Author Archives: hangtime
Cumulative Frequency Distribution
Yet another take on the glory which is frequency tables, this variety is useful in certain types of data. This video shows you how to convert a plain old frequency table into a shiny & new cumulative version.
Relative Frequency Tables
Once you already know how to rock frequency tables, the "relative" variety is just the type where the frequency column contains percentages or proportions rather than raw data, just to make it easier to digest for your reader.
How To Draw A Frequency Table
In this video we get into the nitty gritty of creating frequency tables. And there's plenty to go over, including the subtleties of how to select the number of classes and class width, as well as the one thing you absolutely never do when you cook these things up.
Frequency Table Vocab
This video covers some of the terms you'll need for frequency table problems (and distributions as well) such as: upper class limits, lower class limits, class boundaries, class midpoints, and class widths.
These are those tables where the data are sorted into ranges and then tallied. While these most definitely *look* like tables, some teachers call them distributions. These videos cover standard frequency tables, as well as relative and cumulative versions, and vocab such as upper and lower class limits, class midpoints, class boundaries, and class widths.
Uniform Distributions (for Discrete Variables)
Uniform distributions (aka constant distributions) are ones that are just a straight line rather than bell-shaped. The picture to the right is not a typo! For real-world examples of discrete uniform distributions (roulette, day of the month someone is born on), and how to use area to calculate probability for one of these, check this video out.
Bimodal Distributions
Sometimes spelled bi-modal, this term describes a distribution with two mountain tops instead of just one. Definitely can't assume these are normal enough to use most stats tests on!
Kurtosis & Skewness
Kurtosis and skew are just a couple of numbers you can use to quantify how a distribution is different from the ideal of the normal distribution. Kurtosis tells you whether it's spiky or flat-topped, whereas skewness is about whether the mountain tips left or right.
How To Read A Distribution Graph
A distribution graph -- such as the "normal distribution" -- is basically just a graphical representation of a frequency table. This video explains how distributions are basically just bar graphs on steroids, and how area under the curve can be related to frequency.
The videos in this chapter introduce the basics of what distribution graphs are and what you can figure out from them (hint: area=probability), then applies that knowledge to a bunch of situations, including bi-modal distributions, uniform distributions, kurtosis, skewness, etc.
Chebyshev’s Theorem
This is basically just like the Empirical Rule except that it works for any distribution, not just bell-shaped normal-ish distributions. It even works for bimodal distributions, though it's so vague that it barely matters.
The Empirical Rule
The Empirical Rule is just a really basic rule of thumb for estimating the width of a bell curve based on standard deviation, or estimating standard deviation based on a bell curve (divide the width of the bell by 4).
The empirical rule and Chebyshev's theorem are just a couple of little rules of thumb which tell you some vague things about a distribution. You'll never see these again after the first test!
Sampling Bias
This video covers the particular sources of bias which cause a sample to either not be as representative as one would hope, or to cause the data yielded to not accurately reflect the sample.
Sampling Error vs Nonsampling Error vs Nonrandom Sampling Error
The title says it all, even if you spell it non-sampling and non-random, with a dash in there.
In statistics, you're taking a sample in order to find something out about the population. These videos cover the various ways that either a sample is not representative of the population, or the sample itself is representative yet the data you get from the sample isn't accurate to the sample (thus not to the population either). Random sampling error, Nonrandom sampling error (non-random), Nonsampling error (non-sampling).
Levels of Measurement (Ratio, Interval, Ordinal, Nominal)
These don't make a whole lot of sense when you first learn them, but after this video you'll hopefully see that if you understand the ratio level first, the others fall into place.
This chapter covers the "levels of measurement" -- ratio, interval, ordinal and nominal. They don't make a whole lot of sense when you first learn them, but after this video you'll hopefully see that if you understand the ratio level first, the others fall into place.
Systematic Sampling
Random sampling sounds good on paper, and in medical studies. But in the real world, it's way easier to use systematic sampling to grab a sample that's almost as good as random.