Please enable JavaScript.

Coggle requires JavaScript to display documents.

Research Methods, Generalizablity, More generalizable, Less generalizable …

- - - - Independent variable (IV): manipulated by the researcher
      - Dependent variable (DV): expected to change as IV changes
      - Allows for cause-and-effect inferences, as confounding variables are strictly contolled.
      - Confounding variables: can potentially distort relationship betweem the IV and DV.
      - .
        
        EXPERIMENTAL DESIGN
        
        Independent Measures Design: random allocation of participants into groups and a comparison between these groups.
        
        Experimental group and control group
        
        Potential confounding variables cancel each other out.
        
        The larger the sample, the higher chance that the groups are equivalent.
        
        Matched Pairs Design: similar to independent measures design, but here researchers use matching to form the groups. Expected confouding variable is tested prior to the experiment, and participants are paired together. Each pair is split to a random group.
        
        Good with smaller samples.
        
        Matching variable: the variable that is controlled (for example, memory abilities, language proficiency)
        
        Repeated Measure Design: used when the goal is to compare conditions rather than groups of participants. The same group is exposed to two (or more) conditions, and the conditions are compared.
        
        Order effect: results may be different depending on which condition comes first.
        
        Practice - participants practice, improve their on-task concentration and become more comfortable with the task. Their performance increases in the second trial.
        
        Fatigue - participants get tired during the first trial, and their concentration decreases. Their preformance decreases in the second trial.
        
        Counterbalancing - using two groups where the trials are switched. Data from the two groups is combined, and the conditions are compared.
        
        Comparison is more reliable, as it overcomes the influence of participant variablilty.
        
        CREDIBILITY AND GENERALIZABILITY
        
        Construct Validity: the quality of operationalizations.
        
        When results are interpreted, it is linked back to constructs. Moving from operationalizations to constructs is a bit of a leap.
        
        High validity when the leap is justified and the operationalization provides sufficient coverage of the construct.
        
        Internal Validity: the quality of methodology of the experiment.
        
        High validity when confounding variables have been controlled, and it is certain that the change in DV was caused by IV.
        
        Bias is a threat to internal validity.
        
        External Validity: the generalizability of the experiment.
        
        Population validity: the extent to which findings can be generalized from the sample to the target population.
        
        High validity when the sample is representative and appropriate sampling technique was used.
        
        Ecological validity: the extent to which findings can be generalized from the experiment to other settings or situations.
        
        Low validity when experiments are strictly controlled.
        
        Credibility and generalizability are rarely used, insted the quality is characterized by validity.
        
        BIAS
        
        History: outside event that happened to participantsin during the course of the experiment.
        
        Maturation: during the experiment participants go therough natural developmental changes, such as fatigue or growth. Counter - control group.
        
        Selection: if groups are not equalivent at the start of the experiment. Occurs in independent measures and matched pair design if allocation was not completly random.
        
        Testing effect: first measurment of DV can effect second measurment. Counter - control group.
        
        Instrumentation: instrument measureing the DV changes slightly between measurments. Counter - standardized measurment conditions.
        
        Regression to the mean: the initial score on the DV is extreme, and it is expected that it will naturally go down with measurements. Counter - control group.
        
        Experimental mortality: if participants who drop out during an experiment are not random.
        
        Demand characteristics: participants understand the purpose of the experiment and change their behaviour subconsciously to what they think is expected. Counter - deception.
        
        Experimenter bias: researcher unintentionally affects the results. Counter - double-blind design where information that could introduce bias is withheld from both participants and people conducting the experiment.
      - Quasi-Experiment: allocation into groups is not done randombly, but based on pre-existing inter-group difference.
        
        Cause-and-effect inferences cannot be made.
        
        Pre-existing differences could be age, gender, cultural background.
        
        Natural experiment: conducted in participants' natural environment and researcher has no control over the IV. Used when it is unethical to manipulate the IV. High - ecological validity. Low - internal validity.
      - Field experiment: conducted in a real-life setting. IV is controlled, but many external variables cannot be controlled. High - ecological validity. Low - internal validity.
    - - The researcher does not manipulate any variables (no IV or DV)
      - Variables are measured and the relationship between them is quantified.
      - Cause-and-effect inferences cannot be made, as it is not certain what influenced the results.
      - There are trends in the data: positive or negative and weak or strong.
      - CORRELATION
        
        Correlation: measure of linear relationship between two variables.
        
