Research Methods & Statistics

Semester 1

What is psychology?

• Psychology is the systematic examination of mental processes & behaviour

• It aims to understand & explain thoughts, emotion & behaviour

• There are many different areas of research (e.g. clinical psychology, appetite) & methods (e.g. interviews, experiments)

What is popular/ ‘pop’ psychology?

• popular/’pop’ psychology refers to concepts regarding human behaviour & mental life that become widely popular, yet may be oversimplified, misunderstood or lack scientific backing, despite being presented as based on legitimate psychology

• these concepts form myths & misconceptions of which influence how people understand the world

• e.g. a misleading reported psychological ‘finding’ is that ‘Cannabis use increases the risk of psychosis by 40%,’ however, the statistical likelihood of suffering from psychosis is 0.5%, so it only really increases to 0.7%

What is the difference between ‘hard’ and ‘soft’ science in psychology?

• ‘Hard’ science - objective & measurable 

• ‘Soft’ science - complicated & subjective as psychology studies people 

• It is crucial to have operational definitions of the concepts we are studying e.g. intelligence

How do you run a scientifically sound psychological experiment? 

• You should have a clear definition of the concept your studying e.g. intelligence

• You should have an operational definition of procedure used to measure the variable (DV) e.g. hormone levels – cortisol

• You should have a sample size, bigger is better e.g. Chihuahua vs. Maine coon

• The experiment should be well controlled; the variation should reflect the property you’re interested in.

?

? 

What is the scientific method?

• Initial observations -> Proposed explanation -> Research -> Data analysis (observations & measurements) -> Scientific theory

• e.g. Do men watch more football then women?:

• Observe people in pubs, do they watch the screen? → Do statistical analysis (Chi-Squared) → Is there a difference, does it apply to other settings & sports?

• Theory is continually revised to explain all data, & the process (mechanism) leading to the data

What is a hypothesis? 

• A hypothesis is a prediction, a statement about what you expect, & a way of testing an explanation/ theory.

• In order to be scientifically testable, a hypothesis must be:

• Clearly defined/ operationalised, be non-circular (e.g. sad ≠ sad-looking), deal with observable/ measurable phenomena, & be falsifiable 

What is a theory?

?

Summarise Asch’s conformity experiment (1951)

Asch aimed to investigate the extent to which social pressure from a majority group could influence a person to conform. It was found that 75% of participants conformed at least once.

How was Asch’s conformity experiment flawed?

• 95% of textbooks failed to report that non-conformity was more prevalent (approximately 65% of trials) than conformity

• Biased sample as they were Swarthmore psychology students so low in population validity 

• Task was artificial so is therefore low in ecological validity

• ‘Child of it’s time’ 

• However, it is important to note that there are many valid aspects of the research, conformity just isn't as frequent as the research claimed

Summarise Zimbardo’s Stanford prison experiment (1971)

Zimbardo's study used a sample of Stanford students that were allocated to roles of guards & prisoners. The prisoners were taken to ‘prison’ located in the basement of Stanford Psych lab. It was found that behaviour is driven by the  situation & not personality, & is thus strongly influenced by expectations.

How was Zimbardo’s Stanford prison experiment flawed?

• Zimbardo took part in his study and actively steered it so it isn't a true representation.

• The sample was biased & small (24 students from Stanford University) so wasn’t representative

• Zimbardo's conclusions are highly suspect; they are more of a refection of what he wanted to be true rather than what was true (Only one guard exhibited cruel, unusual behaviour) to get a head-line grabbing result

• Reicher & Haslam (2002) BBC-TV "The Experiment“

• The guards became "increasingly paranoid, depressed and stressed and complained most of being bullied” whereas the prisoners undermined the guards.

• The study has its own limitations e.g. BBC cameras

• Unethical study which couldn't be carried out today (study had to be abandoned after 6 days instead of the planned 14 days due to the cruel behaviour of the guards e.g. making prisoners beg for food)

Why do research?

1. Description

2. (Quantitative) Prediction

3. Understand and infer causation

4. Create change

E.g. Does cannabis cause memory impairments?

Description (observation/ questionnaire), Prediction (correlations/ differences), Understand & infer causation (psychosis), Create change (inform public)

The explicit goal of experimental research is dependant upon who you ask

Why do we need to infer causation?

• Co- variation/ correlation is not sufficient. We need to beware of cofounding variables, e.g. number of books & IQ

• It is important to try & eliminate other possible causes of behaviour e.g. Do heavy cannabis users have more memory problems? - drinking may be an alternative cause as heavy drinkers are more likely to be heavy cannabis users and have memory problems than light drinkers. However, how easy it is to eliminate alternative causes depends on the design.

• Time-order relationship; the cause must come before the effect

What are the different experimental designs?

• Between subjects 

• Natural groups 

• Within subjects

What is a between subjects design?

• Alternatively known as independent groups design

• Groups are made up of different people

• Difference in performance between groups

• Groups can be natural (e.g. old vs young) or manipulated/ allocated (e.g. treatment a vs b)

• Advantage is that  performance is not hindered by :                               1. Boredom/ fatigue as participants only partake in one condition              2. No practice effects/ experience                                                              3. No carry over effects from previous conditions as conditions don’t affect each other                                                         Sometimes this design is the only option

 

What is a natural group design?

• Groups are not created by manipulating an IV

• Based on a participant variable

• Readily formed (e.g. men v women)

What is a within subjects design 

• Alternatively known as repeated measures as you repeatedly measure the same people on the same DV.

• There are 2 procedures for this design: Same time (one session with all conditions) & different times

• Advantages:

1. This design controls for individual differences (e.g. natural memory ability) so participants essentially act as their own control          

2. Participant characteristics are not a problem as they are compared to themselves

3. More powerful

4. Requires fewer participants

5. More convenient to run.

 

What is power?

Power is the probability that you will find a statistically significant effect when it actually exists (i.e. correctly reject the null hypothesis)

What is error variance?

• Error variance = variation caused by individual differences.

• Reducing error variance makes it more likely you’ll find a ‘real’ result.

• Low error variance = more powerful.

?

?

What is the process of conducting psychological research?

1. Choose a Design (between or within)

2. Choose a method (experiments, observations, tests, questionnaires etc)

3. Generate data

4. Analyse

What are the different data levels?

• Nominal

• Ordinal

• Interval

• Ratio

What is nominal data?

• Mostly names/ categories of things

•  You can only really count how many responses we get (frequencies)

• e.g. male or female

What is ordinal data? 

• Data is ordered but cannot assume anything more than this

• E.g. likert scales, age categories

What is interval data?

• Distances between data points are at equal intervals

•  But there is no ‘true zero’ (there can be minus numbers)

• Although intervals are meaningful numbers may be arbitrary (e.g. difference between 12-14 degrees celsius is the same between 88-90

•  The ratio of numbers is not equal (e.g. someone with an IQ of 150 isn’t necessarily 2x as smart as a person scoring 75)

 

What is ratio data?

• Essentially the same as interval

• But is has a true zero and a true ratio (e.g. someone driving at 60mph is going 2x as fast as someone driving at 30mph)

•  E.g. height, weight, score on a test

Why does data level matter?

• The data level determines the type of statistical test that you can perform on your data as different data levels require different tests to answer the same question

• Ratio and Interval data are best.