Research Methods Flashcards
Empirical method
Using a procedure that means you are only measuring what can be directly observed (observable hard evidence)
Replicability
To be able to replicate research and get the same findings.
- this is done in a controlled and standardised approach (control of variables)
- also helps determine causation
Paradigm Shift
When an important change in the basic concepts and experimental practices of a scientific discipline occurs (e.g flat earth —> round earth)
Objectivity
The tendency to base judgments and interpretations on external data rather than on subjective factors, such as personal feelings, beliefs, and experiences.
Falsifiability
The principle that a proposition or theory could only be considered scientific if in principle it was possible to establish it as false.
Reliability
The overall consistency of a study ( can someone else repeat it and get similar results)
Validity
The ability of a test to measure what it was set out to measure
Standardisation
Where the procedures used in research are kept the same (e.g same order, same instructions)
Internal validity vs External Validity
Internal : The extent to which the effects of a study are due to the manipulation of the independent variable - with no influence of any extraneous or confounding variables. (Demand characteristics are a major threat for internal validity)
External : Howe generalisable the findings are
Paradigm
consists of the basic assumptions, ways of thinking, and methods of study that are commonly accepted by members of a discipline or group.
Hypothesis testing
uses sample data to evaluate the credibility of a hypothesis about a population.
Null hypothesis
Predicts no pattern or trend in results.
“There will be no difference/correlation…”
Experimental hypothesis
Predicts a significant difference or correlation in results between conditions.
Independent variable/ dependent variable/ control variable
IV -> change
DV -> measure
CV -> keep
Experimental designs
How participant’s are allocated in experimental condition:
- independent measures
- repeated measures
- matched pairs
Repeated measures + evaluation
Testing a participants under both conditions
Criticisms:
- may perform better on second condition due to practice effect = order effects
- may perform worse on second condition due to fatigue and boredom (order effects)
- cause demand characteristics
Pros:
- one participant does all conditions so fewer participants needed
- no problems with individual differences as same person for each condition, controlling participant variables.
Independent measures + evaluation
The participants in one condition are independent from participants in the other (only participate in one condition).
Criticisms:
- differences in conditions may be due to individual differences
- potentially more sample needed
Pros:
- no demand characteristics as they cannot compare knowledge from previous conditions
- no order effects as they do not know what the other condition is, as only sit one condition
Demand characteristics
Participants changing their behaviour/answers purposefully to either aid or hinder an experiment
Social desirability
Participants changing their answered as they wished to be liked by the experimenter
Types of sample definitions
- Random sample -> a sample selected at chance
- Opportunity sampling ->a sample selected by convenience
- Volunteer sampling -> self selected sample chosen by themselves via eg an ad.
- Systematic sampling-> involves taking every nth person from a list to create a sample
- Stratified random samples -> The composition of the sample reflects the proportions of people in certain subgroups (strata) within the target population or the wider population.
Experiment types/method + criticisms
Lab -> IV is manipulated by researcher in a controlled environment (true exp)
cons: low ecological validity as artificial , demand characteristics + experimenter effects
pros: replicable/reliable due to being well controlled, easy to establish cause and effect
Field -> IV is manipulated by researcher in a naturalistic setting (true exp)
cons: lack of control over extraneous variables so less replicable/reliable, harder to establish cause and effect, issues with getting informed consent
pros: high ecological validity so displays real human behavior, less chance of demand characteristics + experimenter effects
Natural -> IV is not directly manipulated in a natural environment (e.g piaget)
cons: lack of control over extraneous variables so less replicability/reliability, IV is not deliberately changed so we cannot claim the IV has caused any observed change
pros: high ecological validity so displays real human behavior (situational IV)
Quasi-experiment -> IV is already pre-existing (e.g age, gender) and has not been determined. IV cannot be changed unlike natural experiments .
