately. Construct validity is often inferred from a theory from which the concept is derived.
• Criterion validity: a measure is said to have criterion validity if it correlates with a pre-existing criterion or predicts some future event regarded as a consequence of the concept being measured.