Research methods

Paper 2

Investigator effect

conscious or unconscious effect of the investigator on the dependent variable e.g. through leading questions, interactions like frowning or nodding

using double negative questions, self report

hard to decipher e.g. are you not unhappy in your job

correlations

shows strength and association between 2 co-variables

presented on a scattergram suing quanitative data

correlaion coefficient- shows how strong a correlation is

+ correlation= + value (closer to 1, stronger +)

- correlation = - value (closer to -1, stronger -)

no correlation = 0

correlation hypothesis

• directional + or -

• non-directional: is a correlation

• null= none

act of starting point by producing precise quantifiable measure of how co-variables are related- further research and experiments can find causation

using secondary data makes research less time consuming as do not have to gather data

does no explain why so can’t infer cause and effect

cannot explain curvilinear relationships e.g. yerkes dodson law more complex than ±

may be 3rd variable that causes relationship e.g. balding and long marriage → age

using double barelled questioms

hard to answer as 2 questio

pilot study

small scale trial run in investigation before real thing is conducted

• saves time and money

• check procedures and technologies work

• make changes and modifications as necessary

generalisation

extent to which findings can be broadly applied to population

extraneous variable

aspect that is not the independent variable but can affect dependent variable when not controlled but can be controlled easily e.g. the age of the participants

primary data and evaluation

original data collected for the purpose of investigation

authentic data obtained specifically so targets information required

requires time and effort from the researchers to produce

secondary data and evaluation

collected by someone other than the researcher prior to research

inexpensive and easily accessed so requires minimal effort

variation in quality and accuracy of secondary data, may be outdated or incomplete

content of data may not match researcher’s needs or objectives so challenges validity of conclusions

quantitative data and evaluation

expressed numerically

easy to analyse so comparisons and patterns can be easily idenfitied

data less open to bias

narrower in meaning and detail so may fail to represent daily life

qualitative data and evaluation

expressed in words

rich detailed data

chance to fully report thoughts, feelings and opinion on a given subject

often has higher external validity so gives meaningful insight into participants worldview

harder to analyse as cannot be summarised statistically so patterns in data harder to identify

conclusions rely on subjective interpretations of the researcher which could be subject to bias

leading questions

guides respondent to particular answers

behavioural categories (observations) and evaluation

breaking down target behaviour to smaller components

data collection more structured and objective so recording of data is more systematic meaning it is easier to analyse and improve data

difficult to include all possible forms of target behaviour, there should not be an ‘other’ category where many behaviours are deposited despite not being similar.

standard deviation, measure of dispersion

how far scores deviate from the mean

larger standard deviation means a greater dispersion of data:

• not all Ps are affected by the independent variable in the same way

  • results were varied or inconsistent e.g. have had extreme scores

time sampling (observations) and evaluation

target individual or group is established and then record behaviours in a fixed time frame

effective in reducing number of observations made

instances where behaviour is sampled might be unrepresentative of the observation as a whole

mean, measure of central tendency (and evaluation)

calculated by adding up all the values in a set of data and dividing by number of values

very sensitive and representative, as it takes into account all data

easily distorted by extreme scores

median, measure of central tendency (and evaluation)

middle value in a data set when scores are arranged from lowest to highest

not affected by extreme scores as the median remains the same, and easy to calculate

less sensitive- does not take into account all data so is less representative as highest and lowest scores are ignored

double blind

third party conducts this study as neither researcher know actual study

• prevents bias from researchers

• placebo is used as a baseline

single blind

do not tell Ps what condition (if any) that they are in so control demand characteristics

likert scale

respondent indicates agreement to a statement

strongly agree 1, 2, 3, 4, 5 strongly disagree

rating scale

respondent identifies value that represents strength of feeling about a particular topic

very fun 1, 2, 3, 4, 5 not fun

fixed choice option

includes use of possible options and respondents are required to indicate those that apply to them

• maths

• geography

• german

interview schedule

list of questions researcher intends to cover

• start with neutral questions to start building rapport

• done in a quiet room to encourage Ps to open up

randomisation

use of chance methods to control effects of bias when designing the order of conditions

standardisation

using exactlythe same formualised procedures and instructions for all Ps

event sampling (observation) and evaluation

record each time behaviour occurs

useful when target behaviour happens infrequently and could be missed if time sampling was used

if specific event is too complex the observer may overlook important details

confounding variable

extraneous variable that has not been controlled so interferes with results and can lead to inaccurate conclusions

• change in dependent variable may be due to this not independent variable

negative skew

demand characteristics

participants are not passive in experiemtns, they may try to make sense of what is happening

