researchmethodsUPDATED.txt

• 1. Nature is orderly
• 2. Nature is knowable
• 3. All events have natural explanations
• 4. Knowledge is based on observation
• 5. Nothing is self evident

Theory that opinions and actions should be based on reason and knowledge rather than on religious belief or emotional response.

All knowledge is derived from sense-experience

Valid: Affirming the antecedent, denying the consequent

Invalid: Affirming the consequent, denying the antecedent

Moving from the General to the Specific

Moving from the Specific to the General

• To assign symbols/use numbers
• according to a set of rules

• Strictly categorical. It is 1) Exhaustive 2) Mutually Exclusive 3)
• Without Order.

• Example: “What religion are you? 1 – Catholic 2 – Jewish
• 3 – Buddhist 4 – Other"

• This has an order. It is 1) Exhaustive 2) Mutually Exclusive 3) Ordered.
• 4) Cannot measure distance.

• Example: “x=Highest level of education completed.1 – Less
• than HS 2 – HS or GED grad 3 – Some College 4 – College Grad 5 – Advanced
• Degree”

• This can measure distance, including negative values. It is 1)
• Exhaustive 2) Mutually Exclusive 3) Ordered 4) Can measure distance 5) CANNOT
• calculate ratios.

• Example: A number line ordered from -$10k to $10k, with
• subjects placed along the line with their stock market gains.

• This can measure distance and ratios, but not negative values. It is 1)
• Exhaustive 2) Mutually Exclusive 3) Ordered 4) Can measure distance 5) can
• calculate ratios.

Example: Number line ordered from $0 to $100k, people’s incomes.

• Is there a value order?
• Could there be negative values?

• Yes - Ordinal
• No - Nominal

• No – Ratio
• Yes – Interval

Are you measuring what you claim to measure?

• 1. Face validity/does it look valid
• 2. Content validity/does it encompass everything?
• 3. Predictive validity/should be strong correlations with other studies
• 4. Construct validity/degree to which inferences can be made from operationalizations in your study

Does the measure yield consistent and stable results?

• 1. Observer reliability/how observer may have affected data collection
• 2. Instrument reliability/are your instruments reliable and accurate?
• 3. Phenomenon reliability/are there any phenomena that have affected the data?

tentative answer to a research problem, expressed in the form of a relationship between dependent and independent variables

the variable we want to explain or predict

The variable we conjecture explains or predicts Y (dependent)

Any other variable that may impact the x+y relationship

• Limited number of observations
• Selected from a
• population
• On a systematic or random basis,
• Which yields generalizations about
• the population.

• Numerical expressions
• summarizing various
• aspects of the entire population

We use sample stats to infer/generalize about the population parameter

a sample that permits specifying the probability that each sampling unit will be included in the sample

So you can make an inference to the population

Each individual is chosen randomly and entirely by chance, such that each individual has the same probability of being chosen at any stage during the sampling process, and each subset of k individuals has the same probability of being chosen for the sample as any other subset of k individuals.

• 1. Simple random
• 2. Systematic random
• 3. Stratified sample
• 4. Cluster sample

small populations

• You pick a number K at random (usually)
• then every "Kth" number after that

Probability sampling design in which the population is divided into homogenous strata within each of which sampling is conducted

• The entire population is divided into clusters,
• And a random sample of these clusters are selected.
• All observations in the selected clusters are included in the sample.
• D.C.R.S.S.O.C.I.S.

• 1. Practical factors (time/money)
• 2. Math factors (acceptable error margin and sophistication of data analysis)
• 3. Population factors (how diverse/homogeneous is the population?)

• 1. Quota sampling
• 2. Purposive sample
• 3. Convenience sample
• 4. Snowball sample

Getting a certain quota met, e.g. get 200 construction workers

focus groups, precinct exit polls

radio call ins, magazine surveys

Hard to reach populations (e.g. crack dealers)

• 1. Correlation
• 2. Time order
• 3. Non-spuriousness

• 1. Random probability sample
• 2. Real world data collection environment

• Stimulus/response relationship with
• Independent/dependent variables
• Pre/post testing
• Control groups

Internal validity is strong in classic experiment design because you have a large degree of control

• 1. Random population? No probability sample
• 2. Real world setting? Maybe not

• Data are used to
• Examine relationships between properties and dispositions,
• Establish causal relations between them, or
• Describe the pattern of relation

correlational designs are not meant to show causation

• 1. Correlation? Either there or not
• 2. Time order? Often straightforward, not always
• 3. Non spuriousness? A problem for correlational design, hard to prove

By definition, you have a probability sample and real world setting. Strong external validity.

• Two or more groups and or measures over time.
• May/may not have stimulus.
• May/may not have a probability sample.

• Weaker than true experiments
• Stronger than correlational

• Lack of control over rival explanations
• Causation must be logically inferred

• It allows researchers to carry out studies in natural settings with probability samples
• It doesn't require random assignment of individual cases to control groups

• Comparison
• Manipulation
• Control
• Generalizability

It relies on a sample probability or nonprobability

You can't draw a sample, and instead use narrative elements to explain (an individual or leader's decisions etc)

• It draws a random sample at different points in time;
• you might ask random samples to respond to a set of questions about their attitudes (e.g. attitudes about nuclear power)

Where you conduct several observations of the same subjects over a period of time, sometimes lasting many years.

• 1. It is bell shaped
• 2. The mean is in the middle, with 50% on each side
• 3. The mean will equal to zero, with a standard deviation equal to one

Measures that reflect a typical characteristic of a frequency distribution (mean, median, mode)

Statistical measures that reflect the degree of spread in a distribution