Chapter 1 - Intro to Statistics

median

value in the middle when the data is in order

lower and upper quartiles

the medians of the lower and upper halves of the ordered data

baseline rate

rate/risk at the beginning

population

collection of all individuals about which information is desired

random sample

subset of the population selected that every individual has a specified possibility of being in the sample

poll or sample survey

information is gathered from every participant

margin of (sampling) error

number added to and subtracted from sample information to produce an interval that is 95% certain to contain the true value

* in most common sample surveys: 1/(sqrt(number of people in survey))

nonparticipation/nonresponse bias

the people selected for the sample don’t respond

self-selected/volunteer sample

does not tell you about the general population because it’s not random selection

observational study

participants observed and measured, cannot find cause and effect relationship

confounding variable

variable (not IV or DV) not being observed but may be responsible for results

randomized experiment

treatments randomly assigned to participants, can find cause and effect results

treatment in a randomized experiment

specific regimen/procedure

random assignment

each participant has a specified probability of being assigned to each treatment

statistically significant relationship/difference

large enough that it is unlikely to be caused by a confounding variable or be coincidental

multiple testing or comparisons

testing multiple hypotheses in the same study, may result in false positives or data snooping (searching for something interesting in the data to report)

formula to find percentages when given percent change and baseline data

starting population +/- (percent change \* starting population)

Find the percentage of the population that was white in 2020.

Given: 2010 percentage = 66.4% with an 8.6% decrease

66\.4 - (.086 \* 66.4) = 60.7%

formula to calculate percent increase

((new number - old number) / old number) \* 100

Calculate the Percent Increase

Air Traffic Control Errors

1997: 746 errors

1998: 878 errors

((878 - 746) / 746) \* 100 = 17.69 = 18% or 20%

Research Statement: Air Traffic Control Errors have Increased

1997: 746 errors

1998: 878 errors

What’s the baseline?

4\.8 errors per 1 million flights

Find the number of flights for both years based on this information.

1997:

(4.8 errors / 1 million flights) \* (746 errors / x flights)

4\.8x = 746(1)

x = 746/4.8 = 155 million flights

Only 746 / 155 million

\
1998:

(4.8 errors / 1 million flights) \* (878 errors / x flights)

x = 878/5.5 = 160 million flights

Only 878/160 million flights - how much did it really increase

A proposed study design is to leave 150 questionnaires by the checkout line in the student cafeteria were it could be picked up by any student. Is this volunteer sample a poor study design?

Yes, students are not randomly selected, only those who eat in the cafeteria and choose to fill out the survey would respond. This is a self-selected sample.