cons: confounding variables as cannot randomly allocate participants into conditions, IV is not deliberately changed so we cannot claim the IV has caused any observed change./ small sample size therefore not generalisable(individual differences)
pros: controlled conditions so is reliable/replicable (based on the individual)
Extraneous variables
Any variable that you’re not investigating that may affect the dv
Quantitative data vs qualitative data + evaluate
Quan-> numerical data
Pros:
-scientifically objective, easily replicated as the data obtained does not need a lot of interpretation of results so more reliable (easier to identify patterns and trends)
-can use it to reject or accept nul hypothesis
Cons:
-require large samples to get useful data
-poor knowledge of stats can lead to misinterpretation of a=data
-low construct validity = simplifies complexity
Qual -> descriptive data
Pros:
- in depth/detailed therefore more information about a single case (high validity)
- can lead to possible investigations of cause and effect and relationships
Cons:
-time consuming
-expensive
-less generalisable
-no statistical tests or information (easier to understand with numbers> pages of writing)
-samples do not have a large data set affecting reliability of data as it can be subjective in nature
Observer effect
Subjects altering their behaviour when they are aware that an observer is present
Co-variables
Something that changes in relation to another variable
Overt observation( disclosed) vs Covert observation (undisclosed)
Participants are aware their behaviour is being observed and what they are being observed for = overt
Participants are unaware of the presence of the researcher and are unaware their behaviour is being observed = covert
Participant observation
Researcher observes participants while participating in their activities (e.g milligram/ bickman)
- group is usually aware of researchers presence
Non participant observation
Researcher observes participants without participating in their activities (can be either covert or overt)
- e.g OFSTED inspections at school
Primary and secondary data + evaluation
Primary = first hand info collected by researcher
Pro:
-Reliable data as it has been collected by themselves (trust it)
Con:
-Expensive and time-consuming
Secondary = pre-existing data
Pro:
-Saves time and money as data is already pre-existing
-Psychologist may have access to data they would not have been able to collect otherwise
Con:
-Untrustworthy
Experimenter bias vs investigator effect
bias -> The researcher consciously influences the results in order to portray a certain outcome
effect -> researcher unintentionally or unconsciously influences the outcome of any research they are conducting as they know the aims
Observer bias (detection bias)
Happens when the researcher’s expectations, opinions, or prejudices influence what they perceive or record in a study (usually occurs when observer is aware of research, aims and hypothesis)
6 ethical issues in psychology
- withdrawal
- debrief
- informed consent
- protection of participants
- deception
- confidentiality
I don’t pick colourful wellies daily
Types of external validity
Ecological validity
Population validity: is it generalisable to population?
Temporal validity : refers to the validity of the findings in relation to the progression of time.
Types of internal validity
Construct validity: Ability of a measuring tool to actually measure the psychological concept being studied
Questioning bias
To phrase a question to favour one view over others
Correlational study (inc neg and pos)
An analysis of the relationship between two variables .
Nothing is manipulated, two co variables are simply measured to look for an association
neg-> as one variable increases the other decreases \
pos-> as one variable increases so does the other variable /
Self report evaluations
=Methods of gathering data where participants provide information about the cells without interference from the experimenter
Pros:
-For research it is inexpensive and can reach many more test subjects that could be analysed by observation or other methods. It can be performed relatively quickly so a research can obtain results in days or weeks rather than observing a population of the course of a longer timeframe. Self reports can be made in private and can be anonymized protect sensitive information of the haps promote truthful responses.
Cons:
-Honesty: subject may make them more socially acceptable answer rather than being truthful (social desirability)
-Introspective ability: the subject may not be able to assess themselves accurately
-Rating scales: rating something yes or no can be too restrictive, but numerical skills also can be inexact and subject individual inclination to give an extremely middle response to all questions
-Response bias questions are subject to all of the biases of what the previous responses were, Whether they relate to recent or significant experience and other factors.
-Sampling bias: the people who complete the questionnaire are the sort of people who will complete a questionnaire. Are they representative of the population wish to study?
Unstructured interview vs structured interview
Unstructured interviews are the most flexible type of interview, with room for spontaneity. In contrast to a structured interview, where the questions are the order in which they are presented are not set.
Structured interviews are when the interviewer has a set of prepared closed questionsIn the form of an interview schedule, which he/she reads out exactly as worded. Interview scheduled to have a standardised format which means the same questions asked to each interviewee in the same order.
Strength/weakness of correlations
PRO
-High ecological validity as nothing is set up and manipulated(more ethical)
-Can lead to further research (causal rel) via experiment
CON
-Cannot establish cause an effect as we can only make assumptions on the relationship between variables. (may be due to other factors that one variable changes etc)
External validity
The extent to which the results of a study can be generalised to an increase of the situations, people, stimuli, and times. (Bickman high external validity)
Where is the independent variable placed on a graph
X axis (horizontal)
Nominal Data (levels of measurement)
Frequencies in categories (no mathematical value)
Ordinal Data (levels of measurement)
= Subjective/judgement scores (ie on a rating scale) which don’t have equal intervals/ difference between each of them is unknown, just the position they hold within the group
Interval/Ratio Data (levels of measurement )
= Fixed (agreed/universal) units of measurement, with equal intervals
what is a ‘true experiment’
Experiments where the IV is under direct control of the researcher who manipulates it and records effect on DV.
Thus only lab and field experiments are ‘true’ as they involve manipulation of IV by researcher.
Situational iv
Something in the environment that causes a change (e.g in a natural experiment researching the behaviour of football fans with hooliganism)
Hypothesis
A clear, concise, testable statement that states the relationship between IV and Dv to be investigated (written in future tense)
2 types:
- null
- experimental
Types of experimental hypothesis:
One tailed hypothesis/directional:
This states the specific direction the researcher expects the results to move in (e.g higher, lower, more, less and for correlations it would be positive or negative) (evidence/ previous research to support)
= the girls will do significantly better in the maths test scores compared to the boys
Two tailed hypothesis/ non-directional:
This states that a difference will be found between the conditions of the iV but does not state the direction of results. (When you have no evidence to support/ no previous research)
Operationalised
Turning abstract concepts into measurable, clear and specific observations
Order effects
When the order in which the participants are exposed to the different conditions affects the results (e.g fatigue/practice) = in repeated measures
Matched pairs + evaluation
Participants are put in pairs on a characteristic that the researcher has a reason to believe may affect the results. HAPPENS BEFORE EXP. Then one member of each pair is placed into each condition (e.g two people of the same age).
