Week 2: Experimental design and method Flashcards
What are the two types of research?
Quantitative- involves gathering numerical data
Qualitative- involves gathering non-numerical data
What are the principles of quantitative research?
-Involves gathering numerical data
-Always begin with a testable prediction (hypothesis)
What is the difference between a population and a sample?
Can’t gather data from whole population as too time consuming/costly + not all will respond
-Instead select a sample from population + gather data from them
-> instead we take a sample from the population and gather data from them
-We want to be able to generalise the results gained from our sample to the whole population we are interested in
What is random sampling and what are the 2 types?
Quite uncommon in psych
-Leads to a representative sample -> confident in generalising results to entire population
2 most common forms of random sampling:
- Simple random sampling: every single person in population has equal chance of being picked
- Stratified sampling: population is divided into meaningful groups (e.g. where they live in the UK) + simple random sampling is conducted on each group
What is non random sampling and what are the three types?
Very common in psych
-Leads to a less representative sample -> less confident in generalising results to entire population
-Saves time + money + is often the only practical option when population is large (e.g. all people living in UK, or all males)
3 forms of non-random sampling:
- Voluntary sampling: members of the population self-select to participate in the research (e.g. those that reply to student advert). If ppts get their friends/family to participate this is an e.g of snowball sampling
- Convenience sampling: members of the population who are easy to reach are asked to participate in the survey
How big should the sample be?
The size of the sample will determine:
- Extent to which you can generalise findings
- Probability of a ‘chance’ finding or of missing an important finding (larger sample = more power to detect an effect if it exists)
Depends on:
- Size + homogeneity of
population
- Nature of variables measured
- Required precision of results
- How confident you want to be about your results
What are experiments?
-The only way to explore causal relationships
-> Do changes in variable A cause changes in variable B?
Variable = any value that can vary (e.g. score on a test)
-Manipulate the values of one variable (A) and see if it affects a second variable (B), keeping all other variables constant
e.g. does exercise (A) affect memory performance (B)
- Want to control for chance differences between groups - can do this by random allocation to groups
How do we allocate the members of our sample to the groups?
We want to control for chance differences between groups
-> best way to do this is by random allocation to groups
-> each member of sample has equal chance of being allocated to either group
What are the variables in an experiment?
Independent variable (IV):
- The variable which you think may be a causal variable
- The variable that you manipulate (e.g. exercise level)
- Has two or more conditions/groups (e.g. regular exercise vs no exercise
Dependent variable (DV):
- The variable that you think is affected by the changes in the IV
- The variable that you measure (e.g. memory performance)
What are quasi-experimental studies?
A study where the researcher cannot randomly allocate ppts to the conditions of the IV
e.g. when the IV = biological sex
-> split into males and females
What is the difference between experimental and control groups/conditions?
Experimental condition (e.g. regular exercise)
- Condition where the IV (exercise) is present
- Can be several experimental groups e.g.:
group 1 - exercise once a day
group 2 - exercise once a week
Control condition (e.g. no exercise)
- Condition where the IV (exercise) is absent
What are extraneous variables?
Variables that can potentially affect the relationship between the IV + DV -> need to control for as many as possible
e.g. in exercise + memory example, a third variable (e.g age of ppts) may affect memory performance
-> Age would be the extraneous variable
- An extraneous variable is called a confounding variable if it differs systematically with the IV
e.g.all ppts in regular exercise condition are aged 20-22, while all ppts in no exercise condition are aged 80-82
How can you prevent extraneous/confounding variables?
- Match the conditions on key variables
-Balance age, sex, education etc
-Can’t predict all possible extraneous variables - Standardised procedure
- Randomisation of sample to the conditions
-> Ensures equal dispersion of all extraneous variables
Randomisation is the key element to true experiments
What are demand characteristics?
Where ppts guess the study’s hypothesis and (consciously or unconsciously) change their behaviour to support or not support the hypothesis
-> Hiding the study’s true purpose until the end can prevent this
What are experimenter effects?
The experimenter’s desire to support the hypothesis affects their behaviour (consciously or unconsciously)
-> If possible, the experimenter shouldn’t know which condition each ppt has been assigned to - this prevents experimenter effects
What are the 2 types of experimental design?
Between-groups design (or independent measures/between-ppts/between subjects)
Within-groups design (or repeated measures/between ppts/between subjects)
What is a between-groups design?
Compare different ppts in different conditions
Advantages:
- No carry-over effects: Avoids one condition contaminating other condition
- Process is quicker for ppts (less likely to get bored or drop out)
Disadvantages:
- Individual differences have greater effect
- Need more ppts
What is a within-groups design?
Compare same ppts in both conditions
Advantages:
- Effect of individual differences is reduced
- Fewer ppts needed
Disadvantages:
- Boredom, fatigue, drop-out
- One condition may contaminate the other:
-> Carry-over effects (e.g. practice)
How can we solve carry-over effects?
Counterbalancing- Split ppts into two groups
- Half of ppts do condition A first, followed by condition B, then the other half of ppts do condition B first, followed by condition A
-Still a within-groups design as doing both conditions
What is the null hypothesis?
- Manipulation of the IV will have no effect on the DV
- There will be no difference between the conditions
What is the experimental/alternative hypothesis?
- Manipulating the IV will cause a change in the DV
- There will be a difference between the conditions
Hypothesis testing: direction of effect
- All experimental hypothesis must predict a difference
- Hypothesis are either directional (one-tailed) or non-directional (two-tailed)
-> determined by findings of prior research
e.g:
“there will be a difference in the speed of red and blue cars” -> two-tailed hypothesis
“red cars will travel faster than blue cars” -> one-tailed hypothesis
Why do we need to operationalise our variables?
- Not only being clear on what each variable is but also how to quantify the variable you are measuring (DV)
What is the method and what are its 4 elements?
- Gives a clear, succinct and detailed account of how you conducted the research
- Allows a reader to replicate your research
- Report what you did in past tense, and don’t justify why you did these things
4 elements:
Design: overall nature of the design in response to testing your hypothesis
Participants: people you included (and excluded) from study
Materials: what materials/equipment/instruments were used
Procedure: steps taken in putting the above together + running the study
