experimental design and statistics Flashcards
components of a research study
ethics sampling statistical analasys variables design
variables
“If it cannot be measured, it doesn’t exist” (Eysenck..)
• Any characteristic or factor that can vary
• Example: Darley & Latane (1968)
1) Number of bystanders perceived to be present.
2) Helping behaviour
3) How many seconds before helping
• Can refer to non-observable constructs
exam stress example
biological effects
physiological effects
environmental effects
psychollgical
measure meant of variables
self report- psychological test
physiolgical- brain activity, hormone levels
behaviour- Overt Unobtrusive Archives
social desirability bias
answering in question stop keep in line with social expevtations and judgements e.g. people may lie abut how much they smoke/ drink/self harm
populatin
all existing members of a grup
sample
a small subset of the population
aim of research study
to generalise from a sample to the population
Biased sample
may white and eyesnck in 1978
convenience sampling
e.g. your friends
random sampling
- Where every member of the population has an equal probability of being selected.
- Random selection of psychology students
- Stratified random sampling
- Small representative sample better than large unrepresentative sample
- Exit polls in elections • Magazine surveys
research design:observational and correlational
use of observations and studies to describe and predict behaviour
• Observe behaviour while it occurs
• Descriptive method
• Can test hypotheses
research design:single case studies
describe understand and treat individuals behaviour
experimental designs
Manipulate a variable and measure it’s effect in a controlled setting, to establish a cause-effect relationship.
Wild boy of Aveyron
Case studies of feral children support “critical period”
hypothesis. (Curtiss, 1977)
observational studies: example
• Naturalistic observation of chimpanzee (Jane Goodall, 1986)
• Observed them making and using tools
• Falsified previously held view that animals could not use tools.

observer bias
- Expectations of the observer
* Unconscious cues which influence behaviour of participants
ways to overcome observer bias
Coding of procedures
• Observers blind to hypotheses
• Several observers, rate consistency
observational studies: reactivity
Change behaviour when we know we are being observed.
observational studies; minimising reactivity
Use disguised observation
• Get participants used to observer
• Using unobtrusive measures (e.g., archives)
observational studies: example
Hoika & Akhtar (2012): Tested whether 2 - 3 year olds produced novel, copied or cued humour with their parents.
• Participants: 47 parent-child pairs.
• Materials: Toys, video camera.
• Procedure: Parents asked to joke with their child for 10 mins in a playroom setting.
• Coding:Children’s behaviourscoded according to type of humour.
• Objectbased, • Label based, • Conceptual.
• Results: 3 year olds used more label/conceptually based humour than 2 year olds
corelational study
• Relationship between 2 variables
• Naturally occurring variations in 1 variable related to
naturally occurring variations in another variable. • Correlation as a basis for prediction
• E.g., intelligence & job success, exam results and time spent preparing.
correlation; scatter plot
distribution of relationship
strength of relationship
correlational study: advantags
allows prediction
allows study of naturally occurring events
correlation: disadvantages
cannot establish a cause effect relationship
directionality and the third variable problem
experimental research design
either within subjects or between subjects• Investigates effect of an Independent Variable (IV) on a Dependent Variable (DV).
• All other conditions remain constant
independant variable
IV must have at least 2 levels
• Comparison between 2 experimental conditions or groups.
between subjects design
different subjects take part in each group
between subjects design: Advantages
- Each Participant naive to procedures
* Essential when testing naturally occurring variables (e.g., Gender)
between subjects design: Disadvantages
- Large number of Participants needed
* Differences in conditions may be due to differences between each group.
within subjects design
same subjects tested under al experimenrlconditions
within subjects design: Advantages
- Fewer Ps needed
* Solves problem of between group differences
within groups design: disadvantages
Order and Practice
effects
• Use counterbalancing
internal validity
The degree to which an experiment supports a clear causal
conclusion
• IV causes DV
factors limiting internal validity
Confounding variables
• Expectancy effects
• External Validity
tconfounding variables
two interrelated variables IV confound, make i hard o tell which one caused the effect on the DV exaples age education motivation memories
confounding variables environment spacific
ecperimenter
testing
enviroment
aparatus
expecations and demand characteristics
experimenter expectancy effects
demand characteristics
expeimnrwe wxpctancy effets
• Subtle and unintentional cues which influence P’s responses. • Example: E smiling when P behaves as expected.
demand characteristics
P’s know they haven’t been told everything
• Cues participants pick up about the hypotheses which influence behaviour
placeo effect
• A substance with no
pharmacological effect • Placebo effect
• Change in behaviour/symptoms due to expectations
• Influenced by colour of pill, packaging, knowledge of practitioner
placeboeffect: example homeopathy
- Placebo trials comparing sugar pill with homeopathic pills
- Results show an improvement in both conditions.
- Reassurance
- Culture
double blind studies
P’s and experimenter kept blind to the experimental condition.
• Minimises placebo effect and experimenter expectancy effects
• Example
• Subliminal learning tapes
(Greenwald et al., 1991)
• Improvements corresponded to label on tape, not contents.
external validity
- Can we generalise findings of one experiment to other people and environments?
- Replication
- Meta-analysis: combining the results of several studies
- Bystander effect replicated in different populations (Latane & Herrou, 1996) and in a variety of situations (Fischer et al., 2011).
descriptive statistics
Describe a set of data using specific measure
• Summarises and describes characteristics of data
inferential statistics
Allow us to make inferences about a population based on findings
from a sample.