        Negative correlation: as one variable increases, the other decreases.
        
        Positive correlation: as one variable increases, the other increases too.
        
        Correlation coefficient: measure of how correlated two variables are (-1 to +1).
        
        Strong or weak correlation: the more steeper line, the stronger the correlation
        
        Statistical significance: the likelyhood that a correlation has been obtained by chance.
        
        Correlation estimates with larger samples are more reliable.
      - LIMITATIONS
        
        Spurious correlations: when calculating multiple correlations between multiple variables, there is a possibility that some of the statistically significant corrations would be the result of random chance.
        
        The third variable problem: a possibility of another variable that correlates with both X and Y and explains the correlation between them.
        
        Curvilinear relationships: when variables are linked non-linearly. As one variable increases, the other one increases too but only to a certain point and then it decreases. The problem comes when optaining correlation coefficient, that would be small or medium.
      - GENERALIZABILITY
        
        Directly linked with sampling and depends on representativeness of the sample. (similar to population validity in experiemnts)
      - CREDIBILITY
        
        Connected to bias in variable measurment and in interpretation of findings.
    - - Relationship between variables are not investigated.
      - Variables are approached separately.
  - - - Random Sampling: every member of the target population has an equal chance of becoming part of the sample.
        
        If sample is sufficient in size, the results are easily generalizable.
        
        Impossible if target population is too big
      - Self-selected Sampling: recruiting volounteers.
        
        Quick and relatively easy, while having wide coverage.
        
        Low representativeness.
      - Convenience (opportunity) Sampling: recruiting particiapnts that are more easily avaliable
        
        Good when financial resouces and time are limited.
        
        When people are not that different in terms of the phenomenom studied
        
        Low representativeness.
      - Stratified Sampling: essential characteristics the sample has to reflects are decided first, and then their distribution is studied in the target population. Participants are then recruited in a way that the proportions of the characteristics is the same in the sample and the target population.
        
        Theory-driven
        
        Researchers are certain about essential participant characteristics
        
        Avaliable sample sizes are not large
- - - - Theoretical Sampling: purposive sampling that stops when the point of data saturation is reached.
        
        Data saturation: no new information is obtained from new participants added to the sample.
  - - - Method triangulation: use of different methods. If the same reults are obtained using vaiorus methods, credbility increases.
      - Data tirangualtion: using data from a variety of accessible sources.
      - Theory triangualtion: using multiple perspectives or theories to interpret the data.
      - Researcher triangulation: combining observations/interpretations of different researchers.
    - - Researcher should remind participants about voluntary participation and the right to withdraw.
      - It should be made clear that there are no right or wrong answers
      - Good rapport ensures that participants are honest and alter their behaviour in the presence of the researcher as little as possible.
    - - Epistemological reflexivity: the strenghts and limitations of the method used to collect data.
      - Personal reflixivity: the personal beliefs and expectations of the researcher.
      - Researchers should reflect on the possibility that their own biases might have interfered with the observations or interpretations.
  - - - Confirmation bias: researcher has a belief and uses the research in an unintentional attempt to confirm that belief. Solution - reflexivity
      - Leading question bias: responders in an interview are inclined to answer in a certain way because the wording of the question encurages them to do so.
      - Question order bias: when responses to one question influence responses to the following questions.
        
        General questions before specific,
        positive questions before negative,
        behaviour questions before attitude.
      - Sampling bias: the sample is not adequate for the aims of the research.
      - Biased reporting: when some findings of the study are not equally represented in the research report.
    - - Acquiescence bias: the tendency to give positive answers whatever the question.
        
        Avoid leading questions, make questions open-ended and neutral.
      - Social desirability bias: participants' tendency to respond or behave in a way that they think will make them liked or accepted, especially in researching sensitive topics.
        
        Questions should be phrased in a non-judgmental way, or questions asked about a third person.
      - Dominant respondent bias: in group interview, when one of the participants influences the bahviour and responses of the others.
        
