Come Play with Me
[ YouTube.com ]
In connection with development of convenient procedures for non-compensatory modeling estimation and simulation, a need for a simple, sufficiently flexible and operable simulator appeared. No such tool was at hand, so a new one had to be created. Among other, the aim was to avoid inconveniences met with some other solutions.
The simulator was developed for MS Excel 97 and later. The program structure allows for a relatively easy modification of a simulation model or add a new one. Users are not supposed to do such changes.
● Each simulator worksheet is self-contained.
Any simulator sheet can be copied using right-click on the worksheet
lower-tab, selecting "Move or copy" and checking "Make copy". Input
parameters of the copy can be set independently from the source.
● Only one segment variable with non-empty segments is
allowed.
Our experience with more than one segment variables allowed is worse than
bad. Lowly populated combination segments are an often cause of
misinterpretation.
● Simulator uses only clearly identifiable rectangular range
of the worksheet area.
The simulator working range is positioned relative to its upper left
corner cell named "SimulatorRangeStart". This allows to add an arbitrary
number of rows or columns outside the used range. The simulator status
cell (showing "WAITING" or "DONE") is added to the simulator working
range.
● Profile input parameters are
of two types: discrete or continuous quantitative. Input is entered in the upper block of cells of the simulator working range. |
|
- | Discrete levels are recognized by (roll-down) combo-box. |
- | Quantitative parameters are entered directly as cell values. The underlying part-worths are interpolated from the closest values of estimated levels |
● Prohibitions between attributes are respected.
When Run-simulation button is presses a warning is issued if input
parameters are inconsistent.
● User can create a scenario with any number of products up
to a predefined number.
The uppermost parameter is a binary switch for inclusion or exclusion of
the profile from simulation.
● Any row or column of the simulator working range can be
made hidden.
However, deletion of any row or column from inside the range will corrupt
the simulator.
● The simulator is an numeric application.
Formatting of any cell on a simulator sheet is fully in hands of user with
no limitations imposed.
● Simulator sheet is neither locked or protected.
Users can add arbitrary functions anywhere on the sheet except the working
range. Functions can use data from anywhere on the sheet. This is useful
when there are additional data to be used in computations using results
from a simulation.
● The first choice model (FCM) is not included.
With the common number of several hundred respondents in a study the
results have little credibility. The model is a pure wasting the
hard-earned information from the interview. Only probability-based
estimates are computed.
● Descriptive statistics for the
estimates are omitted.
Non-compensatory models would lead to heavily biased statistics. In our
view, even statistics for parameters of additive compensatory model
provide little information.
● The "none" alternative
cannot be set as a profile.
The "none" alternative in CBC is simply an escape for respondents when
they do not know or do not want to give an answer. Validity of its utility
is low. It is not a reasonable replacement for the outer goods it should
be.
A working example of the simulator can be downloaded from page Relaxed Non-compensatory Simulation.