Describe changes that could be made to the data collection process that would result in an experiment rather than an observation study. Also, offer...

In order to
make the observational study of bank tellers more scientific, it would be important to collect
data both before and after the tellers take the advanced training program. It would also be
beneficial to collect data on tellers from multiple banks, some of whom do not receive the
training or who receive a training that isn't intended to build relevant new skills. Comparing
these sets of data allows for more certainty that the teller's error rate is related to their
participation in the training program. This process relies on the idea of isolating the
independent variable (participation in the training program) so that it is the only thing which
could affect the dependent variable (error rate).

In the case of this
experiment, lurking variables which might go unmeasured but actually account for change in the
dependent variable include other training programs that the tellers participate in, life or work
stress faced by a teller, or how recently a teller was trained.

Unseen biases
may be expressed by the person who evaluates errors, as when they are more critical of some
study participant's performance than of others.