PYA4 Research Methods

Replicability- The ability for procedures and/or findings to be reproduced and repeated. Other researches should be able to repeat the study and get the same information. This makes a study more reliable as ideas can be tested. Procedures should be repeatable if we wish to draw conclusions.

Objectivity- Refers to views being based on observable phenomena and not on personal opinion, prejudices or emtion. Objective knowledge is information that can be verified by measurements and is therefore available to other scientists to check and verify it.

Control- All extraneous variables need to be controlled in order to to establish cause (IV) and effect (DV). The greater control we have over variables, the better the internal validity.

Empiricism- Refers to data being collected through direct observation or experiment. Empirical evidence does not rely on argument or belief; experiments and observations are carried out and reported in detail so other investigators can repeat and try to verify the work.

Theory construction- Ability to generate truths from research. Successful scientific theories organise empirical knowledge, guide research by offering testable hypotheses and survive rigorous testing.

Process by which psychological research papers are subjected to independent scrutiny by other psychologists working in a similar field who consider the research in terms of its validity, significance and originality. It takes place before publication.

Why it is undertaken:

• To check whether it would be useful to other psychologists
• To advise whether or not is should be published
• To check validity of research, including accuracy of methodology
• To check for mistakes and suggest improvements to the method

Linked to replicability. In science, researchers should be able to measure or observe a phenomenon time after time and get the same of similar results.

Internal reliability- The consistency of a measure within a test

Assessed by: Split-half method. One half of test is compared with the other to check wehther scores are consistent.

External reliability- The ability to replicate the results of a study and get similar results

Assessed by: Test-retest method. Test is carried out on the same participants several times and the similarity of results is recorded. A correlation coeffiecient could also be worked out between the two scores. The higher the correlation, the higher the reliability.

Improving reliability

• Pilot studies check the method of measurement works and participants can use
• Take more than one measurement from each participant. apparatus as intended
• If extra investigator used, they should collect and record data in the standardised way. In observations or interviews, training is required to improve inter-rater reliability.

The study measures what it is supposed to measure.

Internal validity: The controlling of all variables, except the IV which is manipulated

Assessing: Running a pilot study allows researcher to identify any issues (e.g. confusing items on the questionnaire). The researcher would also operationalise all variables.

External validity: The extent to which findings of a study can be generalised to others

Assessing: The researcher could test the findings from one study in a number of different settings (ecological validity) and with deifferent participants. This would allow the researcher to see whether their results can be applied to the entire population (population validity).

Improving:

• Single-blind technique- Participants don't know what group/condition they're in- removes demand characteristics
• Double-blind technique- Neither participants nor experimenter know which condition the particpants are in- removes experimenter bias and demand charactersitics  

How particiapants are allocated to groups.

Independent measures

+ No order effects, same material can be used in both conditions

- Participant variables as differences may be caused by groups, more participants required

Repeated measures 

+ Participant variables eliminated, fewer participants needed

- Order effects may occur, can't use same materials in each condition

Matched pairs

+ No order effects, attempts to control participant variables

- Difficult to match everything about the participants and can be time consuming

Randomly assignment to groups removes participant variables

Counterbalancing removes order effects

Random sampling + Good technique of providing unbiased representative sample

- Very time consuming to compile a list of the target population

Opportunity sampling + Quick and convenient and therefore the most economical

- Unrepresentative of target population

Volounteer sampling + Convenient and ethical; people choose to be involved and choice not biased on researcher 

- Unrepresentative of target population

Stratified sampling + Representative; tries to ensure characteristics of target population are represented

- Can be very time consuming as categories and proportions have to be identified

Mean + Takes all values into account

- Can be influenced by extreme outliers

Median + Less affected by extreme scores than the mean

- Not suited to being used with small data sets

Mode + Unaffected by extreme scores

- Gives no indication of other values/spread of data and may not be a central measure

Standard Deviation + Tells us how much variation there is around a central score

- Can be skewed if sample is small

Should be written in full detail to make replication possible. A Mad Designer Harry Sells Manky Plastic Coats.

Aim- previous research?

