Resources
The Breakdown
Important
- Interpreting Quasi-Experimental Designs (Post-Test with Control Group) – When presented with a figure showing a treatment and control group with only a post-test, it is crucial to recognise the design and note the absence of a pre-test, which implies a lack of baseline comparison [17:49].
- Interpreting Quasi-Experimental Designs (Pre-Posttest with Control Group) – To interpret results from a pre-posttest design with both treatment and control groups, one must compare changes within the treatment group (pre-test to post-test) and then compare the post-test results between the treatment and control groups to establish the treatment’s effectiveness [20:30].
- Informed Consent Form Components – This document is a must in research and should include the study title, purpose, procedure, duration, potential risks, benefits, compensation, withdrawal procedure, confidentiality statement, contact information (including ethics board details), and a clear consent statement for participants to agree or disagree to participate [14, 16:50, 17:32-19:59].
- Debriefing Form Components – An important document provided after study completion, it should outline the study’s purpose and hypotheses, potential impact, materials for further reading, participants’ right to withdraw data post-completion, contact information, and details for mental health or emotional support if needed [15, 20:09, 20:37-22:05].
- Common Survey Process – The typical sequence for conducting a survey starts with a recruitment notice, followed by an informed consent form, then a sociodemographic form, the main questionnaires, and finally, a debriefing form [16, 22:50-23:50].
Core concepts
- Experimental Research: A research approach where an independent variable (IV) is manipulated to determine a causal relationship with a dependent variable (DV). It involves random assignment or matched groups, provides more control, and typically has higher internal validity.
- Non-experimental Research: A research approach that does not involve manipulation of variables and cannot establish causal relationships. It focuses on showing relationships between predictor and outcome variables, has less control, and generally offers more external validity.
- Quasi-experimental Research: Similar to experimental research in that it involves IV manipulation and aims to establish a causal relationship, but it lacks random assignment or counterbalancing, often due to ethical or practical reasons. This absence of random assignment impacts internal validity.
- Between-subject Design: A research design where different groups of participants are tested in only one single condition (e.g., one group receives treatment, another acts as control).
- Within-subject Design: A research design where the same participants are tested across all conditions.
- Confounding Variable: An extraneous variable that correlates with both the independent and dependent variables, potentially obscuring or altering the true relationship. Strategies to manage them include random assignment, matched groups, or statistical adjustment (e.g., using covariates).
- One-group Quasi-experimental Designs:
- One-group Posttest: A design where participants receive a treatment, and then the outcome is measured once. It lacks a baseline measurement, making it difficult to ascertain if observed changes are due to the treatment.
- One-group Pre-posttest: Involves measuring the outcome variable before (pre-test) and after (post-test) a treatment in a single group, allowing for a comparison of changes from baseline.
- Interrupted Time Series Design: Characterised by multiple measurements of a dependent variable taken over a period before and after an intervention, allowing for the observation of changes over time and assessment of long-term effects and history effects.
- Control Group: A group in a study that does not receive the experimental treatment or intervention. It serves as a baseline for comparison to determine if observed effects in the treatment group are indeed due to the independent variable, thereby enhancing internal and external validity.
- Non-equivalent Group Designs: Quasi-experimental designs that involve multiple groups that are not randomly assigned, meaning their characteristics might not be perfectly balanced.
- Posttest with Control Group: A design with a treatment group and a non-equivalent control group, where the outcome is measured for both groups only after the treatment is administered to the treatment group.
- Pre-posttest with Control Group: Both a non-equivalent treatment group and a non-equivalent control group are measured before (pre-test) and after (post-test) the treatment is applied to the treatment group.
- Interrupted Time Series Design with a Control Group: Extends the interrupted time series design by including a non-equivalent control group that also undergoes multiple observations but receives no intervention, further strengthening causal inference by accounting for external events.
- Pre-posttest Design with Switching Replication: Involves two groups; one receives treatment, then the other (initially a control group) receives the same treatment at a later point, with observations throughout.
- Switching Replication with Treatment Removal Design: A design where the treatment is administered to one group and then removed, while the same treatment is applied to the second group, with observations taken at various points.
- Ex post facto Design: A non-experimental design used to investigate relationships after an event or condition has naturally occurred, where the independent variable cannot be manipulated due to ethical or practical reasons (e.g., studying individuals with pre-existing conditions). It carries disadvantages such as no control over the IV, non-random assignment, and the presence of confounding effects.
- Survey: A method for collecting data from a predefined group, typically through questions, to gather information and insights. It is explicitly NOT a research design. Surveys are commonly used in non-experimental research, both quantitative (cross-sectional, longitudinal) and qualitative, and often employ self-reports (questionnaires or interviews).
- Cognitive Processes in Survey Response: Describes the mental steps participants undergo when responding to a survey item, including question interpretation, information retrieval from memory, judgment formation, response formatting, and potential response editing.
- Context Effects on Questionnaire Responses: Phenomena that can influence survey answers, such as the item-order effect (responses changing based on preceding items) and the tendency for people to choose middle response options to appear normal or typical. Recall bias can also occur if participants incorrectly remember past events.
- Types of Survey Items:
- Open-ended Items: Questions that provide no pre-defined response options, allowing participants to provide detailed, unbiased written answers, though they can be time-consuming to complete and analyse.
- Closed-ended Items: Questions with pre-defined response options, making them quicker to complete and easier to analyse. They include binary (two options, e.g., Yes/No) and Likert scale (multiple points measuring degree of agreement/disagreement) formats.
- Types of Scaling:
- Guttman Scale: A cumulative method where a positive answer to a more difficult item implies positive answers to all easier items. It typically uses binary responses.
- Thurstone Scale: Measures agreement or disagreement on a statement using a binary response type (e.g., Agree/Disagree), often considered a nominal measurement level.
- Likert Scale: Measures attitudes by capturing the degree of agreement or disagreement with a set of statements, using an ordinal or continuous response type (e.g., 5-point scale from “Strongly Disagree” to “Strongly Agree”). More response options generally increase reliability.
- Effective Survey Item Characteristics: For an item to be effective in a survey, it should be brief, relevant to the research, clear in its meaning, specific (avoiding multiple thoughts in one statement), and objective.
- Questionnaire Formatting in a Survey: A well-structured questionnaire includes an introduction, an informed consent form, items grouped logically by topic or type (e.g., Likert, binary), and a debriefing section at the end.