Summary Measures Worksheet
Reflective Piece (Unit 7.1)

What?

This worksheet focused on applying core summary measure concepts in a hands-on way, using Excel functions to calculate and interpret mean, standard deviation, median, quartiles, interquartile range, and frequency and percentage frequency tables. The workbook walked through worked examples and then required me to complete parallel exercises, with my answers entered in red.

The exercises used two types of data, continuous outcomes (weight loss under Diet A vs Diet B) and nominal outcomes (cereal brand preference across two demographic areas). I completed the calculations and then interpreted what the results meant, rather than stopping at the numbers.

So what?

What made this assignment practically useful is that it treated summary statistics as decision tools, not just math to memorise. The Diet A vs Diet B section was a good reminder that “average” is not a complete story. Yes, the mean weight loss is higher for Diet A than Diet B, but the real interpretive step is asking whether that improvement comes with more inconsistency between people. Comparing standard deviations and then comparing the medians and IQRs forces you to think about both performance and reliability in the same breath.

That maps almost perfectly onto my day job in the exams department. When I report pass rates, the temptation is to summarise everything into one clean number, because it is easy to communicate. But a single average can hide important patterns. For example, an overall pass rate might look fine if many students pass on the second or third attempt, while first attempt performance is actually slipping. The worksheet’s split between mean and median, and between standard deviation and IQR, mirrors a real reporting dilemma for me, am I describing a typical student’s outcome, or am I being pulled around by extremes, retakes, or small pockets of unusually high or low performance?

The brand preference exercise felt simple at first, until I noticed the key lesson. Raw rows are useless for decision making, a frequency table is what turns noise into something you can act on. In the worksheet, once the data were summarised properly, the pattern became obvious, Area 1 leaned heavily toward “Other,” while Area 2 leaned more toward the named brands, and Brand B was consistently ahead of Brand A.

Again, this is basically the same move I have to make with exam data at scale. With thousands of records, you cannot “inspect” your way to insight. You have to group, count, and normalise into percentages. And once you do, you start seeing patterns that matter, differences across cohorts, departments, curricula, or changes after a policy update. The worksheet was a small, controlled version of that reality, take messy individual level data, summarise it responsibly, then interpret it without overclaiming.

If I am being honest, the critical part for me was noticing how easy it is to tell a different story with the same dataset just by picking a different summary. Mean vs median, overall rate vs first attempt rate, combined categories vs split categories, none of these are automatically wrong, but each choice pushes the reader toward a different conclusion. This assignment made that tradeoff feel concrete, not theoretical.

Now what?

Going forward, I want to treat summary measures as part of an “honest reporting kit” rather than a checklist. In practice, that means I will default to reporting at least one measure of centre and one measure of spread when I present performance results, especially when results will be used to judge a department or trigger interventions. If I use a mean, I want to sanity check it with a median. If I show an overall pass rate, I want to pair it with a first attempt rate or an attempt breakdown when that context changes the interpretation.

I also want to be more deliberate about when I convert counts into percentage frequencies. In my environment, cohorts vary in size and exam volume can spike, so raw counts can accidentally exaggerate differences that disappear once they are normalised. The worksheet reinforced that percentages are not just “nice formatting,” they are often the only way to make comparisons fair.

Overall, this workbook strengthened something I already knew intuitively from work, but did not always articulate clearly, summary statistics are not neutral, they are choices. The responsibility is to choose measures that match the decision being made, and to present them in a way that makes the underlying story clearer, not cleaner.

References

• University of Essex Online (Unit 7.1). Summary Measures Worksheet (completed workbook submission).