Deck 6

Question 1

humean shizzle

Answer 1

notes

Question 2

humean shizzle 2

Answer 2

notes

Question 3

to understand causality you:

Answer 3

• need to understand it
• can you make it happen (experiment)
• can you see it (observed associations)

Question 4

why/what is sampling

Answer 4

sampling is selecting units from your larger population. why? you want to make inferences about your population so you pick a subset because it is not feasible to study all units in your population

Question 5

steps in selecting a sample

Answer 5

1. determine relevant population and units
2. define a sampling frame
3. determine sampling method
4. select sample size

Question 6

sample size is based on:

Answer 6

• desired precision of population-level estimate
• estimated level of variation/heterogeneity in population
• desired power
• kind of analysis you want to run
• practical considerations

Question 7

random and non-random sampling

Answer 7

in random sampling, all elements in the population have a known chance of being selected. (used to get external validity by statistics and you want a sample that represents the population.

non-random sampling: selection of sample is based on the judgement of the researcher, you get external validity by statistics

Question 8

random sampling methods: simple random

Answer 8

simple random: lottery system

Question 9

random sampling methods: systematic sampling

Answer 9

systematic: you randomly select your first sample, from there you systematically take the further samples (e.g. take a random soil plot at 5 m, so every next sample is selected at every 5 m

Question 10

random sampling methods: stratified random

Answer 10

groups are formed based on relevant characteristics. from every strata, a random amount of units is chosen, so 10 from each strata no matter the proportion

Question 11

random sampling methods: cluster

Answer 11

a cluster is a group that naturally clusters together. you random select clusters and observe all the units in the cluster. samples tend to be more homogeneous

Question 12

multistage clustering

Answer 12

when you also select units within the cluster

Question 13

non-random sampling

Answer 13

snowball
convenience

Question 14

headaches of sampling methods

Answer 14

base rates:
distribution of traits: your fraime should have a discrimination between the differene of distribution
analytic requirements: when you know what is important to analyze in your data, it should be in the sample
nature of the claim:
uncertaintyL the more you know about the samples that you need and where to find the,, you use random sampling. non-random can be used to learn more about your sample

Question 15

different study designs

Answer 15

cross-sectional: you measure once
longitudinal: you measure multiple times (now and later) over time
experiment: you intervene and then you measure
case study: study a specific phenomenon in a certain space and time

Question 16

measurement validity in cross-sectional studies

Answer 16

do your variables cover the concept and is there no systematic bias

Question 17

internal validity in the context of cross sectional studies

Answer 17

can you support your conclusions (assuming that measurement is correct)
- does the study design support the conclusion?
- did you properly deal (= identified them and controlled for them) with confounders?

Question 18

external validity in the context of cross sectional studies

Answer 18

can you generalize your conclusions to other populations, times and contexts?
- problems for external validity are non-response, sampling method, and the quality of your sampling frame