        Researchers should keep dominant responders in check.
      - Sensitivity bias: tendency of participants to answer regular questions honsetly, but distort their responses to questions on sensitive topics.
        
        Building rapport and creating trust between the participant and researcher.
  - - - REASON TO
        CHOOSE
        
        Focus: studying how people interact, iinterpret each other's behaviour and act upon those interpretations in a natural setting.
        
        Researcher believes that meaningful knowledge in the research area cannot be generated without observation.
        
        Allowes researcher to become immersed deeply into the studied phenomenon.
      - Naturalistic VS Laboratory: naturalistic observation is carried out in naturally occuring setting, and is often the only choice (ex. when it is unethical to arrange settings for the bheaviour of interest to occur).
      - COVERT VS
        OVERT
        
        Covert observation: when participants are aware that they are being observed.
        
        Strength - informed consent
        
        Limitation - participant bias related to expectations.
        
        Overt observation: when participants are not informed about the reason of researcher's presence.
        
        Strength: Gaining access to groups that would not agree to participating. Avoidance of participant bias.
        
        Limitation: ethical considerations, participants did not give their consent to participate. Counter - debrefing and ask for consent for using data.
      - Participant observation: observer becomes part of the observed group.
        
        Strenght: allow for first-hand experiences with phenomenon of interest.
        
        Limitation: losing objectivity, no consent
      - STRUCTURED VS
        UNSTRUCTURED
        
        Structured: information is recorded systematically and in a standarized way (ex. checklist, pre-defined time intervals).
        
        Operates with numbers rather than text.
        
        Unstructured: do not have pre-defined structure and observers register whatever they think is noteworthy.
    - - In-depth investigation of an individual or a group.
      - SEPERATE
        RESEARCH METHOD
        
        The individual or group studied is unique in some way, and the purpose is to gain a deep understanding of the individual or the group.
        
        Less focus on generalizability; findings do get generalized, but as a by-product,
        
        Sampling is not an issue.
        
        The case is studied throughly, using a combination of different methods, and often longitudinally.
      - REASONS TO
        CHOOSE
        
        To investigate a phenomena that could not be studied otherwise.
        
        Can contradict established theories and help develop new theories
      - LIMITATIONS
        
        Researcher bias due to longitudinal nature of the study
        (researchers may get too involved)
        
        Participant bias due to longitudinal nature of the study
        (aquisence, social desirability)
        
        Generalization, from case to other setting or wider population.
        
        Demanding in terms of anonymity and confidentiality.
    - - CONTENT
        ANALYSIS
        
        Inductive content analysis: approach to analysing texts produced by interviews.
        
        Data: audio or video recording which is then converted to an interview transcript. Sometimes includes interview notes on observations about the participant and the interview context.
        
        Writing the transcript.
        
        Reading the raw material several times.
        
        Lol-level themes are grouped into a smaller number of high-level themes.
        
        A summary table of themes is prepared.
        
        Conclusions are formulated based on the summary table.
      - REASON TO
        CHOOSE
        
        Individual individual interviews are useful for sensitive topic
        
        Used to understand the meanings participants attach to certain events and their points of view.
        
        Used to get an insight into the nature of subjective experiences and interpretations.
      - Semi-structured: order or set of questions is not specifyied, but there are certain questions that must be asked.
      - Unstructured: participant-driven, subsequent questions are based on the participant's answer.
      - Structured: fixed list of questions that need to be asked in a fixed order.
        
        Useful for big scale research, since it is standardized so it is easy to have many interviewers
      - Focsu group: semi-structured group interview that is conducted with 6-10 people. Participants are encouraged to interact with each other.
        
        Strengths - quick way to get information from several participants, more natural and comfortable setting, multiple perspectives are discussed.
        
        Limitations - dominant participants, dificullity with preserving anonymity and confidentiality, demanding in terms of sampling and transcripting.
- - - - Mental Disorders: accidental reveal of the presence of illness that was previously unknown may lead to negative consequences.
      - Genetic Research: can lead to revealing private information to one individual about other members of the person's family. Such information must be handled with care and sensitivity, and help shoudl be offered if needed.
    - - Possible effects of the way research conclusions are formulated on the scientfic community and society in general must be kept in mind.
      - Possibility of misinformation as the results may not be certain or prone to bias in a signle study.