Untitled

Research Designs: Meaning and Purpose

• Research design is a strategy to answer research questions using empirical data.
• It involves a definite pattern or plan of action throughout the research procedure.
• This pattern acts as a blueprint that guides the researcher in data collection and analysis.
• Research design provides a roadmap, outlining the steps and procedures for collecting and analyzing data.
• It ensures that the study is well-organized and systematic and aims to minimize bias and errors ensuring validity and reliability.
• Well-designed studies increase the findings' validity and reliability.
• Research design helps researchers make efficient use of resources, like time, money, and effort.
• Well-designed research facilitates replication and generalization, verifying findings and extending knowledge and enhances the generalizability of results to broader populations.

Principles and Features of Research

• Clarity of purpose involves clearly defining research objectives, questions, or hypotheses to ensure focus.
• Rigor and validity can be ensured by establishing robust methods and procedures that minimize bias and employing reliable measures.
• An adequate sample size is statistically representative and provides sufficient power to detect effects or relationships
• Proper selection of participants ensures that the subjects are relevant to the research question and possess the necessary characteristics.
• Clear and appropriate data collection methods and tools should align with the research objectives and be suitable for gathering the required information.
• Ethical considerations mean adhering to ethical guidelines, obtaining approvals, protecting participants, and maintaining confidentiality and informed consent.
• An adequate data analysis plan defines a clear strategy for analyzing data and producing accurate interpretations and meaningful conclusions.
• Time and resource management entails planning and allocating resources like time, funding, and human resources, to ensure efficient execution.
• Flexibility and adaptability means designing the study to accommodate unforeseen challenges during the research process.
• Communication and dissemination involves planning for effective sharing of research findings, which includes publishing and presenting reports.

Non-Experimental Designs

• Non-experimental research does not involve the manipulation of independent variables or the control of extraneous variables.
• These designs focus on observing and describing relationships or differences without specific interventions unlike experimental designs.
• Experimental research changes variables and randomly assigns conditions to participants and can determine the cause.
• Experimental research designs are used for research in medicine, biology, and social science, have strict standards, and require resources, they can lead to very interesting results.
• Non-experimental research is descriptive/correlational without explicit changes done by the researcher.
• Validity remains a concern because it is difficult to determine causal effects

Observational Designs

• Scientific observation is performed under defined conditions, in a systematic, objective manner, with careful record-keeping.
• Direct observation involves researchers observing behavior while it occurs and can be classified as either "observation without intervention" or "observation with intervention".
• Observation without intervention (naturalistic observation) describes behavior as it normally occurs and examines relationships among variables and helps establish the external validity of laboratory findings.
• Hartup (1974) chose naturalistic observation to investigate aggression in preschoolers in St. Paul, Minnesota.
• Observation with intervention is frequently used, methods include participant observation and structured observation.
• Drawbacks of naturalistic observation are that the researcher does not have control over the variables and raises ethical concerns about invasion of privacy.

Participant and structured Observation

• Participant observation involves observers playing a dual role, observing behavior and participating actively and can disguised or undisguised.
• Drawbacks of participant observation include observer bias, the researcher's influence of participant behavior and reactivity plus expectancy effects.
• Structured observation involves researcher intervention to exert some control to the events being observed.
• Structured observations are set up to record behaviors that may be difficult to observe naturally. -Jean Piaget is known for using structured observations.

Indirect Observation

• Indirect observation methods are nonreactive, indirect or unobtrusive observations obtained by examining physical traces and archival records.
• Physical traces are the remnants, fragments, and products of past behavior and include use traces and products.
• Use traces are physical evidence from the use or nonuse of an item and are either natural-use or controlled-use.
• Products are the creations, constructions, or other artifacts of behavior
• Rozin et al. examined food products and compared portion sizes in America versus the size of products in France to test their obesity hypothesis.
• Archival records are public and private documents describing activities and comprise running records and records of specific events such as mariage certificates.
• Drawbacks of indirect observation include the potential for bias in the way traces are left and how they survive over time and selective deposit and survival.

Cross-Sectional Design

• These studies are observational, analyzing data from a population at a single point in time and used measure the prevalence of health outcomes.
• They do not follow individuals over time and are generally inexpensive and easy to conduct and are used to understand the prevalence of disease.
• They describe the distribution of variables and the association of an exposure to an outcome and can be descriptive or analytical.
• Descriptive cross-sectional studies characterize the prevalence of health outcomes.
• Analytical cross-sectional studies collect data for risk factors and outcomes and compare outcome differences between exposed and unexposed.
• Strengths include them being quick, inexpensive, ethical, and easy for generating hypotheses.
• Weaknesses include the inability to measure incidence, difficulty in making causal inferences, and susceptibility to biases.

Experimental Research Design Basic Principles

• Randomization involves randomly assigning treatments to groups, and providing an equal probability for each group to receive any of the available treatments eliminating bias and making the research a more bona fide one.
• Replication where the experiment is repeated to increase accuracy.
• Local control to increase efficiency reduce error.

Steps in Experimentation (Little and Hills 1978)

• First you identify the research problem and ensure it's specific, measurable, and researchable.
• Then conduct literature to review existing research and refine hypothesis.
• Formulate a testable prediction about the relationship between variables
• Identify the manipulated variable (independent), the measured outcome (dependent), and the factors kept constant (control).
• Decide the study structure between-subjects or within-subjects.
• Define the target population and sampling method and use random sampling or assignment to ensure generalizability.
• Obtain ethical approval from an ethics board.
• Test the procedure with a small group.
• Execute the experiment and use appropriate techniques to control extraneous variables.
• Use statistical tests to determine if the results support the hypothesis.
• Draw conclusions and assess they align with the hypothesis and consider limitations and alternative explanations.
• Share findings in a structured format, such as a paper, and include an introduction, method, results, and discussion.
• Replicate the Study to ensure findings reliability and validity and replication ensures reliability and validity of the findings.