cons:
- if one half of the pair withdraws, the whole pair is lost
- matching pairs if very time consuming
- even though participants in different groups are matched on their similarities, some individual differences and participant variables may still occur.
pros:
- no order effects (different people are in each condition)
- demand characteristics are less likely
- paired individuals treated as if the same person so they can be compared and individual differences can be controlled.
standard deviation + percentages
How spread out numbers are in relation to the mean (different from range as any anomalies are included and gives us percentages of average)
- cannot be negative
-low sd = data is clustered around the mean
- high sd = data is more spread out
-1 +1 : 68.26%
-2 +2 : 95.44%
-3 +3 : 99.74%
Examples of measure of dispersion
Range - big number - small number
Standard deviation - how spread out the data is, in relation to the mean
Measures of central tendency + evaluation
Mean - add up data then divide by how many there are
PRO: most accurate as includes all data
CON: distorted by extreme values
CON: not appropriate with nominal data
Median - the middle number once put in chronological order
PRO: not affected by extreme values
CON: not as accurate as mean
CON: not appropriate with nominal data
Mode - most common value
PRO: useful for nominal data
CON: not useful if you have more than one mode
Pros + cons of range vs standard deviation
Range:
Pros: easy to calculate
Cons: distorted by extreme values
Standard deviation:
Pros: more accurate as all values are taken into account
Cons: may hide some characteristics of data (e.g extreme values) // harder to calculate and more time consuming
Graphs
Histograms ->demonstrate the ‘frequency density’ of continuous (interval/ratio) data.
- represents the frequency of each score (bell shape curve if less extreme = symmetrical, tails never touch x axis)
Scatter graphs ->used when doing a correlational (relationship) analysis .
- In a scatter graph is doesn’t matter which variable goes on which axis
- the scatter of the dots medicates the degree of correlation between the co-variables (using correlation coefficient—-> -1. -0.5. 0. 0.5. 1.
Frequency Polygon ->demonstrate the ‘frequency density’ of continuous data.
Bar charts ->demonstrate the difference in the frequencies of non-continuous (category/nominal/ordinal) data
Pie Chart ->demonstrate the frequencies of each category as a proportion (%) of the whole.
Line graphs -? Represents continuous data (iv = X axis , dv= Y axis)
Distribution of histograms
Normal distribution = symmetrical bell
Positive skew = most scores on left, elongated tail on right
Negative skew = most scores on right, elongated tail on left
Stratified sampling evaluation
PRO:
As selection occurs from sub-groups within a population and involves random sampling, selection is unbiased and therefore representative, making it easier to generalise findings.
CONS:
- Time consuming dividing population into stratums
- Detailed knowledge of population characteristics are required for stratified samples which may not be available.
Random sampling evaluations
PRO:
- More likely to be representative and not biases to a particular type of person as each individual has the same chance of being selected, therefore easier to generalise findings to wider target population, meaning it has a high population validity.
CONS:
- time consuming
- chance of representation decreases as the sample size increases (therefore may not be representative to wider target population)
Systematic sampling evaluations
PRO:
- unbiased and representative sample as results are representative of the population unless certain characteristics are repeated every nth person which is highly unlikely
CON:
- unbiased selection can only be ensure as long as the researcher has not intentionally biased the sampling system to only include certain types of people
Opportunity sampling evaluations
PRO:
- Efficient and convenient as researcher is not required to identify all members of the target population (unlike in random sampling)
CON:
- Unrepresentative of target population as researcher may only ask people who look friendly/approachable.
- At certain times there might be certain types of people available which may bias the results. Therefore making it difficult to generalise findings to the wider target population, so may lack population validity.
Volunteer sampling evaluations
PRO
- easier to organise, less time consuming, less effort for researcher
CON:
- Bias as only certain types of people are likely to offer their time to undertake a research study (e.g people who are interested in psychology or need the money) as a result, it can be very difficult to generalise findings, so lacks population validity.
correlation coefficient
when the relationship between variables is given a numerical value between +1(strong positive) and -1(strong negative)
= 0 being weak
- shows the strength of a relationship
0 ie the lowest
+1
-1
normal distribution
= symmetrical spread of frequency data that forms a bell shaped curve
- The most people/items are located in the middle area, with very few at the very ends.
- The mean, median and mode located at midpoint of curve// all at highest point
- The tail of the curve never touches the horizontal X axis, so never reaches zero (more extreme values are always theoretically possible).
EG may see a normal distribution when measuring certain variables like heights of all ppl in college
positive skew
=A type of frequency distribution in which the long tail is on the positive (right) side of the peak and most of the distribution is concentrated on the left..
- The mode remains at the highest peak, then the median and mean is dragged across towards the tail .
- used when the mode is the SMALLEST value out of median and mean
EG (and may be the result of a very difficult test). may see a positive skew in a very difficult test where most students score very low, with only a handful of students scoring highly. = REMEMBER -> extreme scores affect the mean (the high scoring students have dragged the mean towards the right). mode and median less affected.
