Quantitative Research Design

Key Features of Quantitative Research Design

• Key questions to consider include whether an intervention will be part of the research and what specific design will be used.
• Broad design options include experimental (randomized control trial), quasi-experimental (controlled trial without randomization), and nonexperimental (observational study).
• The type of comparisons that will be made to illuminate relationships is a key question to consider.
• Design options include within-subjects, where the same people are compared at different times or under different conditions, and between-subjects, where different people are compared.
• Control over variables that may confound the results is important.
• It is important to understand how confounding variables will be controlled and which specific variables will be controlled for.
• Control can be achieved through randomization, crossover, homogeneity, matching, and statistical methods.
• Masking or blinding can be used to withhold critical information to prevent bias.
• Important to understand when data will be collected relative to other events.
• Designs can either be cross-sectional or longitudinal.
• Consider when information on independent and dependent variables will be collected (looking forward or backward in time).
• Relative timing includes retrospective (case-control) and prospective (cohort) designs.
• Location is an important feature; consider where the study will take place, as well as the setting (single site versus multisite).

Causality

• Many quantitative research questions focus on identifying cause-and-effect relationships.
• Causal relationships need to be addressed with appropriate designs
• A counterfactual is what would have occurred to the same subjects exposed to a cause if they were not exposed.
• An effect shows the difference between what happened with and without exposure to a cause.
• Key criteria for causal inferences: the cause must precede the effect in time (temporal), a relationship must exist, and confounders explained by a third variable must be controlled.
• Additional criterion: the causal relationship should align with evidence from basic physiologic studies (biologic plausibility).
• Different designs are appropriate for answering different questions.
• Experimental designs offer the strongest evidence of whether a cause results in a desired effect.

Experimental Design

• Involves intervention, meaning researcher introduces a treatment to some subjects.
• A control, including a control group to provide a counterfactual, is introduced.
• Randomization is when the experimenter assigns participants to a control or experimental condition on a random basis to make the groups equal while factoring in the intervention.

Experimental designs

• Posttest-only (or after-only) design involved data collection only after an intervention.
• Pretest-posttest (before-after) design collects outcome data both at baseline and after the intervention.
• Crossover design exposes subjects to multiple conditions in random order, allowing subjects to serve as their own control.
• Symbol R denotes randomization, X is receipt of intervention, and O is observation/measurement.

Experimental Conditions

• These must receive intense and long exposure to expect effects.
• Researchers describe the intervention in formal protocols that stipulate exactly what the treatment entails.
• Attention needs made to intervention fidelity, showing that the treatment as planned was delivered and received.
• Control group conditions should be considered, especially if the control group receives no treatment, receives "usual care," receives an alternative intervention (e.g., auditory vs. visual stimulation), or receives a placebo.
• The control group may have delayed treatment or attention control.
• Attention control is extra attention but not an active part of the intervention.
• Delayed treatment is when intervention is given at a later date.
• The advantages of experiments are powerful detection of cause-and-effect relationships.
• The disadvantages include not being feasible or ethical, possibility of the Hawthorne effect, and often are very expensive.

Quasi-Experiments

• This involves an intervention but lacks randomization, a control group, or both.
• Categories include nonequivalent control group designs.
• Also involves within-subjects designs.
• Nonequivalent control group pretest-posttest design involves data collection before intervention, in order to examine how the comparability of the experimental and comparison groups at the start of the study can be examined.
• This is symbolized as O1 X O2 and O1 O2.
• Nonequivalent control group posttest only is a weaker study and it is difficult to assume that the groups were similar at the outset
• Symbolized as X O1 and O1.
• One-group pretest-posttest designs typically yield very weak causal relationships.
• Symbolized as O1 X O2.
• Time-series designs gather preintervention and portintervention data over a longer period symbolized as O1 O2 O3 O4 X O5 O6 O7 O8.
• Quasi experiments are easier and more practical than experiments.
• They are more difficult to infer causality because there are often rival hypotheses.

Nonexperimental Studies

• These are observational only and occur when researchers do not intervene by controlling independent variables.
• These are used when independent varibles cannot be manipulated such as gender and ethics.
• Correlational designs are used for cause-probing questions when a situation can not be manipulated and are weaker than RCT's.
• A correlation is an association between variables and is detected through statistical analysis.
• In a prospective correlational design, a potential cause in the present is linked to a hypothesized latter outcome.
• By medical researchers, this is called a cohort study.
• Prospective designs are stonger than retrospective designs in support of causal inferences.
• I a retrospective correlational design, an outcome in the present is linked to a hypothesized cause in the past.
• One retrospective design is a case-control design and often "cases" are compared to with a "control".
• Descriptive studies' purpose is ot observe, describe, and documented aspects of the situation.
• While descriptive correlational studies' purpose is to describe whether variables are related, without ascribing a cause-and-effect connection.
• Nonexperimental Designs do not yield persuasive evidence of causal inferences.
• It does however make large amounts of data collection efficient when intervention is not possible.

Time Dimension in Research Design

• Cross sectional design is when data is collected a single point in time.
• In longitudinal design data is collected two or more times over an extended period of time.
• These use follow-up studies and are better at showing patterns of change and clarifying if a cause occurred before an effect.
• A challenge is loss of participants called attrition.

