Data Analysis Techniques

Descriptive Statistics

Measures of Central Tendency

Mean = the sum of the scores / the number of scores

- = skewed by the extremes

Median = the middle of an ordered list

- =  does not take into account the extremes

Mode = the most occuring number

Bimodal = Two modes


Measures of Dispersion


Range = highest score - lowest score


Standard Deviation =

Small S.D. = scores close to the mean
Large S.D. = far from the mean






Correlation =

- a statistical relationship between two or more variables (anything that can change such as IQ, temperature etc)


- Two variables = co-variates


Positive

Negative

Advantages 
- useful when an experiment is impractical or unethical
- helps prove if it's worth investigating further


Disadvantages 
- cannot tell cause and effect only hints at a relationship... lacks power and rigour of experiment






Presenting Data

Scattergrams

- display results of correlation.
- quick visual impression
- each point on the graph is the point where the scores of the two variables names on the axes cross.
- looking at the general spread of the points tell us the extent the two variables relate.


Tables

Bar Chart

- good when different sets of scores need comparison
- discrete data


Histograms

Don't get confused with Histograms and Bar Charts!
Histograms have history so they stick together. Bar charts are barred from seeing each other. 

Frequency Polygons

- A dot will be placed at the top of a bar or histograms highest point, and connected with a line


Percentages

- Tables, bar charts, histograms, pie charts
- If you scored 43 out of 89 in a test you divide 43 by 89, which equals 0.48. You then multiply 0.48 by 100 = 48%.

Pie Charts

- Visual impression of different proportions which various groups of participants share
- segmants represent a percentage

 
 

Non- Experimental Methods

- do not manipulate the IV
- only describe what is happening


Types of Data
-Quantitative : numbers
-Qualititive : detailed emotional information



Methods

1. Survey
2. Observation
3. Case Study

Survey

Two types of survey

1. Questionnaire
2. Interview

Questionnaire
 
- large numbers of people can take part
- cheaply and easily
- open or closed questions

Open Questions	Closed Questions	Scaling Items
Qualitative Data	Quantitative Data	Quantitative Data
Descriptive information where respondents can explain their response	Numerical information where answers are easily compared	Numerical information where answers are easily compared and indicates strengths
Tells us  WHAT and WHY	Tells us WHAT but NOT WHY	Tells us WHAT but NOT WHY
What do you think about the current state of the health service?	Do you think the health service has a) got better b) stayed the same c) got worse since the collation came to power	I am happy with my body shape 1      2     3    4     5 sa    a    dk    d    sd sa – strongly agree a-  agree dk – don’t know d – disagree sd – strongly disagree

Standardisation

- Pilot study should be conducted to iron out any problems.
- Peer reviewed to identify bias and poor wording
- Randomly allocated so people don't automatically agree.


Avoid!

1. Double Barral Questions (ones that say "do you agree with .... or do you not? yes or no"
2. Presuming Knowledge about topics
3. Emotive questions
4. Leading questions to condition a response

Advantages
-Cheap
-Large amount of data in a short time
-Large sample
-Generalisation
-Highly replicable



Disadvantages
-Sample bias
-Poor questions can cause problems
-socially desirable answers






Interview

-Conversation for research
-Open or closed questions
-Recorded?

Structured
- Specific questions in advanced
-Only questions asked
-Responses outside the questions not recorded


Unstructured 
- Free Range questions
- don't have to stick to those questions
-Respondants can lead interview



Advantages 
- Large amount of detailed, rich information
- Meaningful information feelings, beliefs, motives
-Indicate direction for future research


Disadvantage
- Self report methods can be unreliable (inaccurate memories, lies)
- social desirable bias
-Interviewer effect > extraneous variable



Observation

Coducting

1. Define type of behaviour to be observed
2. Identify time frame
3. Develop "observation schedule"
4. Define observer role
5. Train others
6. Conduct observation 
7. Check data for reliability

Types of Observation

- Natural or conducted in a laboratory


Overt (Disclosed)
- Participants know they are being studied
-Behaviour may be affected (Hawthrone Affect, social desirability, demand characteristics)

Covert(Undisclosed)
- Participants do not know they are studied
-Raises ethical concerns


Observer can be a participant covertly or overtly.
Observer can be a non-participant covertly and sometimes overtly


Observation Schedule

- Audio Recordings
- Video Recordings
- Written Records

Bias
 - behaviour of participants affected if they know they are watched
- Ovbservers consciously or unconsciously record some things and not others
- Different observers fill in records differently.