!happy whale watching tv!
negative skew
= A type of frequency distribution in which the long tail is on the negative (left) side of the peak and most of the distribution is concentrated on the right.
- The mean, median and mode are the same as a positive skew.
- used when the mode is the LARGEST value out of mean and median.
EG (the result of a very easy test). may see a negative skew in a very easy test where most of the students score very high, with only a handful scoring low (these ppl drag down the mean towards the left).
!sad whale not watching tv!
skewed distribution
A spread of frequency data that is not symmetrical, where the data clusters to one end
informed consent (def, issue, overcome, problems)
= Participants should be given sufficient information (rights, aims, procedure and what their data will be used for) so they can consider whether they wish to participate prior to taking part (U16’s must have parental consent)
con: demand characteristics which may invalidate results/difficult to initially obtain informed consent in some research methods (e.g naturalistic observations).
overcome+ problem:
1) presumptive consent, but difficult to envisage (contemplate) how they would actually feel in certain situations.
2) prior general consent, but demand characteristics as participants may be looking out for deception
3) obtain full consent (same as con)
4) debrief, but does not justify for a lack of ethical consideration / difficult to return a person’s psychological state prior to study
deception (def, issue, overcome, problems)
= Deliberately misleading or withholding information from participants
cons: demand characteristics if not deceived
overcome+ problem:
1) presumptive consent, but difficult to envisage (contemplate) how they would actually feel in certain situations.
2) prior general consent, but demand characteristics as participants may be looking out for deception
3) debrief, but does not justify for a lack of ethical consideration / difficult to return a person’s psychological state prior to study
protection of participants (def, issue+ pro, overcome, problem)
= Participants should not be exposed to any physical or mental harm greater than which they encounter in their day to day lives.
con: hard to predict so researchers may be taken by surprise, BUT short-term harm caused may lead to long term benefits to society
overcome+ problem:
1) terminate research , but all research is thrown away so pointless
2) debriefing, ensures they are back to their same mental state + keep an eye on them if not
3) only ‘everyday’ stress, but difficult to know what each participant’s everyday stress is - too subjective
confidentiality (def, issue, problem, overcome)
= The right of participants to have their data, personal information and privacy protected
con: cannot always be guaranteed to participants
overcome + problem:
1) no names (numbers given instead), but sometimes confidentiality must be broken in cases where clients or others are in danger
withdrawal (def, issue, overcome, problem)
=Participants should be fully aware that they are free to leave the study at any point
con: participants may be too embarrassed/afraid to withdraw/ may not understand they can withdraw
overcome + problem
1) right to withdraw by using a standardised procedure, but even if participants know they can withdraw they might feel too uncomfortable to do so as they don’t want to ‘spoil’ the research.
presumptive consent
Where it is impossible to ask real participants so researcher asks a similar profile of person for consent and presume that their participants would also agree
Prior General Consent
Participants are told they may be misinformed or deceived about the true aims of the research prior to the study
BPS code of ethics + cons
British Psychological Society
= A document which instructs (guides) psychologists on what is and isn’t acceptable when dealing with human participants in research
CRITICISMS:
- Not universal as there are other guidelines such as American Psychological Association which has different guidelines on ethics compared to BPS
- BPS is only a guideline, therefore they have been criticised for being difficult to interpret and apply as they can be more vague than ‘rules’
- Only focus on traditional ethics such as confidentiality and protection, ignoring the more complex ethics of negative stereotyping
Positive evaluations of correlations
1) Quick and economical
- There is no need for a controlled environment and no manipulation of variables is required.
- Secondary data can be used, which means correlations are less time-consuming than experiments.
2) Useful preliminary tool for research
- By assessing the strength and direction of a relationship, they provide a precise and quantifiable measure of how to variables are related.
- Correlations are used as a starting point to assess possible patterns between variables before a study takes place.
Intervening variable (third variables)
= Hypothetical internal states that are used to explain relationships between observed variables, such independent and dependent variables. Intervening variables are not real things. They are interpretations of observed facts, not facts themselves. But they create the illusion of being facts.
E.g. measuring the relationship between single-parent upbringings and the likelihood of crime; we cannot say single-parent upbringings cause criminals as there are many third variables at works such as emotional distress from parents splitting up.
similarities of interviews and questionnaires
-need consent
- primary data
- can both use open/closed questions
- both can be structured
differences between interviews and questionnaires
interviews-> spoken (demand characteristics, experimenter bias eg boring or scary)
-semi structured
-time consuming
- usually open questions to gain more in-depth answers
Questionnaires-> written, structured, can get a lot of data quickly from a large sample size, researcher not present but can still lead to researcher bias through leading questions
= used to asses attitudes , behaviours and intentions
open and closed questions +examples + eval
open -> participants choose how to respond, qualitative data (how do you feel…)
closed-> researcher determines possible set of answers, quantitative (yes or no)
- e.g checklist, rankings, likert scale
EVAL:
open-> detailed but difficult to analyse (due to qualitative data)
closed -> easy to analyse but lacks detail (due to quantitative data)
Questionnaire + Interview design considerations
Questionnaires;
- order of questions (e.g qualitative last to not bor ps)
- clarity (clear and concise)
- relevancy (but some random questions to reduce dc)
-bias of questions (should not promote socially desirable answers, no leading questions that promote a certain answer , e.g how do you feel about psychology instead of do you like psychology)
- type of questions (closed/open, checklist, likert scale)
- analyse
-ethics
Interviews;
- location (environmental bias/formality/generalisability of sample)
- open/closed qs
- ethics (privacy)
- clear ,concise and relevant (no leading qs)
Questionnaire Evaluations
strengths:
- cost-effective ->can gather large amounts of data quickly as they can be distribute to a large number of people.
