About the test: Proctorio, 43 questions (T/F, MC, short answer), 90 minutes, it’ll look like your
canvas content quizzes,do not need to know how to use david lane calculator or jmp (will need
to know how to read output)
Ch. 1
Data means nothing without context
Quantitative Variables: possible values must be numbers, usually with units
Categorical/Qualitative Variables: possible values are given by descriptors, can have
numbers (eg zip code) → would it make sense to add/subtract? If no, probably categorical.
Ordinal Variables: order is perceived to have worth (e.g. rank in the military)
Identifier Variables: exactly one individual in each category, no repeats (eg SSN)
We take samples in order to make inferences about the population
Sampling Designs: randomness is key!
- Simple random sample: population is numbered off and a certain amount of
numbers are randomly selected
- Systematic sampling: population numbered off and every k-th number is sampled
- Stratified sampling: population is sliced into homogenous groups/strata and then
simple random sampling is used within each strata
- Each strata is different from one another, but the members within a single
strata all have a shared trait
- Cluster sampling: splits population into clusters which are then randomly
selected. Every member within a selected cluster is then investigated (census).
- Clusters have no distinct features and are more or less alike
- Multistage samples: combine several sampling methods
Ch. 2
Univariate: one variable
Bivariate two variables
Displaying univariate categorical displays
- Frequency tables/Relative frequency tables: shows count/shows percentage
- Bar chart: displays frequency/relative frequency of each categories
- All bars have same width
- Pie chart: presents each category as a slice of a circle so that each slice has a size that
is proportional to the whole in each category
- Ring chart: like pie but displayed as a ring
- Pareto: bar chart with bars put in order from tallest on the left to shortest on the right
with a line above the bars indicating cumulative frequency
, Displaying univariate quantitative displays
- Histogram: plots the bin counts as the heights of bars
- Stem-and-leaf: shows distribution while preserving individual values
- Best for small data sets
- Dotplot: stacks dots to indicate an occurrence of a value
- Best for small data sets
Describing a distribution
1. Shape: unimoda (one peak)l, bimodal, multimodal, or uniform (no peaks, fairly flat)?
Symmetric or skewed? Unusual features (gaps or outliers)?
2. Center:
a. If skewed: median
b. If symmetric: mean
3. Spread
a. If skewed: Interquartile range (IQR=Q3-Q1)
- “The middle 50% of data range from Q1 to Q3 with an IQR of IQR”
b. If symmetric: standard deviation (how far each data value is from the mean on
average)
c. Range (max-min)
We need to account for shape when describing center and spread because mean and standard
deviation are highly sensitive to skew and outliers.
Variance= (standard deviation)^2
Ch. 3
Big takeaway here: the bigger the differences, the bigger the association
- No association/no differences→ variables are independent (have no effect on each
other)
Types of bivariate categorical displays:
Contingency tables:
canvas content quizzes,do not need to know how to use david lane calculator or jmp (will need
to know how to read output)
Ch. 1
Data means nothing without context
Quantitative Variables: possible values must be numbers, usually with units
Categorical/Qualitative Variables: possible values are given by descriptors, can have
numbers (eg zip code) → would it make sense to add/subtract? If no, probably categorical.
Ordinal Variables: order is perceived to have worth (e.g. rank in the military)
Identifier Variables: exactly one individual in each category, no repeats (eg SSN)
We take samples in order to make inferences about the population
Sampling Designs: randomness is key!
- Simple random sample: population is numbered off and a certain amount of
numbers are randomly selected
- Systematic sampling: population numbered off and every k-th number is sampled
- Stratified sampling: population is sliced into homogenous groups/strata and then
simple random sampling is used within each strata
- Each strata is different from one another, but the members within a single
strata all have a shared trait
- Cluster sampling: splits population into clusters which are then randomly
selected. Every member within a selected cluster is then investigated (census).
- Clusters have no distinct features and are more or less alike
- Multistage samples: combine several sampling methods
Ch. 2
Univariate: one variable
Bivariate two variables
Displaying univariate categorical displays
- Frequency tables/Relative frequency tables: shows count/shows percentage
- Bar chart: displays frequency/relative frequency of each categories
- All bars have same width
- Pie chart: presents each category as a slice of a circle so that each slice has a size that
is proportional to the whole in each category
- Ring chart: like pie but displayed as a ring
- Pareto: bar chart with bars put in order from tallest on the left to shortest on the right
with a line above the bars indicating cumulative frequency
, Displaying univariate quantitative displays
- Histogram: plots the bin counts as the heights of bars
- Stem-and-leaf: shows distribution while preserving individual values
- Best for small data sets
- Dotplot: stacks dots to indicate an occurrence of a value
- Best for small data sets
Describing a distribution
1. Shape: unimoda (one peak)l, bimodal, multimodal, or uniform (no peaks, fairly flat)?
Symmetric or skewed? Unusual features (gaps or outliers)?
2. Center:
a. If skewed: median
b. If symmetric: mean
3. Spread
a. If skewed: Interquartile range (IQR=Q3-Q1)
- “The middle 50% of data range from Q1 to Q3 with an IQR of IQR”
b. If symmetric: standard deviation (how far each data value is from the mean on
average)
c. Range (max-min)
We need to account for shape when describing center and spread because mean and standard
deviation are highly sensitive to skew and outliers.
Variance= (standard deviation)^2
Ch. 3
Big takeaway here: the bigger the differences, the bigger the association
- No association/no differences→ variables are independent (have no effect on each
other)
Types of bivariate categorical displays:
Contingency tables:
