Week 3 Lab: Lab 3 — Census Clinic

Bring your assignment data in progress. We break it, together, before the deadline rather than after.

8 March 2027

Before the lab#

Collect something. Twenty minutes of real data minimum, on the queue you intend to use for the Queue Census, which is due at the end of this week. Partial and messy is fine. Nothing is not.

In the lab#

A clinic rather than a class. The format is that you present your data for four minutes and the room tries to break it, and the reason we do this before the deadline is that every objection raised here is one you can still go and fix.

The four questions that come up every year, so you may as well pre-empt them:

Where does your queue start? If somebody is hovering near the end of the line looking at their phone, are they in it? Your answer must be a rule you applied consistently for the whole observation, not a judgement you made per-person.

Did you count the ones who left? Balking and reneging are invisible in a tally of people served, and they are exactly the customers a long queue produces most of. A queue that appears to be coping may simply be shedding.

Was it steady? Show your queue length over time. If it starts at two and ends at thirty, your window contains a transient and Little’s Law will hand you a confident wrong number.

What changed while you watched? Servers speed up when the line grows. A second till opened. Someone went on break. Any of these makes your service rate a moving target, and the honest response is to report it rather than average over it.

Afterwards#

Fix what the room found, finish the observation, and submit by 17:00 Friday. If the clinic convinced you that your queue is unmeasurable, that is a legitimate finding — write it up as one, with the evidence.