• so participant behaviour is no longer authentic

non- participant observations and evaluation

observer is not interacting with people during the observation

maintains objective psychological distance so less chance of going native

too removed from people and behaviour they are studying so less valuable insight

mode, measure of central tendency and evaluation

most frequently occuring value in a data set (2 modes is bimodal)

quick and easy to calculate

less sensitive- does not take into account all data so is less representative, being bimodal is unhelpful

participant observations and evaluation

observer is interacting with people during the observation

can experience situation as Ps do, which increase insight as they get into lives studied which increases validity

‘going native’ - line between participants and you is blurred meaning study loses objectivity

covert observation and evaluation

participants behaviour is watched and recorded without their knowledge or consent, to be ethical this must be happening naturally

no problem of demand characteristics as all behaviour is natural making data more valid

questionable ethics- people do not want behaviour noted down and studied

operationalising

making a variable able to be measured and defined

overt observations and evaluation

participant’s behaviour is watched and recorded with their knowledge and consent

more ethically acceptable as will have given informed consent

participants know they are being observed which may influence their behaviour and make it less valid

controlled observations and evaluation

usually in a lab setting where you want to control aspect of situation

extraneous variables less of an issue so replication is easier

cannot be readily applied to real life settings

using emotive language

portrays researcher’s attitudes too much so needs to be neutral

naturalistic observation? and evaluation

natural environment where target behaviour would normally occur

high external validity as can be generalised

lack of control over research so hard to replicate. uncontrolled extraneous variables means it is harder to judge behaviors

overuse of jargon

too many techincal terms that are only familiar to those in specialist field- means Ps are cut off from understanding

bell-shaped curve

positive skew

line graph

histogram

bar chart

unstructured observations and evaluation

observe all behaviour with no real system

depth of detail covered as produces rich data

only for small scale observations with few people, greater risk of observer bias as no objective behavioural categories- may only record behaviours that catch their eye. qualitative so hard to analyse

cost-benefit analysis

whether research proposals are ethically accessible

weigh up costs and benefits

• may be ethically challenging but could make groundbreaking research that helps others

range, measure of dispersion

highest value- lowest value +1

structured observation and evaluation

only observe behaviour you are interested in

• use behavioural categories

• use sampling techniques e.g. event, time, continuous

• operationalise target behaviours e.g. affection= kissing, holding hands

• important that behaviours are specific so avoids individual interpretation

easier to analyse and compare data observed

5 ethical issues

1. deception- cannot be lied to or misled about study aims

   • midly perimitted if it stops demand characteristics

   • majorly if it benefits research

2. protection from harm- both physical and psychological e.g. injury, stress, embarassment

   • need debriefing to show true aims which always says behaviour was typical and can be offered counselling

3. privacy- need to be able to control information about themselves

   • can only make observations about people in public

   • data has to be stored correctly

4. confidentiality- anonymous, personal details not accessible

   • under data protection act

   • data not shared with other researchers

5. informed consent- must be told nature and purpose of study and their role in it

   • have right to withdraw

   • under 16: parent signs for them

   • presumptive: similar group asked if study is acceptable

   • prior general consent: give permission for several studies where one will involve deception

   • retrospective: asked after study in debriefing

independent groups, experimental designs, and evaluation

participants allocated to different groups which represent one condition

• random allocated: done in an unbiased way to make sure each participant has same chance of being in each group to control order effects

no practice effect as each participant only does each condition once

each group may differ on participant variables e.g. gender, age, ethnicity

more participants needed to get the same amount of comparisons

repeated measures, experimental designs and evaluation

all participants take part in both conditions of experiment

• counter balancing- attempt to control order effects so half do AB and half BA

no issue with participant variables as only compared to themseves

few participants needed so cheaper and easier

order effects- bored or get practice

higher risk of demand characteristics as easier to work out aim

matched pairs, experimental designs and evaluation

pairs of participants matched in terms of relevant variables then one member does A and one B

deals with order effects and demand characteristics

matching variables reduces problems of participant variables

expensive for little gain

takes time to match and never exact

labaratory experiments and evaluation

highly controlled environment, maniuplate independent variable and see effect on dependent variable

can infer cause and effect as highly controlled

easily replicable so more valid

demand characteristics

lack generalisability due to artificial environment and low external valdiity as does not reflect daily life

field experiment and evaluation

natural every day setting, manipulate independent variable and record effect on dependent variable

natural environment means more natural behaviour so higher validity

do not know they are being studied so no demand characteristics

loss of control over extraneous variables

harder to replicate

ethical issues due to a lack of consent

natural experiment

pre-existing independent variable and examine effect on dependent variable e.g. covid

chance to research things we could not usually

high external validity as on real world issues

limited chance to study as event is rare

participants are not randomly allocated so might be confounding variable

actual study done in a lab so may be demand characteristics

quasi experiment

independent variable based on permanent difference e.g. gender

done in a controlled environment so easily replicable which increases validity

participants are not randomly allocated as permanent previous difference so may be confounding variables

independent variable is not deliberately changed so do not know it caused observed change

questionnaires, self report techniques and evaluation

asking people about attitudes or behaviours from pre-set list of questions, either open or closed

open question- researcher does not restrict how you answer the question

closed question- researcher decides a range of possible answers

• face to face, online or post

fast and cheap as multiple participants can do them at once and no need for lots of materials- even the researcher does not have to be there. means bigger samples can be gathered so more representative.

