SOHP503 Lecture No.6 Design a Research Project 0125 PDF

6. Design a Research Project SoHP503 Project studies Jon Green 8th January 2025 What is this lecture about? From the problem/ question and the solution to the problem/ methodology, we now need to define the details of the study. What study design are we going to use? How big will our sample be? Wha...

6. Design a Research Project SoHP503 Project studies Jon Green 8th January 2025 What is this lecture about? From the problem/ question and the solution to the problem/ methodology, we now need to define the details of the study. What study design are we going to use? How big will our sample be? What do I need for resources and how good are they to answer the question? What are the conditions around the study? Do I need to have a control measure so that I can interpret my findings better? What exactly can I measure and how are those measures defined? When am I done? Aim: develop an 1. Learn and differentiate most common study designs and understanding about identify the most suitable depending on a research problem research design and its 2. Understand the concept of sample size and the components. Based on a theoretical background research question/ 3. Explore methods to define an appropriate sample size for problem you should be an UG research project able to define the study 4. Identify resources necessary to conduct a research design, including all project elements. 5. State and define all study conditions 6. Identify necessary control measures 7. Define endpoints and outcome measures 10/01/2024 SoHP503 No.6: Design a Research Project| Jon Gree 2 n Review – what do you know? Scientific theory Literature search/ review Question – PICO / PEO / SPIDER Aim/ objectives/ hypothesis Population/ sample? Rationale/ impact on population Data collection method Definition of success/ endpoints? Ethical considerations 15/01/2025 SoHP503 No.6: Design a Research Project| Jon 3 Green Why choose a particular design? Design = recipe describing the components of an investigation and how they will be applied - i.e. detail of which methods will be used and how e.g. placebo trial, clinical chart review, cluster randomisation, focus groups, semi-structured interviews, online (survey) - Choices may be influenced by specific constraints/ opportunities of logistics, practicalities, politics etc. - Should be guided by – established strategy/ investigative framework (methodology) with strong track record e.g. randomised controlled trial - (see critical analysis checklists e.g. CASP, STROBE, equator, SIGN etc.) - Which itself will sit within/ across, and be guided by, research paradigms e.g. positivism 15/01/2025 SoHP503 No.6: Design a Research Project| Jon Green 4  What makes research meaningful? Episteme – Ology Knowledge – Science of How can know about things? 15/01/2025 SoHP503 Introduction | Dr Daniela Oehring 5  Worldviews/ Philosophical paradigms (epistomomological/ ontological) Post-positivism/ Realism Pragmatism Determination Action consequences Reductionism Problem-centred Empirical observation and measurement Pluralistic Theory verification Real-world practice Generally Quantitative 10/01/2024 SoHP503 No.6: Design a Research Project| Jon Gree 6 n Constructivism Advocacy Interpretivism/ Relativism / Participatory Contextual Political Social/historical construction Empowerment issue- oriented Theory generation Collaborative Multiple participant meanings Change-oriented Generally Qualitative Adapted from table published in: Creswell J. W. (2008). Research Design: Qualitative, Quantitative, and Mixed Methods Approaches (3rd ed). Sage: California 15/01/2025 SoHP503 Introduction | Dr Daniela Oehring 7 Broad division of strategies of inquiry/ methodologies Quantitative Qualitative Mixed Methods Experimental Narrative research Sequential Non-experimental Phenomenology (IPA) Concurrent e.g. surveys Ethnographies Transformative/ Action Quasi-experimental Grounded theory e.g. Cross-sectional Case study Service evaluation* Audit* *Not considered Research (with a capital R) – instead ‘quality improvement’ 11/01/2023 SoHP503 No.6: Design a Research Project| Jon Gree 8 n  Qualitative research Gather/ analyse non-numerical (descriptive) data May be naturalistic/ observational Understand individuals' social reality Including attitudes, beliefs, and motivations – social phenomena Hypothesis generation & contextualisation/ significance of quantitative findings Employment of rigorous/ robust strategies and methods – not just dense description! Outcomes – Transferable (generalisable?), Trustworthy, Transparency, Validity, reliability 11/01/2023 SoHP503 No.6: Design a Research Project| Jon Gree 9 n  Qualitative Research methods Interviews Focus groups Diary methods Observation Thematic analysis (independent methodology?) e.g. reflexive (Braun & Clarke) Content analysis – communication patterns Discourse/ Conversation analysis 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 10 n  Quantitative/ experimental research Obtain empirical data - Validity/ reliability? Develop, refine or test a theory Applied, experimental, non-experimental Descriptive/ Exploratory Experimental/ Quasi-experimental Describe populations (epidemiology) Randomised Controlled Trials Find relationships May be Naturalist/ clinical/ pragmatic Diagnostic accuracy Cross-sectional Single subject/ cohort Parallel groups (Propensity matched scoring?) Case control/ series ‘quasi-experimental’ (no randomisation) Secondary analysis One group e.g. test - re-test Survey Association Causation 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 11 n Quantitative research methods Randomisation – computer generated, envelopes Blinding - participant, clinician/ data collector, analyst/ adjudicator Single, double, triple Use of validated measurement instruments e.g. surveys Intention to treat analyses Power calculations to estimate required sample size Descriptive & inferential statistics Intervention specification Control groups Control measures Prospective vs retrospective Descriptive vs inferential statistical analysis - between groups/ association 11/01/2023 SoHP503 No.6: Design a Research Project| Jon Gree 12 n Study conditions/ trial arms Comparison of alternative interventions Must be clearly described for reliability Comparison with control Usual care Sham intervention/ placebo 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 13 n  Control measures Spectrum: Optimal control of all variables except the one which is changed (laboratory/ basic science) – universal truths Pragmatic trials in naturalistic clinical settings – real world (can still be robust, but likely to require larger sample size to see beyond confounding factors) Observation 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 14 n  Outcome measures and endpoints Outcome measures Endpoints Patient reported outcome measures Clinical significance Survival (good) Statistical significance HCP contacts a priori (deduced/ predicted) Ambulance attendance vs post hoc criterion/ testing Reduced pain Bonferroni adjustment Increased strength/ mobility Improved eyesight Weight gain/loss Reduced pathology 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 15 n Sampling methods (quant or qual) Probability Non-probability Simple random Convenience Systematic Quota Stratified Purposive Cluster Snowball/ referral 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 16 n  Sample size Two questions: 1. What is the optimal sample size needed to be confident of producing scientifically meaningful/ valid results? 2. What is the maximum sample size that is logistically feasible within the scope of a undergraduate group project? 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 17 n  Optimal sample for…. Qualitative research – saturation – homogeneity? Systematic review Scoping review Observational studies (cohort/ cross-sectional/ case control/series) Experimental/ quasi-experimental Survey Audit/ service evaluation 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 18 n  Power calculations for quantitative research AS AN UNDERGRADUATE STUDENT YOU ARE NOT EXPECTED TO BE ABLE TO CARRY OUT A POWER CALCULATION! However….power calculation is a key concept in quantitative research, so you should have some understanding of them and why they are so important. Power calculation = estimate of minimum sample size to have confidence of avoiding type (I &) II errors. If the authors of a study fail to persuade you that they have an adequate sample size, the you should not accept their results! 08/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 19 n Factors affecting power calculations 1. Variance/ standard deviation/ precision/ margin of error of measurements Larger sample = more precision (narrowing CI) – estimate from previous studies 2. Size of clinically significant difference Larger difference requires larger sample size 3. Likelihood of type I or II error Type I – typically 0.05 (probability of type I error = 5%) Type II – typically 0.8-0.9 (probability of a type II error = 20%/10%) 4. Type of statistics used e.g. parametric? 5. Prevalence of event of interest 6. Design effect (adjustment for deviation from simple random sampling) 7. Expected withdrawals, missing data & losses to follow-up Usually calculated for primary outcome 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 20 n  Research Resources Time Travel IT Expert advice Equipment Printing Administration Venue 11/01/23 SoHP503 No.6: Design a Research Project| Jon Gree 21 n Any Questions? Thanks  [email protected]