Video Transcript
A lamp manufacturer has two lines
of fabrication, line one and line two, and the work is organized in two eight-hour
shifts. The quality manager plans to
introduce new quality checks using new power measurement devices. They decide to measure the power of
the lamps exiting line one between 10:00 O’clock and 10:15 AM every day for one
week. Is their sample biased?
In this problem, there is a
manufacturer with two production lines or fabrication lines, and each of these lines
produce lamps. The staff working on the lamps work
in eight-hour shifts, and there are two of these shifts per day. For example, the first shift of
staff might work from 6:00 AM to 2:00 PM and then be replaced by the second shift,
which works from 2:00 PM to 10:00 PM.
We are told that the manager is
going to perform a quality check using new power measurement devices. So that means that they will take a
sample of the lamps. And so they measure the power. That is, they take a sample of
lamps exiting line one between 10:00 O’clock and 10:15 AM every day for one
week.
Let’s recall that a biased sample
is a method of forming a sample which favors certain values of the variable of
study. The variable of study here is the
power of the lamps. So are there any problems with the
approach which is taken by the quality manager?
Well, firstly, the manager is only
sampling from one line. We don’t know if there are any
problems with the equipment used in this line or the workers on the line. Secondly, the manager is sampling
at the same time of day every day between 10:00 and 10:15 AM. That means that the people working
on the line may not change. For example, they might be working
on the same shift every day. This is a problem because then we
are sampling lamps which may be made by the same group of people. For example, there could be one
worker who sometimes wires the lamps with mistakes. If they always worked the morning
shift, then the manager is more likely to sample lamps that have problems with the
power.
Therefore, this sample is
biased. We need some randomization in order
to create an unbiased and representative sample. Importantly, each lamp produced by
the manufacturer must have an equal chance of being selected for the sample. And so the answer is yes, the
sample is biased.