Within - Observer Reliability

- Observer rates a behaviour consistantly on different occasions

Between - Observer Reliability

- Two or more observers obtain the same results when measuring the same behaviour on the same occasion

Inter-Rater Reliability

- Greater the statistical relationship of "within" and "between" observer ratings, the greater the level of inter-rater reliability



Advantages 
- high ecological validity
- covert observations eliminate demand characteristics and unnatural behaviour
- ethical


Disadvantages
- replication
-no IV
- less control of confounding variables



Case Study

- highly detailed in-depth study of an individual, group, family, organisation or animal
- retrospective or longitudial
- collection of methods:

Records and case histories - social workers, dr's, psychiatrists
Questionnaires + Psychometric Tests - quantitative measures
Structured or Unstructured Interviews - pastor present thoughts, feelings and behaviour
Experiments - accuratly test performance
Diaries -  track thoughts
Observation - behaviour

- involves self report
- carried out in developmental psychology and individual differences


Advantages
- Detailed, rich information on a single case
- high ecological validity
- reveals new information not previously thought of


Disadvantages
- lacks scientific validity
- cant generalise
- unreliable self report
- impossible to replicate
- open to different interpretations

The Experiment

Types of Experiment

1. Laboratory 
2. Field
3. Natural

Laboratory Experiment

- Laboratory environment
- experimentor can control situational and participant variables

Examples
"The Strange Situation"
"Harlow's Monkey's"
"Pavlov's Dog's"




Advantages
- demonstrates clear cause and effect
- easily control extraneous variables
- objective measures ensure validity
- allows replication


Disadvantages 
- low ecological validity
- impossible to control all random variables
- demand characteristics
- sample bias






Field Experiment

- Takes place in an outdoor/indoor real life setting.

Examples 
"Glasgow Study" Shaffer and Emerson


Advantages
- high ecological validity
- low demand characteristics
- avoids sample bias


Disadvatages
- less control of extraneous variables
- diffcult to replicate
- ethics : consent, privacy, deception







Natural Experiments


- semi-experiment
- cannot manipulate IV
- experimenter measures DV

Example
Hodges and Tizard


Advantages
- high ecological validity
- no sampling bias
- low demand characteristics


Disadvantage
- no control of IV
- no control of extraneous variables
- cannot replicate unless naturally occuring
- ethics : consent, privacy, deception






Experimental Designs
  

1. Independant Measures
2. Repeated Meaures
3. Matched Pairs

Independant Measures

- Participants only undergo one condition of the experiment.
- Randomly allocated to the experimental or controlled group
- Ideal when there are two IV conditions


Example
Test whether people stop for strangers
1. tramp
2. buisness man




Advantages 
-Quick
-Cheap
-Easy
-No order effect
-good for experiments involving deception




Disadvantages
- less control of participant variables
- large sample size





Repeated Measures

- Participants do both conditions of the experiment to test themselves
- Eliminates participant variables


Practise Effect - people do better the second time around. Must be the same both times or this will be an extraneous variable.


Boredom Effect - self explanitory


Counterbalancing

Half of participants do first condition, the other have do the second condition.
They then swap to do the other condition they have not already done.
Can eliminate order effect.






Advantages
-eliminates individual differences
-twice amount of data with small sample size




Disadvantages
- cannot be used when subjects must be different (men and woman)
- order effects
-boredom effects
-practise effects
- need to counterbalance



Matched Pairs

- Participants matched with someone of the same intelligence, score, ability etc.
- Identical twins the best!
- Used to avoid participant variables and order effects.


Advantages
- good for misleading experiments
-eliminates individual differences
-no order effects


Disadvantages 
- cannot perfectly match participants on ALL variables
-big sample size
 

Experimental Method

The Research Method

Theory
Hypothesis
Method
Sample
Process of Results
Interpretation of Results
Conclusion
Revise Theory?





Theory

- explain and predict a variety of phenomenon in general terms
- made up of different ideas/models



Hypothesis

- a statement
- Null Hypothesis and Alternative Hypothesis

Three Hypothesis Conditions

1. Predict that a difference will occur between one group and another
2. A precise statement with an IV and DV
3. Refutable - capable of being wrong

IV's And DV's

Independant Variables - the variable that the researcher manipulates.
Dependant Variables - the behaviour of the subject.

Dr. I. Vee likes to experiment on his friend Davey.

"The Herion Addict"
The doc says I'm a IV Drug User
Whether I am high or die 
Is dependant on how much I use

Examples

"Smoking causes cancer"
IV - Smoking (the Experimentor will control group of smokers and non-smokers)
DV - Cancer (the "behaviour" of subject)


"Materal deprivation in the first five years of life in delinquent behaviour during adolescence"
IV - Materal Deprivation in first 5 years
DV - Deliquent behaviour in adolescence


"People who have low levels of intelligence watch three soap operas regularly"
IV - Hours of TV watched
DV - Intelligence


Operationalisation

- Attempting to define a variable in order to measure it's occurance.

