Unit 1: Intro to Stats & Data Analysis

statistics

science of variability

First principle (examining a sample’s external validity)

What was compared?

Second principle (examining representativeness of a sample & where the data came from)

Who’s not here?

Third principle (considering variability & uncertainty)

Incorporate ish-ness

causal inference

claims that some choice we make will cause some effect to happen

causal claims consist of…

a dependent variable, an independent variable, & a claimed effect

generalization inferences

claims about an entire group of people, based only on information from a sample of people

generalization claims consist of…

the population of interest, an attribute, & an estimate(#)

data

representation of someone or something

observations

things we are interested in (rows)

attributes

pieces of information about the observations (columns)

measures

the way we collect information about observations

quantitative data

refers to data in which the values of an attribute are numbers representing a quantity of something

categorical data

refers to data in which the values of an attribute for an observation are selected from a set of different category labels

rating scale data

refers to data in which the values of an attribute for an observation are selected from a predetermined list that has an order to it

ex. (strongly agree, agree, disagree, strongly disagree)

text data

refers to data in which the values of an attribute for an observation is text, typically from an open-ended question

time series data

refers to data in which the values of an attribute for an observation indicate a moment in time

reliability

refers to the extent to which the data you collect from a measure truly represents and reflects the real world characteristics of the observation

data validation

the act of ensuring that the values collected from each observation for each attribute are valid

data cleaning

removing invalid values from a dataset

univariate analysis

the analysis of a single attribute or variable at a time

standard deviation

a measure of spread

boxplots

graphic depiction of a five-number summary

dot plot

a graph where each observation is displayed as one point on the graph

density plot

a dot plot with a line drawn across the top of all of the stacks of dots

frequencies

the total number of observations whose response is equal to a particular value

bar graph

graph based on a frequency table, and has one bar for every response opinion

word cloud

graph that depicts the words across the responses from the observations, size of the word varies by the frequency of the word

distribution

the pattern that the responses from all the observations make

quantitative measures include…

count, mean, standard dev., 5-number summary, box plot, density curve

categorical & rating scale measures include…

count, frequencies, percentages, bar graphs

text data measures include…

count, word clouds

quantitative data characteristics include…

shape, center(location), spread

categorical, rating scale, and text data characteristics include…

frequencies and percentages

normal distribution

distributions with a bell-shape curve shape

right skew distributions

distributions that look like the right side of a normal distribution has been stretched out

left skew distributions

distributions that look like the left side of a normal distribution has been stretched out

middle 95%

part of the distribution that represents most of the observations

what are typical values?

the mean, median, & mode

two way table

a frequency table where the attributes are presented as different rows, and a different attribute is presented as columns

column percents

relative frequencies based only on the total from a single column

absolute difference

whether two column percents are similar or dissimilar

relative difference

what to compute instead of absolute difference when the column percents are small

line graph

what type of graph to use when dealing with time series attributes

grouped density plot

graph to compare two distributions → two density plots in one graph

utilizing shape, spread, and location

if the graph’s shapes are very different…

then it doesn’t make sense to compare the locations with the spread

standard deviation formula

(upper limit of middle 95% - lower limit of middle 95%) / 4

absolute difference formula

Group 1’s column percent - Group 2’s column percent

relative difference formula

Group 1’s column percent / Group 2’s column percent

ratio of standard deviations formula

largest standard deviation / smallest standard deviation

if the ratio of standard deviations is 1…

the standard deviations are pretty similar

if the ratio of standard deviations is 3+…

the standard deviations are pretty different

effect size formula

mean difference / larger standard deviation OR

(Group 1’s mean - Group 2’s mean) / larger standard deviation

if the effect size is .25…

then there is a pretty small distance between the means

if the effect size is .75…

then there is a pretty large distance between the means