quantitative research methods
Characteristics of Quantitative Research
Collection and analysis of numerical data to describe, explain, predict, or control variables and phenomenon of interest
Underlying tenet is the philosophical belief that world is stable and uniform, that we can measure and understand it, and make broad generalizations about it--contrasts with qualitative approaches
Focus is usually on the singular perspective of describing and explaining--sometimes in a definitive manner--phenomenon under investigation
Operates under widely agreed-upon steps that guide research process
Research process composed of widely agreed-on steps with little flexibility in strategies
Quantitative researchers believe nothing should be left to chance; therefore, no aspect of design is permitted to emerge during process
The major goal is for researchers to remain as objective as possible
Steps of process are much more linear
Focus on objectivity permits generalizability of results beyond particular research setting
Additional characteristics
Literature review is more central to research design than in qualitative research
Research purposes/goals are typically specific and narrow
Data collection procedures are specified prior to beginning study
Sampling strategies differ from qualitative studies.
Tend to focus on random selection in order to improve generalizability
Typically collect data from a large number of individuals
Data analysis and interpretation of results are entirely statistical in nature
Reporting of results is very standardized and almost “preformatted”.
The Quantitative Research Process
Identification of the research problem to be studied
Tend to be narrow in scope
Focus on a handful of key variables
Statement of one, or several, pertinent research questions and/or hypotheses
Research questions and hypotheses have to be stated clearly
Failure to clearly state these at outset may lead to problems as study progresses
Review of related literature: Can inform decisions regarding research design, sampling,instrumentation, data collection, and analysis
Development of a written literature review: Synthesize pertinent body of literature
Development of a research plan: Strategies for selecting sample of participants, research design, and data collection and analysis need to be included
Collection of data
Does not take a lot of time, depending on the design
Typically taken directly from participants using a variety of instruments
Analysis of data
Conducted through the use of statistical analysis software programs
Focus on numerical descriptions, comparisons of groups, or measures of relationships
Development of conclusions and recommendations
Drawn from the interpretation of result
Typically connected back to the body of literature from the review
Preparation of a final research report
Approaches to Conducting Quantitative Research
Nonexperimental research designs: No manipulation to variables
Descriptive research
Goal to describe and make interpretations about individuals, settings, and so on
Includes observational research and survey research
Observational research
Focus on quantification of observations in order to measure complexities of human behavior
Yields data that depict complexity of human behavior
Survey research
Goal is to describe characteristics of group or population
Representative respondents selected from population (return rate is very important)
several modes of data collection
direct administration: access to all, or most, members of a group in one place
mail surveys: sending a hard copy to each individual
telephone surveys: administered individually and may be expensive
Interviews: most costly because most be done individually and face-to- face
electronic surveys: e-mail, web-based
Types of surveys
Descriptive survey: One-shot survey to describe characteristics of a sample at one point in time
Cross-sectional survey (census): Examination of characteristics of (and differences among) many samples or populations measured at one point in time
Longitudinal survey: Group or cohort studied at different points in time
Trend study: Changes within specifically identified population over time
Cohort study: Specified population within a subgroup studied whose members have common characteristic
Panel study: Same people examined over specified length of time
Survey research process
Identification of the topic to be studied
Review of related literature
Identification and selection of participants: Identify the target population (larger group of people to whom to generalize the results of the study)
Determination of the mode of data collection
Drafting the cover letter and instrument
Pilot test of the instrument: Trial run of the data collection process to determine if revisions need to be made before true data collection happens
Collection of data
Analysis of data
Strengths and limitations
Efficient data collection from large number of people
Allows for greater degree of generalizability
Options allow for customization of survey research process
Potentially low response rates
Requirements of some modes of survey delivery • Surveys result in self-report data
Correlational research
Measure of relationship between two or more variables
Explanatory correlational studies (goal to comprehend and describe certain related events, conditions, and behaviors) versus predictive correlational studies (predict future conditions or behaviors in one variable from what we know of another variable)
No manipulations of variables; measure as they “exist in nature”
Relationships measure with correlation coefficient (r): Measure of
strength and direction of relationship
Correlational research process
Identification of the topic/problem to be studied
Review of related literature
Identification and selection of participants
Specification of the design and procedures for data collection
Collection of data
Analysis of data
Answering research questions and drawing conclusions
Strengths
Simplicity of design
Often appropriate for novice researchers
Limitations
Correlation ≠ causation
Must ensure validity and reliability of data
Resulting correlation coefficient will look meaningful but can actually be misleading
Causal-comparative research
To determine why or how two existing groups are different from one another
Also known as ex post facto research: First observe differences; look back in time to try to determine possible conditions that led to differences
Most often, presumed cause (independent variable) has already occurred; this is sometimes called the grouping variable
Dependent variable is ultimate variable of interest.
