Introduction Flashcards
What are statistics?
Science that relates data to specific questions of interest by devising methods to:
- gather data relevant to question
- summarize and display the data to shed light on the question
- draw conclusions to questions supported by the data
Why do we need statistics?
Uncertainty in the data means that we need to determine whether the effects we see are due to chance factors or systematic (nonrandom) factors.
What are three primary sources of uncertainty in data collected in social sciences research?
- sampling techniques
- measurement error
- random variability
Summarize the main points of Abelson (1995) Chapter 1
- the general public tends to mistrust statistics; however this is not reflective of statistics itself.
- statistical claims have an argumentative nature, some subjectivity is unavoidable.
- making a claim with a single statistic can be deceiving (e.g., average life expectancy of conductors)
- comparisons are important, provide context and reduce likelihood of misinterpretation.
- data analysis & statistical inference help us choose among explanations for observed comparative differences.
- logic of null hypothesis testing (see other card)
null hypothesis testing
- testing for systematic differences, two chance factors are at play (sampling, measurement)
3 possible explanations
- variability due to systemic factor (e.g., all control scores are the same)
- variability due to chance factor
- variability due to both chance and systemic factor
First we test all-chance explanation to see if systemic factor should be invoked.
Accept/reject null hypothesis is too strong - use “retain/discredit” instead.
Data
numbers that you get from measurements
Constant
construct that only has 1 value (e.g., all male sample)
Variable
anything that can be codified, has more than 1 value
Qualitative/Categorical variable
assigned values don’t mean more or less (e.g., gender)
Quantitative/Continuous variable
values indicate an amount (e.g., age)
Population
individual or group representing all members of a group of interest
Parameter
value generated from or applied to a population
Sample
subset of a population
Statistic
values derived from sample
What are the four scales of measurement and why are they important to know?
They determine what statistical test is needed:
Nominal – different in kind not degree (e.g. gender)
Ordinal - values have weight but don’t tell you distance between scores (e.g., Likert/rating scales)
Interval - values that have weight and equal distances between each unit (e.g., temperature)
Ratio - an interval scale using an absolute zero which means absence of characteristic (e.g. weight)