Standard Deviation Etc.
(y-bar y)/s
What is a standardized score or z-score?
It has mean = 0 and standard deviation = 1
What is the standard Normal distribution?
When designing a sample survey, it is crucial to avoid this. If your survey has it, it can render your results meaningless.
What is bias?
A (flawed) sampling design in which a large group of individuals is invited to respond, and those who choose to respond are included
What is voluntary response sampling?
A special sample in which the entire population is surveyed
What is a census?
The effect that adding a constant to every data value has on position (mean, median, percentiles)
What is 'it adds the same constant to the measures of position' ?
In a Normal model, the % of values that fall within 2 standard deviations of the mean
What is 68% ?
A key aspect of designing a sample survey that allows the sample to represent the rest of the population
What is 'randomization' ?
A (flawed) sampling design which includes individuals in the survey who are easy to include
What is convenience sampling?
An observational study that gathers data from the past
What is a retrospective observational study?
The effect that adding a constant to every data value has on measures of spread (range, IQR, standard deviation)
What is 'it has no effect' ?
Suppose it takes you 15 minutes to walk to school, on average, with a standard deviation of 2 minutes. If a Normal model is appropriate, what % of the time would it take you more than 17 minutes?
What is 16% ?
A sampling scheme in which the population is first divided into homogeneous groups. Simple random sampling is then done within each group. (like making horizontal slices through a Boston cream pie)
What is stratified sampling?
A (flawed) sampling design in which some portion of the population is not sampled at all or has a smaller representation in the sample than in the population
What is undercoverage?
An observational study that plans to gather data in the future
What is a prospective observational study?
The effect that multiplying every data value by a constant has on position (mean, median, percentiles)
What is 'all measures of position are multiplied by the same constant' ?
For a Normal distribution, the z-score for a value at the 84th percentile
What is 1 ?
A sampling scheme in which the population is first divided into representative groups. A census is then performed within 1 group or a few groups. (like a vertical slice through a Boston cream pie)
What is cluster sampling?
A form of bias in which some people included in the survey don't respond
What is nonresponse bias?
Along with direct control, randomization and (possibly) blocking, a key component of an experiment
What is replication?
The effect that multiplying every data value by a constant has on measures of spread (range, IQR, standard deviation)
What is 'all measures of spread are multiplied by the same constant' ?
For a Normal distribution, the z-score for a value at the "2.5th" percentile? (97.5% of the data are greater than this value)
What is -2 ?
A sampling scheme in which the survey begins with a randomly-selected person and then samples (for example) every 10th person on a list.
What is systematic sampling?
A (flawed) sampling design in which something in the design (for example, people's unwillingness to reveal personal facts) influences their responses
What is response bias?
Differences we observe between groups that are greater than what we might get just from random variability
What are statistically significant differences?