• Statistical significance: unlikely findings would occur by chance alone.
bobo doll experiment
- Bandura, Ross & Ross (1961)
- Children observe adult with Bobo-doll
- Aggressive behaviour recorded
- IV=Observation Group (Aggressive/non- Aggressive
- DV=Number of aggressive acts.
measures of central tendency
mode is the most common score and mean is sum of results divided y how may results there is
median is the point a which 50% of the scores fall
standatd deviation
square root of x-xbar all squared over n-1
normal distribution the bgger picture
Bell shaped curve
• Defined by the mean
and standard deviation
• Example: I.Q of a large sample of the population.
• We can make inferences about a sample if it has a normal distribution.
ow do we decide the level of probability to accept or reject the null ypoothesis
If there is a less than 5% probability the results are due to error, we reject the Null Hypotheses & infer that the Experimental Hypothesis is correct.
• Expressed as p<.05
t test
• Comparing differences
between two samples or conditions to calculate a t-value.
• Based on the probability of this value we can then make a decision about whether to retain or reject the null hypothesis.
tvalue
• t value is a ratio of:
• t= difference between means
variability about the means • t-value large
• the difference in means is large and the variability within the groups is low
• t-value decreases
• As differences in means decrease, or variability increases
ttest: low variability sample
andura et al., (1961) 18 IV: ObservationGroup • Aggressive Male (AM) • Non Aggressive Male (NAM) DV: Number of Punches to Bobo Doll
ttest: high variability sample
- As the Variability (SD) increases, the t value decreases.
* Are these differences in the means significant?
testing hypothesis using t tests
Using a t-distribution (or statistical software) we can obtain the probability of a t value occurring.
• If the p value is <.05, we can reject the null hypothesis.
• Significant result cannot “prove” a hypothesis.
an indépendant sample ttest
An Independent Samples t-test is a parametric test which compares means from two separate samples/groups.
indépendant ttest: example low validity sample
Example: LowVariabilitySample Bandura et al., (1961) 18 IV: ObservationGroup • Aggressive Male (AM) • Non Aggressive Male (NAM) DV: Number of Punches to Bobo Doll
high validity sample example
- As the Variability (SD) increases, the t value decreases.
* Are these differences in the means significant?
parametric tests
make assumptions about data
nonparametric tests
do no make assumptions about data
use t test when
making a comparison
use correlation wehn
looking at relationship between iv and dv
paired t test
A paired t-test is a parametric test which compares means from the same sample tested under two different conditions
assumptions of parametric t tests about data
normal distribution
homogeneity of variance
interval/ratio scale
non parametric tests
Less powerful
• Data that does not meet assumptions of parametric test
not interval radio scale
not normal distraction
mann whitney U test
- Scores from each group are combined and ranked from lowest to highest.
- Does one group have consistently lower ranked scores than the other?
- Calculate a U value
Wilcoxon signed ranks test
Calculate difference between each pair of scores
• Difference scores are ranked from lowest to highest.
• Add together ranks of positive scores and of negative scores.
• The smallest value of the summed ranks = T value.
why do we need scientific methods
• Beliefs, Intuitions, Feelings influence how we explain the behaviour of other people.
• Is our personal (naive) approach valid?
1. Risk of confirmation bias : Only accounting for evidence that
supports our beliefs
2. Hindsight (explainingbehaviouraftertheevent)
example: confirmation bias
- Astrologists argue that levels of extroversion and neuroticism are influenced by star sign.
- Odd numbered signs = tendency to Extraversion
- Even numbered signs = tendency to Introversion
- Water signs=tendency to Neuroticism
- Belief in astrology may have affected results • Eysenck & Nias (1988)
- Retested children and adults with no knowledge
- No effect of star sign on personality
- Example of Confirmation Bias due to belief in astrology
hypothesis testing
• Darley & Latane (1968)
• Created an emergency situation in a controlled
setting
• Manipulated the number of perceived bystanders
• Measured helping behaviour
ethical values i research
- Respect
- Scientific Value
- Social Responsibility
- Maximise Benefit and Minimise Harm
zero correlation
variables c and y are not related statistically therefore there is no relationship between the two
second stage in a research study
gather information and form a hypothesis
fourth step in the scientific process
analyse data draw a tentative and report the findings
informed consent involves
informing participants of the risks, benefits and procedures of a study
third step in the scientific process
test a hyspothesis in an ecperiment
first step in the scientific process
observe something and ask why
three essential characteristics of an ecperiment
researcher measures effect over one variable
one of more variables re controlled
extraneous factors are controlled
the bidirectinality problem
being unable to determine a cause effectrealationship
in determning a cause effect relaitonship
an experiment is an effective tool
confounding of variables is …
hen two variables are intertwined in such a way its hard to tell which one has actuallyy effect the dependant variable
the limitations of hindsight
there is no definite way to know which one of the explanations is correct
past events can usually be explained in more than noe way
generalisability
can information found in a study be imposed on the whole population e.g. this can be difficult in a case study
disadvantages of a case study
there may be lack of objectivity
findings are often not generalisable
its a poor method of determining cause and effect relationship
how critical thinking be applied to science and everyday life
it enables individuals to debunk false beliefs
it can assist scientists in solving puzzles of mind and behaviour
it encourages the belief that beliefs and emotion can act a psychological blinders