Independent variable also might be a preexisting condition.
Results might only suggest cause-and-effect type of relationship (identify possible causes).
Causal-comparative research process
Identification of the topic/problem to be studied
Review of related literature
Identification and selection of participants
Specify the design and procedures for data collection
Collection of data
Analysis of data
Answering research questions and drawing conclusions
Strengths and limitations
Solid alternative when experimental designs are not feasible or possible
May not be possible to manipulate the grouping variable
Cause has already occurred; no control over it
Experimental research design: Researcher establishes different treatments or conditions and studies their effects on participants
Categories: Pre-experimental research designs, quasi-experimental research designs, true experimental research designs, and single-subject research designs
Demonstrate cause and effect most convincingly of all research designs
Generally, require main components
Sample with comparison group
Independent variable, or treatment variable
Dependent variable, or criterion variable
Random selection (process of choosing, in random fashion, individuals for participation, in order for each member to have equal chance of selection) versus random assignment (each individual randomly selected to participate has equal chance of being assigned to any group compared in the study)
Two categories of experimental research design
Single-variable designs: Involve only one manipulated independent variable
Factorial designs: Involve two or more independent variables and at least one is manipulated
Pre-experimental research
Do not do a good job of controlling for extraneous variables
Appropriate for preliminary investigations of research topics
Designs include one-shot case study, one-group pretest–posttest design, and static-group comparison design
Strengths and limitations
Good for preliminary investigations
Weak designs
No control over extraneous variables
Very difficult to draw definitive conclusions
Quasi-experimental designs
Closest to true experimental designs
No random assignment of participants to groups
Appropriate for designs requiring the use of participants in existing groups
Often used for research in school settings; designs include matching posttest-only control group design, matching pretest–posttest control group design; counterbalanced design, and time-series design
Strengths and limitations
Good option when random assignment is not feasible • Vast improvements over pre-experimental designs
Still do not control extraneous variables
Still difficult to draw definitive conclusions
True experimental research
Involve both random selection AND random assignment of participants
All designs have at least one comparison group
Simplest experimental design is posttest-only control group design
Other designs include pretest–posttest control group design and Solomon four-group design.
Experimental research process
Identification of the topic/problem to be studied
Review of related literature
Identification and selection of participants
Specification of the design and procedures for data collection
Collection of data
Analysis of data
Answering research questions and drawing conclusions
Strengths and limitations
Capacity to draw strong conclusions about cause-and-effect relationships
Requirements are extremely stringent (sometimes prohibitive)
Must take care to counteract threats to validity
Single-subject experimental research design
Experimental-type studies conducted on individual participants
Focus on promoting a change in behavior
Non Treatment phase usually symbolized by “A” and treatment phase symbolized by “B”
Various types, including A–B design, A–B–A design (reversal design), alternating treatment design, and multiple-baseline design
Strengths and limitations
Ability to focus on an individual
Ability to study effectiveness of a treatment on only that participant
Alternative explanations will always exist
Must find ways to factor out these alternatives
Threats of Validity in Quantitative Designs
Refers to degree research conclusions are considered accurate and generalizable
All types of quantitative research are subject to threats of validity
Internal validity
Degree to which measured differences on dependent variable are a direct result of manipulation of the independent variable and not another influence or variable
Main threats
History: When treatments extend over longer time, factors other than treatment may influence results
Maturation: If treatments extend over longer periods, participants may change physiologically and influence the dependent variable
Differential selection of participants: Selected participants may have differences accounting for variations
Testing (pretest sensitization): Participants may learn from the pretest to improve performance on the posttest, even when the treatment has no effect
Instrumentation: Sometimes instruments for measurement are unreliable or lack consistency in measurement, which affect variables of interest
Statistical regression: This occurs when participants have very high or low scores on a measure
Attrition (mortality): Individuals may leave the experiment as it is conducted due to illness, moving, dropping out, and so on
Selection-maturation interaction: Differential selection of participants may interact with other threats, such as history, maturation, or testing, in which a group may perform worse or better with a particular treatment.
External validity
Extent to which results of particular study are generalizable to other groups or settings
Main threats
Population validity: Degree of similarity among sample used, population from which sample was drawn, and target population to which results are to be generalized
Personalogical validity: A research finding may apply well to some individuals and poorly to others
Ecological validity: The situation, physical or emotional, existing in the experiment--may be different from the sitting where results are to be applied