(2) Rationale, Q\'s & Hypotheses PDF

Preview

Summary

Download (2) Rationale, Q's & Hypotheses PDF

Description

QUANTATATIVE: LECTURE 2- RATIONALE, RESEARCH QUESTIONS AND HYPOTHESES: Rationale: - Why are you conducting the research? - Important for proposals and grant writing. -

Based on theory and research. - Extends from critical appraisal and evaluation on research. Applicable to the wider social context. - Needs clear implications for knowledge, practice and/or impact. Needs to be evidence based, using an appraisal tool. What are the implications?

Research questions, and how they extend from the rationale: - It will follow from the rationale, needs to incorporate the key findings or gaps in the literature. - The wording needs to be precise. - the influence of… (relationship or association study) - the effect of… (direct effect of the IV on the DV) - Exploratory or explanatory? - impacts how concise the focus and wording should be. - If it is relevant, directly relate to theory. - i.e., Fear and functional form (Shen & Dillard, 2014): based on inverted U theory (Yerkes & Dodson) Hypotheses: - Builds on RQs but allows for predictions. - association v causal? - directional? (lots of research and literature) - It is not always statistical (at first) - different statistical tests can be used to answer the same question. - It requires: - clear and unambiguous IV and DV - answerable - measurable and operationalizable. - Pre-specification of a hypothesis is good practice. Appropriate design: Types of design: Observational (association): - cross sectional design - longitudinal (cohort) - naturalistic observation Experimental (comparison): - randomised control trial - between participants - within participants (or crossover) - Quasi experiment Mixed designs (some combinations of above, but less likely to be used. Observation or Experiment?

-

-

-

-

QUANTATATIVE: LECTURE 2- RATIONALE, RESEARCH QUESTIONS AND HYPOTHESES: Does it answer the question? - some designs naturally fit with a given question… - consider examining the impact of an intervention. Consider the claims you intend to make… - correlation/causation - applicability of findings Practicality… - financial and time constraints - ethical barriers Validity/reliability of measures implemented… - specifically content and ecological validity - are the measures that you choose appropriate to the design?

Operationalisation: - How to measure the variable of interest. - there are numerous methods for achieving this. - Consider definitions, the operationalised definition needs to match the design. - Consider previously established/validated measures. Formulating a quality design: Design elements: participants, IVs, outcomes. Measurement error: - Psychologists are interested in “latent” variables (inferred not observed- such as personality) - They always contain error, but this can be reduced due to varying forms of validity and reliability. - “triangulation” could be useful (multiple methods) Increasing quality = reducing error - Can use the PIDO framework. Participants and sampling: - Selection bias: potential participants from populations have unequal chance of participating. - researchers happen to select certain individuals (opportunity sampling) - disproportionate - certain individuals are more likely to “self-select” - Ideally you need random or stratified sampling. - Participant screening: you have to decide who in included or excluded. - inclusion/exclusion criteria. - screen for ethical design reasons: specific population of interest or responses may differ from the general population (you may be looking at fear and how the general population experiences fear; so, you would potentially want to exclude those with phobias for example, they may have a disproportionate fear response) - How will you determine eligibility? - standardised scales - measure objectively (try to avoid self-report) - Sample size: how many do you need? - it is directly related to study power (your sample has to be based on the statistical analysis as results can vary) - over or under sampling affects the likelihood of finding a significant effect in one test over another. - How do you find an appropriate sample size? - examine/imitate studies that are similar.

-

QUANTATATIVE: LECTURE 2- RATIONALE, RESEARCH QUESTIONS AND HYPOTHESES: - use a stats textbook: tell you an approximate number of participants for each statistical test and each number of variables included within the study. - conduct a power analysis. Participant attrition/response rate: bias in the individuals that do not complete the studies (more common in longitudinal studies) - you can overcome this by over-recruiting participants to account for the likelihood of dropouts, based on previous power analysis (often 10% over-recruitment) - be transparent about the number of participants initially sampled. - try and determine if you have any information on the characteristics of non-responders (when did they drop out, why?)

IV- Operationalisation and Validity: - Conduct a comprehensive review of previous interventions. - Examine previous validity checks. - Conduct reliability/validity checks. “Blinding” and social desirability: - Blinding: concealment of group allocation to minimise biased Reponses. - for participants and researchers. - assess the extent of social desirability. - This is achieved by: - using a method of group allocation (computers or a third party) so you are unaware of the groups. - design the study to minimise participant knowledge of the groupings. - assess participant knowledge of the manipulation (eg: funnelled debrief) - IV: manipulation checks: - how do we test if the group allocation was effective? - demographics - pre-test/post-test difference scores. - “triangulate” by measuring the factor - Manipulation checks are being questioned regarding their impact (Hauser, Ellsworth & Gonzales, 2018). - good cover story (stop participant questioning) - unobtrusive and/or (possibly) continuous measuring, to allow people to forget what factor you are measuring (eg: measure galvanic skin response based on fear instead of giving a questionnaire after experiences a fearful stimulus as the participant could alter their answers based on the knowledge that the study is looking at their fear response- instead measure while the fear is being experienced for It to be more reliable) - question if the response can be adjusted by the participants (could lie in questionnaire, but not in behavioural measurements ^^) - measure the check after the DV. - run a manipulation check in a pilot study, to increase validity. Outcomes (Mediating and moderating variables): - Additional sources of error: - spurious correlations (there is a third variable e.g.: ice cream consumption and shark attacks- at face value to relate, but it is because of heat; the more sun, the more going out and eating icecream at the beach = more shark attacks) - when you control extraneous variables its stronger in design. - check for previous behaviours (people may drink more alcohol by chance and so if they’re all in the pro-alcohol group, amount will be higher and past experiences are likely to effect the results)

-

QUANTATATIVE: LECTURE 2- RATIONALE, RESEARCH QUESTIONS AND HYPOTHESES: Theory of planned behaviour, attitudes, beliefs, social norms are all likely to relate to a particular outcome based on intention. - put controls in place for these variables prior to study....