Example

"Younger girls talk more during lessons than older boys"
IV - Age and Gender
DV- Talking

Operationalise: definition of young and old; the amount of talking


"Eating cheese before bed causes nightmares"
IV - Eating cheese
DV - Nightmares


Operationalise: what classes as a nightmare? how much cheese? type of cheese? delay before bed?



Null Hypothesis
- No observable effect on DV

Alternative Hypothesis
- Predicts affect on DV



Types of Hypothesis

One tailed - ONE OUTCOME Directional Hypothesis
- predicts direction of result

"Bullying has a NEGATIVE effect on exam performance"


Two tailed Non-directional Hypothesis
- Predicts effect but not direction

"Being bullied WILL AFFECT exam performance"






The Experiment

- where a researcher manipulates the independant variable to see if it has an effect on the dependant variable.

The Three Conditions

1. Generalisation - must be able to generalise results to the whole population (biggger sample size helps this)
2. Replication - results set down so it can be repeated exactly (reliable experiement - same results all the time)
3. Validity - measure what you think you are measuring!
   E.G. "Middle aged men who have more sex live longer"   Does sex REALLY cause longevity? what else? diet, exercise? etc

Ethics

Consent 
- participants informed of the nature of the experiment. Particularly if information will influence willingness to take part.
- with children, special safeguards must be used
- under 16's parents consent


Deception

-cannot withhold information if it affects decision to take part
- if nessesary, consider alternative procedure, give sufficient information as soon as possible, consult experienced researcher


Debreif

-participants should understand the research
-discuss with participants


Withdrawl


-inform of "right to withdraw" procedure
-before, during or after investigation
-data MUST be destroyed


Confidentiality


-information is confidential unless informed beforehand
-information must not be identifiable


Protect From Harm


-participants must not experience any distress that is outwith any level they expect in everyday life
-contact number if they experience stress or questions


Experimental and Control Groups

Experimental group experiences the prescence of an IV
Control group comparison experiences an abscence of IV

"Smoking causes Cancer"

Experimental Group - those who smoke
Control Group - those who do not smoke


Extraneous Variables 
- variables that may cause the result other than the independant variable

"Puddles cause umberella's to go up"
IV - puddles
What else? - RAIN!




Three Types of Extraneous Variable

1. Participant - intelligence, memory, personality, mood, motivation
2. Situational - light, temperature, time of day, noise, place
3. Investigator

Investigator Variables

1. Demand Characteristics - participant searches for cues to work out how to behave. They behave artifically because they think it's DEMANDED
2. Experimenter Effect - characteristics of the experimenter (age, sex, race, behaviour)
3. Experimenter Bias - experimenter's expectations affect participant's performance

The Hawthrone Effect

George Elton Mayo et al (1924 - 1932) 
Aim - find out which variables affected employees work
Hawthorne Plant - Western Electric Company - Illinois
Manipulated work enviornment (light, temperature etc)
All manipulation did not effect work performance.
Conclusion - workers felt special to be chosen for experiment. Increased status amongst workers




Confounding Variable
-  a variable that definately DOES affect the IV 


Sampling Methods


Random Sampling

People in the target population have an equal chance of being chosen.

Advantages
- Unbiased sample
- 


Disadvatanges
- Impractical for large sample sizes
- Time consuming 






Stratisfied Random Sampling 

Breaking down the population into sub-groups (e.g. age) then random participants are taken from those sub-groups to represent the population

Advantages
- representative of population


Disadvantages 
- time consuming
- lack of participant details can make this challenging




Quota Sampling


Once subgroupd have been identified, the researcher picks people from the groups.


Advantages
- less time consuming than stratisfied where the subgroups are randomly selected


Disadvantages 
- No proof sample represents population




Systematic Sampling


Researcher uses a system to pick participants from the population
e.g. picking every 10th person from a list


Advantages 
- whole population is easily sampled


Disadvantages
-




Opportunity Sampling


Researcher picks those who are willing and able to participate 
OR
Participants who take the opportunity to do the experiment


Advantages 
- easy to carry out
- cheap


Disadvantage 
- biased sample






Self- Selected Sampling


Participants who respond to adverts, or volunteer to take part in an experiment. 


Advantages
- cheap
- quick
- an opportunity to reach a wide population


Disadvantages 
- biased sample - those who participant have a hidden agenda!






Snowballing 


Word-of-mouth ; researcher tells participants, who then tell friends/family, who will then participant


Advantages
- quick way of gathering participants
- good for reaching hard-to-get participants