Method- reseach method, IV and DV

Design- state conditions

Hypothesis- precise, testable

Sample- recruitment, sampling technique, age, size, gender of particpants, how selected

Materials- consent forms, instructions, advert, how many questions if questionnaire, likert scale, question example, what high scores indicate

Procedure- recruitment, consent, instructions, allocation to groups, durations, debrief

Controls-  2 things controlled e.g. counter balancing, instructions to remove researcher effects

Deception – a participant must not feel deceived, can be dealt with consent or a debriefing
Right to withdraw – participant must know they can withdraw from the experiment at any point
Protection from harm – The participant must not be harmed in the experiment
Informed consent – The participant must give their consent that they want to participate 
Confidentiality – The participant must have the option of remaining anonymous or withdrawing their data from the findings
Privacy – The researcher must respect the participant’s privacy

Experiments with animals

The BPS has some guidelines that must be followed when using animals, 3R's:

Reduction – Use as minimal amount of animals

Replacement – Try and use alternate methods if possible

Refinement – Use improved techniques to reduce stress

Experimental hypothesis= Prediction of what will happen, precise, testable, IV and DV included

Null hypothesis= Prediction that there will be no difference, precise, testable, IV and DV included

One-tailed= Makes prediction about difference, Two-tailed= Makes prediction about difference, but not what kind 

Type one error= Null hypothesis wrongly rejected, reduce change by setting p value higher

Type one error= Null hypothesis wrongly accepted, reduce change by setting p value lower

Lab= +High internal validity, +Causal relationships, -Ecological validity, -Demand characteristics

Quasi= +Real problems, +High internal validity, -Ecological validity, -Causal relationships

Natural= +Real problems, +High ecological validity, -Internal validity, Causal relationships

Field= +Causal relationships, +High ecological validity, -Internal validity, -Hard to replicate

Random errors= Extraneous variables, cannot be predicted, hard to control e.g. room temp, state of mind

Constant errors= Extraneous variables, affect DV consistently e.g. individual differences, not counter balancing

P value= probability 

5% level= there is a 5% chance that results occured by chance, or a 1 in 20 chance

1% level= there is a 1% chance that results occured by chance, or a 1 in 100 chance

Nominal= putting data into catagories e.g. number of males and females in a class

Ordinal= data ranked so put in order of scores e.g. 1st, 2nd, 3rd in a race

Interval= rank and precise intervals between scores e.g. positions of time in a race

Chi squared= nominal data, testing for a difference, independent/repeated measures

Spearman's Rho= at least ordinal data, testing for a correlation

Mann-Whitney= at least ordinal data, testing for a difference, independent measures

Wilcoxon= at least ordinal data, testng for a difference, repeated measures

Abstract= Summary of the whole report, useful to psychologists to see if they want to read about it, clear and concise synopsis of research question, method, findings and conslusion

Introduction= The introduction is an important part of the report that provides background information on theories and studies relevant to the investigation

Discussion= explanation of findings, relationship o background research, strengths and limitations of method, modifications, implications and suggestions for further research

• Data collected by semi-structured interviews with open questions, participant observation, focus groups
• Small, defined sample- purposive sampling
• Organise data by preparing a transcript
• Code the data- identify catagories, themes 
• Top down approach (thematic analysis)- codes represent ideas from existing theory
• Bottom-up approach (grounded theory)- codes or catagories emerge from data
• Represents true complexities of human behaviour
• More difficult to draw conclusions

Consent form: Purpose is..., procedure is..., confidentiality. As am informed participant, I understand that all info will be confidential, participant is volountary so can withdraw at any time, I can ask for a 'time out' period, more comprehensive details of the study will be given to me later, If I have any concerns, I can contact..., I confirm I am over 18

Instructions: Thank you for choosing to take part, before we begin you have the right to withdraw, on your desk is a sheet of paper...any questions? 

Debriefing: Thank you for taking part, aim was..., done by..., results confidential, can choose to withdraw results. Any questions?

• Naturalistic vs. controlled
• Participant vs. non-participant observation
• Structured observation vs. unstructured observation
• Time sampling vs. event sampling
• Overt observation vs. covert observation