lecture 4- data collection

What makes a sample appropriate?

• goal- gaining knowledge about population

• We can generalise from sample to population

• Sample is representative of population

What samples are representative?

• Large samples

• Sample’s composition is similar to population

What is selection bias and what are examples?

Occurs when sample is non-representative, and this influences results.

• volunteer bias- certain types of people are more likely to volunteer and may behave differently

• Attrition bias- people who drop out of study are different from people who don’t. E.g. if investigating side effects of drug and some people drop out, may conclude that drug has no side effects.

• WEIRD ppts- not representative of world population

What is probability sampling?

• random selection

• Each member of population has specific probability of being chosen as ppt

• Makes it possible to know likelihood of sample being similar to population

• High degree of representativeness

What is non-probability sampling?

• non-random selection

• Individuals have unknown probability for being selected

• Impossible to know likelihood of sample being similar to population

• Usually low degree of representativeness

What are the sampling methods for probability sampling?

• simple random sampling

• Cluster probability sampling

What is simple random sampling?

• all individuals have known and equal probability of selection

• Many statistical methods assume simple random sampling took place

• On average, samples are perfectly representative

• True random sampling difficult to achieve as it requires list of all individuals

What is cluster probability sampling?

• divide population into clusters

• Select random sample of clusters

• Every individual in cluster is part of sample

Example: how good are primary school students at reading?

• no need for list of all students

• Known probability of selection (1/number of schools)

• Sample representative of school level characteristics

What are the sampling methods for non-probability sampling?

• opportunity sampling

• Purposive sampling

What is opportunity sampling?

• ppts are conveniently available

• Restricted sample

• Probability of inclusion is unknown

• Selection bias is likely

What is purposive sampling?

• ppts are included/excluded based on researcher’s judgement

Example: how do senior managers cope with work-related stress?

• who should be included in sample e.g. senior developers, professors, headmasters, CEOs?

What is physiological data?

• quantitative

• Body’s natural response to study environment

• Requires specific apparatus

• Usually continuous over time

• Often requires intensive preprocessing to make it interpretable

• Most objective type

• No self insight required

Methods of collecting physiological data?

• EEG

• fMRI

• Eye tracking

• Skin conductance

• EMG

What is behavioural data?

• quantitative

• Measurable behaviour in online/in-person settings

• Small range of observable behaviours in controlled environments

• Amount of data from each ppt can range from one data points to hundreds

What is survey data?

• typically quantitative, can have qualitative elements

• Predetermined questions with limited answer options

• Can be collected via mail, online, phone, f-t-f

• Efficient but may be selection bias/response bias

• Huge samples possible

• People may give socially desirable answers

What is test data?

• typically quantitative

• Predetermined questions with limited answer options that measure ppt’s ability, or a psychological construct, and make comparable to others

• Combined to a single test score per ppt e.g. total points in math test

• Psychometrics= sub-discipline of psych for developing good tests

What is interview data?

• qualitative

• Largely predetermined questions, answers are completely open

• Data emerges from verbal info from audio/video recordings

• Many pages of text per ppt

What is observational data?

• quantitative or qualitative

• Observation of natural behaviour in natural or artificially created environment

• Behaviour not restricted, but researchers focus on specific aspects

• Qualitative data- behaviour described verbally i.e. researcher diary

• Overt/covert observation- balancing observer effects and ethical data collection

What are cross-sectional studies?

Researchers collect data from ppts at single point in time.

What are longitudinal studies?

Researchers collect data from ppts on multiple occasions over longer period of time.

• data always repeated-measures

• Different types- cohort, panel, intensive longitudinal

What are cohort studies?

• longitudinal

• Ppts share common experience or demographic trait

• Ppts are followed forward and regularly assessed

• Shows distinction between cohort effects and age/time effects e.g. old people more conservative

What are panel studies?

• longitudinal

• Panel= group of people with specific composition of demographic traits who agreed to take part in multiple studies

• Ppts who drop out can often be replaced with other people with same traits

What are intensive longitudinal studies?

• ppts take same survey at least once a day over extended period time, often via an app

• Opportunity to collect data about behaviour and cognitive processes in natural environment, reducing inaccuracies due to memory bias