used to gain qualitative and quantitative data whoch are rich and indepth, and easy to analyse and spot trends

may be ignored leading to a lack of participants

misinterpreted and not understand question

participants may not tell truth as impacted by social desirability bias: want to look good so lie so less valid

structured interviews, self-report techniques and evaluation

1-1 conversation between researcher and participant

online or over the phone

pre determined questions and wait for response

easy to replicate as standardised

interviewer needs less training

cannot elaborate further

unstructured interviews, self-report techniques and evaluation

1-1 conversation between researcher and participant

online or over the phone

only topic of interview decided in advance then participants expand and elaborate

greater flexibility as points can be followed up to gain insight

build rapport and use empathy

risk of intervewer bias

have to sift though answers to find relevant information

semi structured interviews

mixture of structured and unstructured as the questions are decided in advance but the interviewers can ask follow up questions

random sampling and evaluation

each member has equal chance of being selected

no bias

extraneous variables should be equally divided enhancing internal validity as you know only the indendent variables had an effect

time and effort

may not be willing

could be unrepresentative

systematic sampling and evaluation

every nth person chosen from a list

avoids researcher bias

may coincidentally still be biased

opportunity sampling and evaluation

whoever is available at the time

easy quick and cheap

unrepresentative as drawn from one place so cannot be generalised

researcher bias as they control selection

volunteer sampling and evaluation

participants self select

easy and quick

want to take part so more engaged

volunteer bias- attracts certain demographic so less generalisable

stratified sampling and evaluation

target population are made into strata and proportionally chosen for sample

most representative as all subgroups represented in proportion

long process and may not want to take part

strata cannot reflect all ways people differ

meta analysis and evaluation

uses secondary data through a number of studies which investigate the same aims

these results can be pooled together to make a joint conclusion

can use statistical anlsysis to calculate and effect size which is the overall relationship across a number of studies

creates a larger, more varied sample meaning you can generalise results which increases validity

may be prone to publication bias “file drawer problem”- the researcher may not select all relevant studies and choose to leave out the studies with negative or non-significant results so the meta analysis may only represent some of the relevant data.

peer review: aims and issues

before a piece of research can become part of a journal all aspects must be objectively scrutinised by a small group of usually 2-3 specialists in the field (anonymous to researcher) to ensure publicated research is of high quality

1. allocate research funding- whether to award funding for a proposed research project

2. validate quality and relevance of research- all elemts assessed for quality and accuracy

3. suggest amendments and improvements- may suggest minor revisions of the work

   • in extreme situations conclude the work is inappropriate and should be withdrawn

anonymity- some use anonymity to criticse the writer (may be rival) as can get away with it and may be in direct compeititon for research funding

publication bias- tendency to publish significant headline grabbing findings to increase the credibility of their publication. also a preference to publish good and positve results- research which is not tends to be ignored.

burying opposing research- may suppress opposition to mainstream theories as want to maintain the status quo. researchers tend to be critical of research that contradicts their own view which slows down the rate of change.

implications for the economy

psychological research can help to influence, affect, benefit or devalue our economic prosperity

E.G.

attachment role of the father- both parents are equally capable of providing emotional support needed for development meaning parent’s working arragements can be more flexibile to maximise their income and contribute more to the economy e.g. mum as breadwinner okay, share roles more

treatment of mental health problems

• absence from work due to mental health costs £5 billion annually

• research can help support/ treat sufferers so they can return to work which helps the economy

sign test

nominal

difference

repeated measures

1. calculate difference between each category

2. count number of + and - (ignore any with scores of 0)

3. select smallest number= calculated value

4. check number of participants (N)- ingore any with scores of 0

5. check kind of hypothesis: directional (one tailed) or non directional (two tailed)

6. refer to critical values table- use kind of hypothesis, significance level (0.05 unless told otherwise) and N which gives critical value

7. calculated value must be equal or less than critical value to be significant
   

cannot find statistical certainties, this suggests not this proves

statistical testing

level of measurement

• ordinal

• nominal: categories of data (discrete) so only appears once

• interval

difference or correlation

experimental design

• matched pairs

• repeated measures

• independent groups

inferential- makes inferences and predictions about wider opulation based on a sample

done to see if difference is significant or just chance and whether the hypothesis should be rejected or accepted

1. sign test