Intro to Statistics at IU Indianapolis - Connecting Measures of Central Tendency & Dispersion
Overview
Andy Baker connects measures of central tendency and dispersion to visual summaries, showing how to build a five-number summary and box plot, apply the 1.5 × IQR outlier fences, and compare box plots with histograms. He uses the shape and measurement level of data to guide choices: median and IQR for ordinal or skewed data, and mean, standard deviation, and histograms for symmetric data; course logistics also preview an online normal-distribution lecture and SPSS labs.
Key takeaways
- The five-number summary is minimum, Q1, median, Q3, and maximum; Q1, median, and Q3 correspond to the 25th, 50th, and 75th percentiles.
- John Tukey’s outlier rule sets cutoffs at Q1 − 1.5 × IQR and Q3 + 1.5 × IQR; whiskers reach the most extreme observed values within those cutoffs.
- Ordinal or skewed data is generally summarized with the median and IQR, while symmetric data is summarized with the mean and standard deviation.
- Histograms show distribution shape and clusters, whereas box plots efficiently show quartiles, median, spread, and potential outliers.
- Income is often right-skewed, making median and IQR more representative; approximately symmetric standardized test scores are better described with mean and standard deviation.
Chapters
- Andy Baker explains partial credit on Quiz 1, including that ratio is the best measurement-level answer for a variable with a meaningful zero.
- A histogram displays bars, while a frequency polygon shows the same frequency pattern as a line graph.
- Discrete values are whole-number counts, such as number of siblings; continuous values can include decimals, such as a temperature of 23.5°.
- Online Lecture 2 covers the normal distribution and is the sole basis for Quiz 3, which will be completed online; the next in-person paper quiz is Quiz 4.
- Lab 3 is available in SPSS and applies the measures covered in class, but unlike earlier labs it has no Clickology walkthrough video.
- For large SPSS variable lists, right-click to display and alphabetize variable names, then use Ctrl+F on Windows instead of scrolling through thousands of entries.
- For some IU Anywhere SPSS problems, Baker suggests opening Utilities, then OMS Control Panel, and selecting End All.
- The five-number summary consists of the minimum, Q1, median or Q2, Q3, and maximum.
- Q1 marks the 25th percentile, the median marks the 50th percentile, and Q3 marks the 75th percentile.
- Range, IQR, variance, and standard deviation require numeric data, so they cannot be calculated for label-based nominal variables such as rank or type of work.
- Baker starts by sorting seven exam scores from lowest to highest, then finds the median, Q1, and Q3 from the appropriate halves of the data.
- In the worked example, Q1 is 15, the median is 18, and Q3 is 22, giving an IQR of 22 − 15 = 7.
- The box spans Q1 to Q3, with a line at the median; the plot is based on medians and quartiles, not means.
- The example’s IQR of 7 gives a fence distance of 7 × 1.5 = 10.5, using the rule associated with John Tukey.
- The lower fence is Q1 − 10.5 = 4.5 and the upper fence is Q3 + 10.5 = 32.5.
- Whiskers extend to the most extreme observed values still within the fences; values beyond a fence, such as the example’s 45, are plotted as outliers.
- The fence values are cutoffs, not necessarily whisker endpoints: the minimum of 12 is within the lower fence, so the lower whisker reaches 12.
- A histogram reveals distribution shape and possible bulges, while a box plot gives a compact view of spread, quartiles, median, and outliers.
- Baker compares both displays using class data, including commute times and sleep hours; the commute distribution reflects students with zero-minute trips from dorms.
- The displays complement one another: histograms help reveal skew and clusters, while box plots make the middle 50% and extreme values easier to summarize.
- For ordinal measurements, including ranked responses such as strongly disagree to strongly agree, use the median rather than the mean.
- For skewed data, use the median and IQR because extreme values can distort the mean and standard deviation.
- A box plot is a useful display for skewed data, and visualizing a distribution helps identify skew that may not be obvious from raw numbers.
- A symmetric distribution has similar sides and often resembles a bell-shaped mound, with many observations near the center and fewer at either end.
- For symmetric data, Baker recommends the mean and standard deviation, with a histogram to show the distribution’s shape.
- The normal-distribution material in Online Lecture 2 builds on recognizing whether data is symmetric or skewed.
- Income is typically right-skewed because many people earn comparatively less while a small number earn very large amounts; median and IQR are therefore more appropriate than mean and standard deviation.
- Baker illustrates the effect of extreme wealth with the example of Bill Gates entering a bar and sharply raising its average income.
- Standardized test scores are presented as a roughly symmetric example, with most scores near the middle and fewer near the lower and upper limits; use the mean, standard deviation, and histogram.
- The choice of summary depends on the distribution: skewed income calls for median and IQR, while approximately symmetric test scores call for mean and standard deviation.
Transcript excerpt
0:00 Um, so I'll only do it for this first one and I I take it as kind of a a moment to kind of get to know how I do things and how how we do things, how we roll. So, uh, but I'm not going to come in here have a set time and organize all of that. That's going to have to be, uh, led by you. So, when you're ready to do that, let me know. Uh, and so who, you know, we'll we'll we'll figure out that time. The ideal time, uh, to ever do something like this would be before class. So, if you don't know, now you know that my office hours are in here the hour before this class. So, I get it. Some of you might have a class during that time. Um, so we can work something out. Obviously, we've got time after as well. Um, so, you know, there's starting to see a pattern here where students come in and ask questions
0:47 during that 9:30 to 10:30 period. Um, and so, yeah, I I highly suggest doing that if you want to. not required at all, but that would be a good time to do the makeup or I'm sorry, a redo of of quiz one. So, there's the results. Like I said, um, as far as what you see, if you have any questions as far as what the handwriting, what you see in red is, uh, just let me know. Uh, I do do, as you can see, I do do, uh, I, uh, I have partial credit for certain things. Um, so yeah, you get some weird scores, 75s. Uh so I do you know like for example um oh an example of something I would give
This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.
Related notes
Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, Andy Baker.