Controlling the Study Context

• Controlling external factors can by achieving constancy of conditions.
• One can maintain control over environments, settings and time frames.
• Another method is control over intervention by forma intvention fidelity and adherence to protocol.
• Controlling Participant Factors Randomization occurs when subjects are their own control.
• Homogeneity is restricting sample size.
• Can achieve control via matching and or a statistical control.
• Statistical conclusion validity is the ability to detect true relationships statistically.
• Internal Validity-the extent to which it can be inferred that the independent variable caused or influenced the dependent variable
• External Validity the generalizability of the observed.
• Construct validity is the degree to which key constructs are adequately captured in the study

Threats to validity

• Threats to Statistical Conclusion Validity include the low statistical power, and weakly defined "causes".
• Another is Unreliable implementation of a treatment leading to low intervention fidelity

Threats to Internal Validity

• Temporal ambiguity-not always able to determine a cause if something proceeded for a longtime.
• Selection threat includes biases from pre existing differences between groups being compared.
• Selection is the biggest potential threat in studies without experimental bias.
• History threat shows other events between studies being compared.
• Maturation threats include processes that result simply from the passage of time.
• Mortality or attrition threats include a differential that shows loss of participants of a different group.

Threats to External Validity

• Internal Validity can sometimes counteract External Validity.
• Inadequate sampling of study participants includes those that pertain solely to External Validity.

Threats to Construct Validity

• A good representation of the underlying construct?
• Is an intervention or an awareness of the variable.
• An idea of weather the constructed variable is really a valid choice.

Basic Sampling Concepts

• Population is the entire group of interest based on eligibility criteria which can split to inclusion or exclusion criteria.
• Sampling is the section of the of the strata (subpopulations of a population such as male or female).
• Target population is portion of a accessible population that is accessible to the researcher from which the population is drawn.
• The representative sample should closely match characteristices of the population.
• This goal is achieved more in homogenous setting and larger sampling sizes.
• Randomness can occur through power analyzation.

Sampling Designs

• Non probability sampling doesn't involve selection of elements at random is rarely the population.
• Probability sampling selection through equal chance in sample.

Types of sampling

• Convenience sampling select the participants that are available.
• Quota sampling identifying population strata to see how many people are needed from which stratum.
• Consecutive sampling recruiting all people from an accessible population over a.
• Purposive sampling hand picking sample members.
• Simple random sampling establish from a sampling frame a list of population elements.
• In a stratified random sampling from that set elements are randomly selected.
• Through systemic sampling that case is systematically selected from the list.

Sample sizes requirements

• Should be adequately represented for the quality and should be considered through powe analyzation
• Not meeting the requirements will alter the test.
• Considerations should go toward a sampling approach,the population for eligibility and the characteristics.

Data collection in Quantitative Reseach

• With existing or new data.
• Either records, historical or with existing data.
• Self and or patient reporting is a collection method.

Overview of Data Collection

• Structure
• Quantifiability
• Objectivity
• Should be considered on their respective sources.

Structured self Reports

• Collected with formal instrument or interview
• Quuestions are presepcfied but orally asked

Types of questions

Closed ended (fixed alternative) questions Open ended

Disadvantage

• In questionnaire geographically dispersed with anonymity in obtaining certain opinions or traits.
• Higher repsonse rates are appropriate

Composite Psychosocial Scales

• Assign numerical scores across a continuum
• Allows those to determine others traits and atttributes

Likert Scales

• Declarative statements and responses are summed into a score

Visual Analougue scales

• Is to measure subjective experiences

Response set biases

• Social desirability
• Exclusive response set

Types of observations

• Through reactivity and with observation biases.
• Observation bias probably won't be completely eliminated but minimized through awareness etc.

Biospyiologic Measures

• In vivo measurements are perfromed directly or on living organisms while in virto is not.
• Strength based accuracy.

Factors affecting data Quality

• With procedures and measurement qualities.
• Physchomentric assessment evaluates measurements.
• Reliablility is extent to which is free from error.

Data Quality in Quantitative Research:

Validity.

• Face validity: weather like it is measuring
• Content validity: To the instruments
• Critserion Validity: Score of how well it is rated.
• Construct Validity: Relates well to what it is presented.

SAQ

• The SAQ (Seattle Angina Questionnaire) is a well-established disease-specific measure.
• The SAQ includes 19 quantified tests.
• SAQ tests include physcial and mental.
• SAQ tests are done throught ordinal and value summing with five sub scales.

Ethical Research

• Respect for those as human beings.
• And respect the choices.
• And limit who has those autonomy.
• Consider personal research and privacy.
• Follow Benefiance with the right research risk.
• Equitable selection of research.
• And all information of the participants.
• And a research board to oversee everything.
• And vulnerable parties.
• And code of conduct depending the the country.

Reseach boards

Composed of a mix of many.

Recruitment is also reviewed by said boards.

All information should be reviewed by the boards. -Full and unbiased

Special Consideartions.

• Chilrden

• Prisoner

• Disablities

• Disadvantaed Pregences that would cause it.

• And Code to keep. CNA's Code of Ethics from 2008.

• Promoting safe, compassionate, competent- Care Promoting well being- Promoting ethical decisions while well with dignity Maintaining all forms of care and promote.