Research Design

• Independent Groups
• Repeated measures
• Matched pairs/participants

Independent groups - Each person does 1 condition individual differences (problem).

Repeated measures - All participants do all conditions order effect (might get better with practice/bored/worst...) doing two conditions.

If not fairly distributed: counterbalancing AB|BA (evens out the problem).

Matched pairs/participants - Independent groups but participants are matched.

• low individual differences
• no order effect

Hypothesis and Variables

In psychology a hypothesis is:

• A clear statement
• A prediction
• Testable
• Formulated at the beginning of the research process

Psychologists start with a theory which a general idea about a behaviour and then develop a hypothesis which makes the theory testable.

Experiments compare 2 conditions that are identical in all respects, expect for the factors being investigated.

The experimenter controls the presence, absence or intensity of factors - variables - thought to affect the behaviour being studied.

To effectively study variables they need to be operationalised (made measurable)

IVs and DVs

 

IVs are also known as Independent Variables - these are manipulated by the experimenter.

 

DVs are dependent variables - these are dependent on the IV manipulations. They could be scores, results from tests, observations etc.

 

E.g. Students wearing underpants get better grades than students wearing no underpants. *

IV= underpants

DV= grades

 

Null Hypothesis

*Null hypothesis (add 'not') - There will ('not'-null hyp) be a significant increase (directional - if you change it you can make it non-directional) in the number of words recalled by people aged 18-44 years, when compared to the number of words recalled by people aged 65 and over.

In psychology, there are two types of Hypothesis. It is important to distinguish between them, especially for four coursework:

 

The Experimental/Alternative Hypothesis - makes a prediction about how an experiment will turn out. They should always be specific.

 

Experimental Hypothesis (Alternative):

• Directional (when previous research indicates an outcome) - Specific prediction about which condition will do best (one tailed test)**
• Non-directional - not specific highlights that there will be a difference between 2 conditions (2 tailed test) **

Null Hypothesis

We would use Null hypothesis:

• When there is no previous research or highly contradictory research
• Same statistical tests require non-directional hypothesis

**One-tailed (direction of result is anticipated)
"There will be a significant increase in (scores in condition one), when compared to (scores in condition two)"

OR

"There will be a significant (positive/negative) correlation between (variable one) and (variable two)"

**Two-tailed (direction of result not anticipated)
"There will be a significant (difference/correlation) between (scores in condition one), and (scores in condition two).

For Null hypothesis add 'NOT'

E.g.

• Altering background noise (IV) affects memory skills (DV) - two-tailed
• Rearing a donkey in a darkened environment (IV) retards the development (DV) of it's eyes - one-tailed  

Don't forget! You only use the term 'Experimental Hypothesis' if you are looking at an experiment - if you are looking at some other kind of study (observation, survey, correlation ect) you should use the term 'Alternative Hypothesis'




Sampling

• People who take part in psychology research are called participants.
• Before selecting a sample, the target population needs to be identified; this is the group of people whose behaviour the experimenter wants to investigate.
• Sampling method/technique - because it is not possible to test everyone
• Actual sample
• Representative sample - A sample that represent the population

There are several different types of sampling, find out about the following types and give and example of when they may be useful : Opportunity, Self-Selected or Volunteer, Stratified, Random and Snowball

Opportunity - A sample made up of anybody who is suitable and available.
E.g: Most researchers are based in university psychology departments - samples are usually composed of undergraduate students.
Problem: They are hardly representative of the population as a whole - lacking external validity + Ethical issues

Self-Selected or Volunteer - A sample drawn from people who volunteer to take part in the research.
E.g: Questionnaires / Newspapers or magazines
Problem: The participants are self-selected - they may differ in important respects from those who did not volunteer + not representative of the population

Stratified - A sample which reflects the proportion of certain characteristics in the target population.
Problem: Your various subsets might be unequal in size (e.g. male/female, smoker/non-smoker, etc)

Random - A sample which is randomly selected from the target population.
Problem: Can we meaningfully generalise data from a small sample, even if it is a random sample? Very impractical.

Snowball - A sample drawn from introductions provided by members of the target population
Problem: It is impossible to determine the possible sampling error and make statistical inferences from the sample to the population. As such, snowball samples should not be considerate to be representative of the population being studied.

Relational Research (Correlations)

A study that investigates the connection between two or more variables is considered relational (or sometimes correlational) research. The variables that are compared are generally already present in the group of population.

In correlations it is proposed that two variables are co-related, i.e. they go together in the sense that either:


a) as one variable increases (or decreases) so does the other = positive                 OR
b) as one variable decreases the other increases = negative

The strength of a relationship between two variables is the degree to which one variable does tend to be high if the other is high (or low, for negative correlation). The strength is expressed on a scale from + 1 (perfect positive correlation) through 0 (no relationship) to -1 (perfect negative).

The figure arrived to express this relationship is the correlation coefficient and this can be calculated for the relationship between any two variables.

Problems with correlation!

It is a common mistake to automatically assume that one variable causes the other to vary.
In fact in correlations a CAUSAL RELATIONSHIP is impossible to determine and there are 3 possible reasons for why 2 variables might be correlated:

1. A causal relationship
2. Both variables are affected by an unknown third variable
3. The correlation occurs by chance

For example: There is a positive correlation between bio-yogurt eating and the birth of 2 headed mutant babies. (sorry for the horrible example!)

There is a positive correlation between bio-yogurt eating and the birth of 2 headed mutant babies, (2)eating bio-yogurt can cause the birth of 2 headed mutant babies, (3) living right next to a nuclear power station can lead to people eating bio-yogurt, which can lead to the birth of 2 headed mutant babies.

A piece of Correlational research
Fourty-four thieves
MDH- Maternal Deprivation Hypothesis

An introduction to Research in psychology - Causal Research

1. Experiments on causal relationships investigate the effect of one or more variables on one or more outcome variables. This type of research also determines if one variable causes another variable to occur or change. In order to do this, experiments should be carried out under highly controlled conditions.

A piece of Causal Research:

Loftus and Palmer (1974) The effects of leading questions
Here, Loftus is suggesting that Leading Questions actually cause people's memories to be distorted.

For further information: http://www.simplypsychology.org/loftus-palmer.html