Basic HTML Version
Q
uantitative
R
easoning &
P
roblem
S
olving
208
© 2014 Pacific Crest
O
ops
! A
voiding
C
ommon
E
rrors
●
Data values are clearly outliers from existing values
Example
: The height of the 10 year-old female was reported as, “78 inches.”
Why?
In statistics, we study variation and outliers (data that is outside of a common or
expected range) have a significant contribution to variance. Therefore we need to be
sure that this very unreasonable value is accurate.
●
The data doesn’t represent the purpose
Example
:
The restaurant is determining if the menu matches their current and future clientele
but asks questions such as, “Was the temperature comfortable?”, “Was the lighting
adequate?”, “Would you appreciate live music during your dining experience?”, and
“If we extended our Friday and Saturday hours to 1:00am, would you be more likely
to visit us later in the evening?”
Why?
If the data collected does not target the problem identified, we have done nothing to
solve the problem. The design of the instrument for both measuring and collecting the
data must align to the purpose.
●
Mistakes in the recording of data where one value was missed, affecting all subsequent data
Example
: When Mindy took her exam, she accidentally skipped question 17 and filled in the
scantron bubble for 17 with the answer for 18. Every subsequent response, and there
were 100 of them total, was off by one, something she didn’t notice until she got to
question 100 and noticed a left over blank on the scantron.
Why?
Mindy did fail the exam, though she actually knew all the correct answers. The align-
ment of data is a critical part of the validation process when it comes to recording data.
●
Didn’t follow pre-defined conventions in use of the forms (survey or observational)
Example
: Cody created and distributed a survey. When he analyzed the results, they didn’t
make much sense. His roommate looked at the survey and saw that while Cody’s
survey asked good questions, the five options offered for every question’s response
were organized in a way that differed from most surveys.
Cody’s survey:
agree
completely
disagree
completely
disagree
somewhat
agree
somewhat
neutral
Typical surveys:
agree
completely
agree
somewhat
neutral
disagree
somewhat
disagree
completely
Why?
In the case of Cody’s survey, most people saw that the first option was “agree
completely” and assumed, based on past experience, that the option at the other end
of the scale would be, “disagree completely”, as those are the two extreme answers,
at either end of a spectrum of agreement. In general, having various conventions that
can’t be normalized after the data collection makes the integration of sub-sets of data
into a master data set just about impossible. It is important to establish and follow one
set of conventions from the beginning to get seamless integration of the data.