Please enable JavaScript.

Coggle requires JavaScript to display documents.

Research into NDCs - Coggle Diagram

- - - - Construct validity - face and predictive criterion validity; gathering data from multiple methods (triangulation) can increase validity
        
        However, this can be impractical or inaccurate
        
        Can still have causation issues even without a validity issue
        
        If a test can establish causation, it has internal validity
        --> Impact of ‘history’ - other coinciding events which could explain findings over the independent variable
        
        12 threats to internal validity -
        
        History
        
        Testing - how pre-testing could impact the results by causing more thinking from participants (only exists in tests using a pre-test)
        a. Equally, using a pilot study can identify these issues
        
        Instrumentation - measurement changes
        
        Regression - if an atypical population is selected, they can perform relatively worse than their matched typical population
        
        Mortality - losing participants, sometimes due to their perception of failing
        
        Maturation - growth, change or development in participants unrelated to treatment
        
        Selection
        
        Selection by maturation interaction - predisposition of groups to grow apart
        
        Ambiguity about causal direction (reverse causality)
        
        Diffusion of treatments - receiving aspects of treatment from another group
        
        Compensatory equalisation of treatments - special treatment to one group creates pressure for a control group to receive it which could impact results
        
        Compensatory rivalry
      - Methods to offset threats - taking specific steps to deal with a threat, or use of randomisation:
        
        Non-experimental and quasi-designs have to be particularly wary of threats
        
        Causal description - controlling for possible alternative explanations to try and make sure you measure the intended variables (multi-strategy design approach gives the what and the why)
        
        Validity threats are made implausible by evidence, not methods, which try to establish the evidence used to erase these threats (positivist approach to think method rules out validity)
      - External validity and generalisability - experiments in a lab lack ecological validity
        
        If sample is representative from a known population, then the generalisation to that population can be done through statistical inference - generalisation to other groups has to be non-statistical and based on subjective probability
        
        Threats to generalisability
        
        Selection - specific to studied groups
        
        Setting - specific to study context
        
        History - specific experiences can determine effect of findings
        
        Construct effects - particular constructs studied may be specific to group studied
        
        Solving generalisability issues - direct demonstration and making a case:
        
        Direct demonstration involves somebody else extending the results by testing them in other participants, or contexts
        
        Making a case is concerned with persuading that it is reasonable to results to generalise, with arguments that the group or time is representative
        
        Attempting replication can also be useful, and sometimes necessary for particularly controversial or impactful findings
  - - - Unreliability can be caused by -
        
        Participant error - tiredness, pre-menstrual tension, and although some considerations can be made it is not foolproof
        
        Participant bias - problematic for validity - seeking to please with good results and therefore making more effort
        
        Observer error
        
        Observer bias - problems with interpretations, or they were biasing observations in way of ideological commitment
        -> Introduction of blinding assessment to combat this
- - - - Identifying a research area has three issues:
        
        Researcher bias - a narrative analysis is confounded by the researcher selecting what papers to select and how to create a story that works with their research gap
        --> Sometimes, the literature is reviewed after a study rather than before, and thus does not include all relevant research or an objective analysis
        --> Bias, beliefs, perspectives and motivations therefore influence the direction of the literature review
        --> Systematic reviews can overcome this by ensuring studies can be replicated by an independent other, but there is still researcher bias, own conclusions can be drawn
        
        Apples and pears problem - systematic reviews are introduced to answer very specific research questions and synthesising papers with the same questions, same designs and same samples
        --> Sometimes, the method used the synthesise studies uses different questions, designs and samples
        --> Variability in findings can just be variability in studies themselves
        --> Therefore, most reviews are premature and state more research is needed
        
        Premature reviews - meta-analyses and systematic reviews require similar studies with similar designs and samples, and although initial searches lead to many possible papers, the majority do not meet stringent inclusion criteria
        --> This is more practical to synthesise than many papers, but by including more papers you would erase the issue of future directions being the requirement of more research
      - Sample
        
        Not representative of a wider population - tend to use opportunistic or convenience sampling, and many use psychology students or market researchers
        
        Response bias - do not show how many responded compared to how many asked - those who respond are vastly different to those who do
        
        Under/over powered - not enough people to show an effect, leading to type 1 errors, and over powered studies lead to Type 2 errors
        
        Anecdotal - qualitative studies provide rich and detailed data, but data is not representative of a wider population
        --> Problem is tackled by re-researching the same area with different people until no new ideas emerge, and small samples are justified because all aspects of emotional experience have been captured
        --> Anecdotes may not be representative, but they provide insight
        
        Not generalisable - either to sample or to context
        
        Questions to ask - was the research actually answering something useful, and was it worth it?
        
        Was it worth the cost, impact and reach?
        --> Some studies may have low cost, and low reach, but massive impact
        --> Others have high cost and have huge reach, but little impact
        --> Could also have been a pilot study