- reduces chances of experimenter bias -> absence of researcher (still could happen due to leading questions but is more time effective)
-if closed qs -> low effort as it’s easy to analyse + can make comparisons between groups due to statistical analysis , so graphs/charts can be made
weaknesses:
- social desirability bias/ demand characteristics (lack internal validity)
- response bias -> Acquiescence bias (the tendency to agree with items in a questionnaire regardless of content)
3 main types of interviews
structured -> interviewer has preplanned questions to ask the same participants the same questions, cannot deviate from
semi-structured -> interviewer still has preplanned questions but they can ask other ones
unstructured -> no preplanned questions
leading questions
a question that prompts or encourages the answer wanted (e.g “do you like psychology” asked by an interviewer that teaches psychology)
Interview Evaluations (structured/unstructured)
Structured:
- easily replicated due to standardised format
- reduces differences between interviewers (reduces exp bias)
x not spontaneous-> limits the richness of data
Unstructured
- more flexible which allows interviewers to gain insight into the worldview of interviewees
x interviewer bias
x difficult to replicate
General interviews
X social des bias -> but skilled interviewers should be able to establish sufficient report with ppts so if a sensitive/personal topic is discusses, they can gain a more truthful answer
Content analysis procedure
- Familiarisation: Read/review the qualitative source
- Determine specific coding categories 3.
- Re-read/re-review qualitative source tallying each time a coding category is identified
- Total up the frequencies of each coding category.
= Qualitative -> Quantitative
content analysis
Whereby qualitative information can be systematically converted into quantitative data, achieved through a process of coding to determine frequencies of certain features within the qualitative source.
Pros of content analysis
Pros:
1) Replicable -> easy to compare in quantitative form (e.g graphs and relationships) therefore there is a straightforward systematic procedure, meaning your research can be easily replicated by other researchers.
2) less time consuming -> can easily be done at anytime and anywhere — and for any cost, too. All you really have to do is get access to the sources you need, which can be as easy as a collection of books, a streaming service, or the internet.
Cons of content analysis
Cons:
1) Subjective -> certain people may interpret categories differently if unclear (e.g ‘tally when aggressive behaviour is seen’, some may interpret a playful slap as aggression whereas others may classify it as ‘banta’. Therefore, unreliable due to the presence of researcher bias.
2) simplistic -> small samples sizes means that data cannot be generalised to wider population
3) missed categories if unclear if only conducted by one researcher, thus data is skewed
4) social desirability bias -> people may display certain characteristics (e.g not be aggressive) if they are overtly observed as they want to seem likeable to the researchers.
Thematic analysis procedure
- Familiarise: read/review source
2.Agree on themes eg (ie opinions) - Re read source with themes in mind , quoting whenever theme is presented
- Present an overall theme -notes by researcher (usually in quotes) .
-> themes : ideas that are either explicit or implicit that are evidenced across the sources
Thematic analysis aim + def
Aim: involves the analysis of qualitative data, and presents results in qualitative form.
- involves identifying themes ; implicit/explicit ideas within data. Eg via quotes
-For example, mental health issues may be misrepresented in an article as a “threat” and “use up NHS resources”. These themes develop development, broad categories like “stereotyping”.
Pros of thematic analysis
Pros:
1) Produces rich qualitative data
2) flexible
3) can generate unanticipated insights
Cons of thematic analysis
Cons:
1) Difficult to define themes effectively, thus some material may be incorrectly allocated or nor included due to bias
2)Time consuming due to having long, descriptive, written texts to analyse repeatedly.
3) Difficult to reliably analyse
Case Study
Descriptive, in depth research of an individual, group or phenomenon that involves collecting data using a range of techniques
- usually involves qualitative data and a history of the individual
- longitudinal studies
evaluations of case studies
pros:
- rich and detailed qualitative data
- high ecological validity
- can generate hypotheses for future studies
- contributes to understanding normal functioning
cons:
- subjective, cannot be generalised
- difficult to replicate
- time consuming
structured vs unstructured observations
structured -> researcher specifies in detail what is to be observed and how the measurements are to be recorded
unstructured -> researcher monitors all aspects of the phenomenon that seems relevant to the problem at hand, very rich in detail but can be overwhelming as there may be too much information to note down all behavior
difference between observations and experiments
observations -> will never influence the response of participants ( where a researcher observes and records participants behaviour, but does not manipulate and variables)
experiments -> researcher manipulated the situation and observes the effect under controlled settings
Naturalistic observation vs controlled observation
naturalistic -> watching and recording behavior in the setting within which it would normally occur
controlled -> watching and recording behaviour within a structured environment, ie where some variables are managed
Behavioral categories
When a target behavior is broken up into components that are observable, clear and measurable (operationalisation )
- ie an observer may be observing aggression in teenagers, so the unambiguous, coherent, behavioral categories used may be kicking, punching, sniggering etc
event sampling and time sampling
event -> A target behavior or event is first established then the researcher records this event every time it occurs.