Between Groups Design

• In this design the variable independent is tested between the subjects.
• Randomized groups design involves the independent variable being manipulated and assignment happens through random assignment.
• Each participant is part of only one group therefore it is independent of the other groups.
• A common procedure for carrying out random assignment is block randomization and is used to balance potential confoundings.
• Matched groups design or Randomized block design matching in pairs done on pretest scores assigning one to each condition.
• The independent task can be used as the variable needed to be tested.
• Methods of Matching include matching by pairs or on the basis of mean and standard deviation.

Latin Square Design

• The researcher is studying the effects of four different teaching methods (A, B, C, D) on students' memory retention, while controlling factors time of day and order of presentation.
• Each method is tested at all times (morning, afternoon, evening, night) and is presented in all session orders (1st, 2nd, 3rd, 4th).

Factorial Design

-(Between group designs for more than two independent variables:) This design is to examine how two or more independent variables (factors) influence a dependent variable and how these variables interact.

• An example is studying the effects of sleep and study method on test performance so you would:
  • Recruit participants.
  • Randomly then asign participants.
  • Control sleep duration which will determine how many hours of sleep.
  • Asign them the study method for rote memorization
  • Measure their test scores after the study session to determine results for each assigned group.

Within Group Design/Repeated Measures Design

• Participants are exposed to experiment conditions and good for reducing variability.
• Can be either complete or incomplete conditions, depending on experience.
• To ensure practice effect control material has the same level of difficuly.

Complete and Incomplete Within-Group Design

• Complete Within-Group is where each participant experiences every condition multiple times, counterbalanced within to control for order and sequence counterbalancing methods with the advantages being it controls practices and reduces fatigue for each participant.
• ABBA counterbalancing presents the conditions in one order then the reverse
• Block randomization randomizes the conditions with blocks that are then repeated over multiple conditions.
• Incomplete within groups is where each participants only experiences each once a limited number of times and conditions with the participant can be repeated to account.

Pre-Experimental Designs:

• Consist of designs which do not qualify for the experimental designs because they do not provide a control group/control threats of internal validity pre designs do incorporate the least basic elements of an experimental design: -One Shot Case Study: treatment X is given to a single group and observation O is made to assess the effect of treatment on said group
• One-Group Pretest Post-Test Design: treatment X is judged by making a comparison between pretest and post-test scores, there is no control as effects is not diagrammed.
• **Static Group Comparison: ** 2 groups are taken;
  • one group experiences experimental treatment X and another does not to compare the end results , the design may be that the dashes indicate that the control group is not randomised

Quasi Experimental Designs

• Applies experimental interpretation to results that means at least one requirement of a true experiment, creating equivalent groups.
• Still stimulates true experiments but will always be a compromise where they control some but not all extraneous variables and is often called as quasi 1 independent variable.

Types of Quasi Designs

• Time-series Design :Control/Comparison group cannot be included in the experiment but you do control for extraneous 01020304 X 05060708 where a new traffic law reduces road accidents.

Equivalent time Sample Design:

• Like time series but the group can repeat to give a systematic way the way the design is below. Steps:
  • Group of 30 students who frequently study in library is observed
• Over a week period the presence for lack of a classical music.

Non Equivalent Control Group

• Used educational Research often needs to work with intact groups because reconstitution and or subject shufflers is not allowed- The design is
  • 01 X O1 O3. O4
• Above design is similar to the pretest-posttest control group.

Seperate Post Test Design

• The separate Sample allows expirementer to assign subjects and treat them where treatment x is given but its a one way and the pre/post tet designs repeat with only pretest and posttest done the 3 factos of invalidity are still not checked.
• O1 X O2
  • O3. X O4

Patched Up Design

• The Experimenter combines features to ensure factors producing invalidity the the best they can , This design features and experiments in combination.
• **Longitudal Design-**Subjects are observed to effect the passage of time but they do not generalize over time due to anomolies and events out of the study

Longitudinal and Hoort Design

• Longitudinal design*:
• The Researcher measures a group of people to pass the time on the study and cannot account for extraneous events.

Cohort Study

• *In design the researcher conducts a longitudinal time to see a generation that happened at 10/20 years to see what was missed.

True Experimental Designs

• Have three key important factors within which include control groups, and expiremental being formed for their equivalent through randomsaztion. A few reasons these factors will not account for invalidity or invalidating data which can often cause conflict amongst the process of research.

Post Test

• There exist are 2 important groups: one group given treatment usually an expirement, the other group not treated/used in the control ; use and control often makes auto correct/check 2 extraneous, historical etc if not both groups random then subject is check not only that but the subject random is check which provides a save of money for groups test expirement has seen the treatment

A. PreTest /Post Test design

• Where the design has 2 where 1 get treatment , 1 Doesn't where the control group by default checks sources that invalidate due to history.

The Solomon Four Design

• Solomon that developed with with which is combination for the two equivalent groups the design represents is one of if not, best design due to the randomized expirement setup and it's the most known.

Ex Post Factors

• Ex Post Factors -* The researcher, will give treatment but they instead relate to it with already set to measure.
• *Criterian Approach- **