Social Science

Front 1

Targeted Sampling

Back 1

Non-probability.

People you can identify who are of particular significance to the study.

Front 2

Convenience sampling

Back 2

Non-probability.

People you know or come across

Front 3

Quota Sampling

Back 3

Non-probability.

People you come across until you meet a predefined sample size for each of two or more subcategories you want to compare (ex: men and women)

Front 4

Differences between Questionnaires and Interviews

Back 4

Questionnaire: Can't probe for more info.
Can't ask follow-up questions.
Set series of questions.
Easier to compare answers.
Quantitative; can run statistical analysis'
May lack interview bias (unless done face-2-face)

Interviews: More in-depth.

Front 5

Questionnaire Limitations

Back 5

Low response rate
Self-completion not right for all ppl (illiterate)

Question format can bias answers

Can't probe

No control of unanswered questions

Front 6

Questionnaire question types

Back 6

Open-ended: write out their answer

Scale: Likert, symmetrical (always odd number)

Ranking: by preference or frequency

Closed check-list: ex. Yes/No

Choices: choose between statements (which best represents your feelings)

Front 7

Types of Response Bias

Back 7

Yes effect: 60% will answer 'yes' to neutral ?'s

Order effect: will be influenced by previous ?'s

Self esteem/deference effect: say what makes them look good/what you want to hear.

Audience effect: answers differ from when asked alone

Front 8

Triangulation

Back 8

With-in subject: asking the same person more than once.

Between subject: asking different people.

Cross method: using more than one method

Front 9

Types of Errors

Back 9

Non-intentional: don't know

Intentional: don't want you to know

Non-directional: lack of knowledge

Directional: response bias

Front 10

Validity

Back 10

Internal: strong internal logic that allows you to draw conclusions with high certainty

External: rational logic for generalizing results from sample to broad population

Contextual: degree to which results are reflected in the 'real world'

Front 11

Experimental Research Design

Back 11

Good for deductive research

Has high internal validity

Usually gathers quantitative data

Front 12

Quasi-Experimental Research Design

Back 12

Higher external and contextual validity

Still can use quantitative data

Front 13

Observational Research Design

Back 13

Qualitative data

High contextual validity

Weak on external validity (not good for generalizing)

Good for inductive research

Bad for statistical testing

Front 14

Types of Observational Research Designs

Back 14

Single case study

Comparing case studies

Cross-sectional: representative sample from large population

Front 15

Cluster Sampling

Back 15

Probability sampling.

Separate population into mutually exclusive clusters (two different villages), take a simple random sampling of clusters, then create a sampling frame for each cluster and take a simple random sample separately from each cluster.

Front 16

Simple Random Sampling

Back 16

Probability sampling.

Create list of all 'cases' in a population you want to sample (sampling frame) then use a random number table to select individuals.

Front 17

Systematic Sampling

Back 17

Probability sampling.

Determine the number of sample units, then decide how many samples you want, figure out the fixed sampling interval (ex 100 houses, 20 samples, 100/20=5, 5=sampling interval), use random number generator to pick starting point (ie. 3, so start at 3rd house and go to every 5th house)

Front 18

Stratified Sampling

Back 18

Probability sampling.

Put study pop. into different strata (categories), construct a sampling frame and take a simple random sample from each strata.