time -> A target individual or group is first established then the researcher records their behaviour in a fixed time frame, say every minute
Inter-observer reliability (aka inter-rater reliability)
Used to make data more objective and unbiased, observations are carried out by at least two researchers in which data from each observer is compared to check for consistency/reliability (Where two or more research is agree on a set of results)
observational studies evaluations
PROS:
1) high ecological validity -> Observations have the benefit of capturing what people actually do, which may be unexpected behavior. People often do not act the same as they say they would so observational methods are useful because they give special insight into real behaviour.
2) ethical consideration -> Observations allow us to study variables that would be unethical to manipulate otherwise (e.g prison behaviour)
CONS:
1) observer bias -> Observers’ interpretation of a situation may be affected by their expectations. They may look for certain behavours over others etc. This may be reduced however, by using more than one observer (inter-observer reliability)
2) cannot establish cause and effect -> Due to no manipulation of variables, so even though two variables maybe related, both may be caused by third variables (e.g there may be a relationship between single parent upbringings and the likelihood of crime but we cannot say single parent upbringings cause criminals as there are many third variables such as emotional distress from parents splitting up that may have created criminal behaviour) .
Peer review
=The assessment of scientific work by others who are specialists in the same field to ensure that any research intended for publication is of a high quality.
- it is part of a verification process where research is deemed by scientifically acceptable or not (occurs before new research is published)
The process of peer review
1) An expert in the same field is selected, who works for the journal the paper has been submitted to
2) Read the unpublished paper checking for mistakes
3) Either say it 1. Accept the work unconditionally. 2. Accept it as long as the researcher improves it in certain ways. 3. Reject it but suggest revisions and a resubmission. 4. Reject the report outright.
= The process can be carried out via single-blind, double-blind or open review
Main aims of peer review
1) To allocate research funding -> Decides whether or not to award research funding to a particular project.
2) To validate the quality and relevance of research -> In relation to the formulation of the hypothesis, the methodology, the statistical tests used and conclusions drawn.
3) To suggest improvements or amendments -> In some extreme cases, they may conclude that the work is inappropriate for publication and should be withdrawn.
Issues with peer review
1) publication bias/
Distorted understanding of a behaviour
-> It is a natural tendency for editors of journals to want to publish significant ‘ headline-grabbing’ findings to increase their credibility and circulation of their publication. This could mean the research which does not reach these criteria is ignored. Creating a false impression of the current state of psychology as journal editors are being selective with what they publish.
2) anonymity -> Usually the peer doing the review remains anonymous throughout as this is likely to produce a more honest appraisal. However, minority of reviewers may use their anonymity as a way of criticizing rival researchers. This is more likely by the fact that many researchers are in direct competition for limited research finding. Therefore, some journals favor a system of open reviewing whereby the names of the reviewers are made public.
3) preserving the status quo -> Peer review can be biased towards preserving the status quo and favouriting safe research that goes with existing theory rather than unconventional work.
4) it is not always possible to find an appropriate expert to review a research proposal or report if it is a highly specialist area
MCT, MD, LOM ( for mean median and mode)
mean -> standard deviation -> interval
median -> mean -> ordinal
mode -> mean -> nominal
correlations vs experiments
correlations -> investigate the relationship between co variable/no manipulation (cannot establish cause and effect)
experiments-> manipulate IV to see if DV is affected and establishes cause and effect
Statistical Tests table (+ why used)
Name: Difference (exp) or Relationship (corr) LOM Rel/Unre
unrelated difference interval unrelated
related difference interval related
pearsons r relationship interval related
mann-whitney difference ordinal unrelated
wilcoxon difference ordinal related
spearman’s rho relationship ordinal related
chi-squared difference via association nominal unrelate
sign test difference nominal related
unrelated -> independent measures
related -> repeated measures, matched pairs, correlation
= used to determine whether a significant difference/correlation exists (consequently whether the null hypothesis should be accepted or rejected)
Counterbalancing
Counterbalancing is a technique used to deal with order effects when using a repeated measures design. With counterbalancing, the participant sample is divided in half, with one half completing the two conditions in one order and the other half completing the conditions in the reverse order. E.g., the first 10 participants would complete condition A followed by condition B, and the remaining 10 participants would complete condition B and then A. Any order effects should be balanced out by this technique.
