<<

. 4
( 62 .)



>>

2. The second version is the version for correcting the
calculation errors as well as the gross shortcomings in terms
of its usability. This version is a little easier to use and
has better accuracy in its calculations. It is also often at
this point that there is a sudden understanding into what
the model should have been all along, which leads to. . .




TLFeBOOK
Chapter 1
10




3. The third version is much easier to use and more elegant
in structure. Often, this is a radical departure from the
first two versions and comes after a smack-your-hand-in-
the-middle-of-your-forehead moment of insight. And
strangely, this is the one that comes much closer to what
the original concept of the model was.

MOUSE OR KEYBOARD?
The byword is ˜˜whatever works for you.™™ As you become more
and more expert at developing and working with models, you
will begin to find yourself spending more time with your PC.
This brings us to the question of whether it is better to use the
mouse or the keyboard to operate the menus and work with the
worksheets.
Using the mouse has the advantage of getting to some of the
commands more quickly and ˜˜intuitively,™™ but it has the disad-
vantage of taking more time and hand motion: your hand has to
leave the keyboard, find the mouse, position the cursor, click, and
then return to the keyboard. In addition, the mouse can lead to
wrist and elbow strain when you need to extend your
arm to handle the mouse, especially when there is little or no
support to the forearm. Using the keyboard has the advantage of
being quicker, and learning this method gives you the advantage
of being able to continue your work if for some reason you
cannot use the mouse. The disadvantage is that it can be quite
tedious to step through the menu system, especially when you
are confronted with a menu box with drop-down lists, tabs,
checkboxes, etc. However, some practice can make the hand
movements automatic, so that your hands will seem to have a
˜˜keyboard memory.™™
I do not recommend one over the other and can only say use
whatever works for you. Indeed, it might be that the best method is
a combination of the mouse and the keyboard.


A Suggestion for Mouse Placement
If you place your mouse to the side of the keyboard, an arrange-
ment that most people use, you can have overworked shoulder




TLFeBOOK
A Financial Projection Model 11




and elbow joints because your shoulder has to support your arm
as you work with the mouse. Additionally, this position forces
your hand to point outward as you work, creating an angle at the
outer edge of where the hand meets the wrist. It is possible to get
tendonitis at the point where the tendon kinks through the angle.
To minimize strain, place the mouse in front of you, between you
and the keyboard, rather than to the side. So, a view from the top
of the desk would be as follows:
Monitor screen
Keyboard
Mouse
Edge of desk
You
There are several advantages to this:
The arm can be supported by the elbow on your desk.
u

The position of the hand directly in front of you is also
u

more natural and closer to the center of your body. You
are more ˜˜centered,™™ to use a martial arts term.
It is just as easy, if not more so, to move your hand from
u

the keyboard toward your solar plexus than to move it out
to the right and putting your elbow and your shoulder in
a twist.
In this position, the hand holding the mouse will tend to
u

point toward the left side of your body (if you are right
handed), extending the outer edge of the hand and wrist
and reducing the possibility of tendonitis at this point.
In this location, given the curve of your arm, the most
u

natural position for the mouse is ˜˜sideways™™, with the
cable leading off to the left (again, if you are right-
handed). You will move the mouse to the left in order to
get the cursor to move ˜˜up™™ on the screen. This adjust-
ment, however, will be an almost instantaneous one.




TLFeBOOK
This page intentionally left blank.




TLFeBOOK
CHAPTER 2


Design Principles for
Good Model Building




This chapter covers the principles you should keep in mind.
These are meant to minimize confusion in building the model
and in using it. Remember, the confusion you avoid may be
your own.


KEY PRINCIPLES
When we design something that exists in a physical form in the
world, we have the benefit of having something to pickup, turn
over, peer into, kick, or thump when something is not working.
Additionally, if we are designing something like a car and find
that the dashboard lights are not working, it is a safe bet that
the problem lies with the electrical wiring or switches in the
vicinity.
Not so with spreadsheet modeling. Despite the fact that we
can see a model, it™s not actually ˜˜there,™™ and when problems
arise, we have only our mental map of it to use in figuring out
what is wrong. And, unlike a physical counterpart, a problem
in one area of the model can be caused by something else not
seemingly related to the problem at hand.
So the design principles we apply as we build our
model are critical. The more we can do things correctly the
first time around, the less trouble and confusion will result.
13



Copyright © 2004 by John S. Tjia. Click here for terms of use.
TLFeBOOK
Chapter 2
14




Some principles to consider:
KISS”Keep it simple, stupid.
u

Have a clear idea of what the model needs to do.
u

Be clear about what the users want and expect.
u

Maintain a logical arrangement of the parts.
u

Make all calculations in the model visible.
u

Be consistent in everything you do.
u

Use one input for one data point.
u

Think modular.
u

Make full use of Excel™s power.
u

Provide ways to prevent or back out of errors.
u

Save in-progress versions under different names,
u

and save them often.
Test, test, and test.
u




KISS
The overriding principle in model building is the ˜˜Keep it
simple, stupid™™ principle. The KISS principle does not mean
that a model should be simplistic and do nothing but the most
rudimentary of calculations. Rather, it means that whatever you
need your model to do, keep it simple. A variation of this is the
principle of Occam™s razor: the best solution is the simplest one.
Keep the formulas simple, even if it means using one or
u

more lines to break up the calculations. If you write a
formula and then look at it again 10 minutes later and
have a hard time understanding it, that is a sign that you
may want to break up the formula into two or more cells.
Keep the structure of the model simple, with a flow of
u

calculations that, as much as possible, go in one consistent
direction in the model, from the ˜˜beginning™™ to the ˜˜end.™™
Generally, you can consider the ˜˜top™™ sheet in Excel”
whose screen tab is at the leftmost at the bottom of the
screen”to be the beginning. The ˜˜bottom™™ sheet is at the
end. This will give the user a sense of the start and the end
of the model. A ˜˜simple™™ structure will mean different




TLFeBOOK
Design Principles for Good Model Building 15




things to different people. On the one hand, it may mean
that there should be only one sheet, with the beginning of
the model at the top and the results at the bottom. On the
other, it may mean that there should be several sheets,
with each sheet containing particular blocks of inputs or
calculations.
Keep your formatting simple, with just enough to make
u

visual distinctions on the screen to help your users,
without going into a psychedelic mix of florid colors and
heavy lines. Bold type is helpful for highlighting items
on the screen, but use it sparingly. If the screen holds a
profusion of bold type, then the highlighting effect is
gone, and the screen now looks visually ˜˜heavy.™™
KISS is a very beneficial principle to follow. Determine what
˜˜simple™™ means to you and those who will be using your model.
If there is a difference, go with the ˜˜simpler™™ of the two. The
more you can follow that standard, the more your work will
be used.
A good model should be powerful and fulfill its analytical
goals, allow its settings to be changed quickly and with reliable
results, and be fun to use. A truly great model ˜˜disappears™™:
the users use the model to get the results they want without
the model™s functions or interface design intruding into their
consciousness.


Have a Clear Idea of What the Model Needs to Do
Having a clear idea at the outset of what your model needs to do
is an absolute requirement. If you do not have a clear idea, the
best thing to do is to step away from the computer and continue
to think out what the model should be. A good way is to build a

<<

. 4
( 62 .)



>>