Descriptive vs Inferential statistics/analysis
Descriptive
- used to quantitively summarise data
- results in form of charts, graphs and tables
- measures of central tendency and dispersions
Inferential
- provides data from a sample that the researcher studies to make conclusions about the population
- results in forms of probability scores (stat tests)
How to calculate the sign test
When to use -> Difference, Nominal, Repeated measures or matched pairs (related)
- hypothesis : directional or nond
- how many + - = n (0 don’t count)
- smallest number is (s) value
- compare on critical value table using 0.05p if not told otherwise
- use n value
= (s) value must be less than or equal to the critical value to be significant (not a coincidence)
Type 1 vs Type 2 errors (stat tests)
1-> A false positive : when you think your findings are significant but they are not (reject the nul and accept the exp)
2-> A false negative : when you think your findings are not significant but they are (reject the exp and accept the nul)
Stat tests -> which hypothesis should you accept/reject when your results are significant?
You reject the null hypothesis if your results are significant .
You accept the experimental hypotheses if your results are significant.
Features of scientific study’s
- empiricism
- falsifiability
- objectivity
- replicability
- paradigm
Pilot Study
A pilot study is the first step of the entire research protocol and is often a smaller-sized study assisting in planning and modification of the main study
Ways of assessing/checking validity
1) face validity -> whether it appears ‘on the face of it’ to measure what it’s suppose to measure (does an anxiety test look like it measures anxiety? - can eyeball it, no professional needed).
2) concurrent validity -> how a new test compares against an old validated test measuring the same construct. ( Agreement is indicated if correlation between the two sets exceeds +.80)
Ways of assessing/checking reliability
1) inter observer reliability: observations carried out independently by at least two researchers to compare data and check for consistency.
- may be done as a pilot study to check observers are applying behavioural categories in the same way.
- gain a correlation co efficient using statistical testing (at least +.80 for reliability)
2) test-retest : a way of assessing the external reliability of a research tool (questionnaires/psychological tests). Iinvolves presenting the same participants with the same test or questionnaire on two separate occasions, and seeing whether there is a positive correlation between the two.
= Shows the extent the test/questionnaire produce the same answers
- gain a correlation co efficient using statistical testing(at least +0.80 for reliability)
operationalisation
Turning abstract concepts into measurable observations. Where even the most basic concepts are defined through the operations by which we measure them.
Examples of research methods
Common research methods in psychology include surveys, case studies, experimental studies, content analysis, meta-analysis, correlational research, quasi-experiments, naturalistic observation, structured observation and neuroimaging.
Lab experiments are considered ‘scientific’ as they allow for greater levels of standardisation of procedures and precise recording of results. Which TWO feature of science does this relate to?
Control
Replicability
What do the three levels of measurements correspond with…
-> central tendency
-> level of dispersion
-> graph
(used for design a study)
Interval
-> mean ( No extreme scores)
-> standard deviation
-> all graphs
Ordinal
-> median (extreme scores as mean may be distorted)
-> range
-> bar chart
Nominal
-> mode
-> N/A
-> bar or pie chart
Why would a histogram be inappropriate for comparing means across two independent groups?
Histograms represent continuous data (ie interval) and data obtained is not continuous. The data from an independent measures is categorical/nominal (ie with or without music).
Therefore, it would be best to use a bar graph.
What does a 0.05 (usual) level of significance mean
There is a 5% probability the results came about by chance but a 95% probability that the IV caused the DV.
This Level of significance Best balances the risk of making a type one or type two error.
Why may a psychologist decrease the level of significance?
Reduces the chance of a type one error, which is useful when the findings may be used for high-risk situations. For example, we may reduce the level of significance for a vaccine to a more stringent level of significance like 0.01 because you are more certain IV causes DV. Therefore, there is less harm to humans if incorrect.
Examples of self report techniques
-> questionnaires
-> interviews
Ethical issues def
Arises as a conflict between the participants rights and the researchers goal to produce authentic, valid, and worthwhile data.
Which two statistical tests can only be used to analyse correlational data?
Pearsons + Spearman’s
When to use a positive/negative/normal distribution graph?
Positive -> when the mode is the SMALLEST value out of the mean and median.
Negative -> when the mode is the LARGEST value out of the mean and median
Normal -> when values are close together
stratified sampling (how to conduct)
The composition of the sample reflects the proportions of people in certain subgroups/stratas within the target population/wider population.
How to conduct:
1) Strat/subgroups are identified
2) calculate proportion of each strata (10% 90%)
3) calculate number of participants needed from each to make proportional. (1 and 9)
4) Random sampling is then used to select the number of ppts required from within each stratum.
For example:
10% graduates
90% undergraduates
= in target population
But a researcher wants a sample of 10 (so using random sampling choses 9undergrads and 1grad to represent target pop)
target population meaning
The group that the researcher draws the sample from and wants to be able to generalise the findings to.
Parametric vs Non-parametric Tests
Parametric :
- related t test
- unrelated t test
- pearsons
Non-parametric :
- sign test
- chi squared
- wilcoxon
- mann whitney
- spearman’s
What are parametric tests and why are they good?
CRITERIA:
-> interval lom
-> homogeneity of variance (the same level of variance/ SD is the same in both conditions)
-> normal distribution (mean median mode all the same, data must have been from a population that has a normal distribution)
= They detect significance where non-parametric test may not be able to as they take into account all values so are more robust and sensitive to data).
continuous vs non continuous data
Continuous -> interval/ ratio
Non continuous -> ordinal/nominal (categorial)
Test-retest
a way of assessing the external reliability of a research tool. It involves presenting the same participants with the same test or questionnaire on two separate occasions, and seeing whether there is a positive correlation between the two.
How to improve reliability (for questionnaires /observation/interviews/experiments)
- Questionnaires -> Items should be deselected/re-written. E.g. replace open questions with close questions to make less leading and ambiguous.
- Interviews -> Same interviewer, but if this is not possible, higher trained professionals to use structured interviews.
- Observations -> operationalise behavioural categories to make more descriptive and less subjective - so more measurable.
- Experiments -> Standardised procedures.
How to write a statement of significance?
- The result is significant/not significant.
- This is because the calculated value (x) is equal/higher/lower than the critical value (x).
- For the 0.05 level of significance, N/DF = (x) for a one/two tailed hypothesis.
- Therefore, the null hypothesis is supported/rejected.
What could you do if you found the observed value/calculated value to be way beyond the critical value threshold?
Compare the calculated value to the critical value at a more stringent level of significance, for example, 0.01.
Inter rater reliability; how to use to check reliability ?
-> two observers
-> train them to operationalise the behaviour they want to observe (eg anger -> punching, going red etc)
-> ask them to complete observations at the same time; using a tally
-> then ask them to compare results
-> they will use statistical tests to check the reliability
=if the correlation coefficient is significant (+ 0.8 and above) then it is likely the data is reliable/consistent
Economical implications of psychological research
1) Attachment research into the tole of the father
- Bowlby; childcare = mother // provider = father
-New research; although a father’s role is different from a mothers, they’re both as equally valuable for providing emotional/psych support.
- Therefore, households today are better equipped to maximise their income + contribute more effectively to the economy as parents have more flexible working arrangements.
2) The development of treatments for mental health disorders
- Absence from work costs the economy £15 billion a year (1/3 due to mental health)
- Psych research into causes/treatments of mental health plays as role in supporting a healthy workforce.
- EG -> drug treatments (SSRI’S for OCD and Depression)
-> referrals from GP’s for psychotherapies (systematic desensitisation + CBT).
= Therefore, individuals suffering from mental health issues are better able to manage their condition effectively and return back to work (economical benefit).
coding
The initial stage of content analysis
- some data sets are very large, such as transcripts, and so there is a need categorise this information into meaningful quantitive units.
For example, counting, the number of times of particular, word of phrase appears in a text.
How to improve validity (for questionnaires /observation/experiments/qualitative research)
1) Experiments :
-Control group; measures whether IV caused DV?
-Single/double blind procedure to decrease investigator effects and DC
2) Questionnaires :
- Distraction questions to decrease DC as less likely to guess aims
- Tell participants all data is anonymous
3) Observations:
-Covert observation
-Operationalise behavioural categories
4) Qualitative research (case studies/interviews):
-Triangulation: the use of a number of different sources as evidence, e.g. diaries, and observations, etc
threats to validity
- Extraneous variable
- Confounding variable
- Demand characteristics
- Investigator effect
- Leading question
confounding vs extraneous variables
extraneous variables : any variable, other than IV, that may have an effect on DV if not controlled (can be sorted out in a pilot study BEFORE research takes place).
confounding variables : any variable, other than IV, that may have affected the DV so we cannot be sure IV caused DV (occurs AFTER research has taken place; effects study negatively)
= an extraneous variable can become a confounding variable if it was not picked up on in a pilot study.
How to conduct a matched pairs study
1) baseline assessment (on relevant trait) by using a QUESTIONNAIRE/INTERVIEW
2) Give an EXAMPLE of a question that may be used in gathering the data
3) MATCHED into pairs based on a similarity
4) RANDOMLY allocate one p from each pair to each condition
when is a scattergraph appropriate to use?
When the study is correlational looking at the relationship between two co variables.
meta analysis
= Researchers collect and collate a wide range of previously conducted research on a specific area
* collated research is reviewed together
* combined data/effect size is often statistically tested to provide an overall conclusion
What does the distribution of scores (mean, median and mode) mean for the data? (in relation to skews)
1) data is normally distributed if the mean is equal to the mode/median
2) data is positively skewed if the mean is greater than the mode/median
3) data is negatively skewed if the mean is smaller than the mode/medianWhat
How to answer a ‘using the data in the table’ question… (4)
Point
Justification
Point
Justification
(PJPJ)
Evaluation of event and time sampling
Event :
Pro -> it helps to preserve the integrity or wholeness of the event being studied (ie not restricted to a certain time frame)
Neg -> may be difficult to assess all the behaviours as they occur (to many to count etc)
Time:
Pro-> observer has time to record what they’ve seen
Neg -> some behaviours will be missed outside the intervals - observations may not be representative
percentage increase/decrease
new value - old value divided by old value x 100
when is a type 2 error most likely to occur
When the probability level is very low (1%>5%)
