In the end, much of your data quality will depend on the forms from
which it is entered, who does the entry, what kind of checking you do,
and how good your review of the entered data is.  There will always be
errors, you might better spend your time trying to identify them than
writing data entry front-ends.  It depends on the magnitude of your
data entry task.

Excel can also be set up to present a pull-down list of values, where
that's appropriate (e.g., words or phrases).  If you have data with a
limited number of descriptive answers (e.g., Likert scale items), code
them numerically, enter that, and reassociate the lables later.
Simply setting the size and spacing of the cells in the sheet can
increase accuracy quite a bit.  The defaults are relatively small and
cramped for accurate, fast scanning and entry.  Increase the font
size; use pull-downs where appropriate; use two people to enter, one
to read values one to type, code complicated non-open-ended responses.
That can go a long way to making things more accurate.

What will work for you will depend on the quantity of data that needs
to be entered, also.  If you have 30 subjects with 50 values each, you
may want to just do it and check harder.  If you have 3000 subjects
and 500 values each, then that would justify more time spent designing
a front-end, perhaps.

A well-designed data entry front-end can be a boon, but they can also
be complex to program, and it's not clear that design principles are
easily found or followed, as so much depends on what the source from
which the data is being entered is and who is doing the entry.  The
requirements for production data entry from record forms is quite
different from designing web forms to be completed once per person (or
person/observation).

In the old days, we had data entered twice, then ran a program to
identify discrepancies between the two sets of entered data, as well
as to check for legal values and other easily detected anomalies,
e.g., mammalian fish.  We had access then to experienced data entry
people, though, and we replaced this system with the
one-reader/one-typist system and had them alternate about every 10-15
minutes, which seemed to work better with younger people.  Sampling
subjects from the entered data for full review is also useful to get
an idea of overall accuracy.

Not all data entry systems will work well with all data entry people
not with all kinds of data, so spend some time doing some
experimentation with it, and you'll probably come up with something
relatively smooth.  Having at least two people try any proposed system
is a huge help.

Just some opinions; feel free to take anything useful and leave the rest.

Good luck!

-- bennet




On Mon, Feb 2, 2015 at 2:56 PM, Ethan White <[email protected]> wrote:
> On Mon, Feb 2, 2015 at 12:39 PM, Caroline Li <[email protected]> wrote:
>>
>> I've heard that entering data manually into spreadsheets (which I've
>> always done) is frowned upon. Is there free software available for data
>> entry? Apparently some people use Google Forms, but maybe there's an option
>> better geared to scientific data?
>
>
> I prefer entering data via a database management system because they can be
> configured to offer a lot of quality control to prevent bad data from being
> entered. We have tended to use MS Access, but there is an open source
> equivalent in Libre Office Base. Any data entry into a database management
> system will at least enforce types (you can't accidentally enter a letter
> when it should be a number) and will force the data to be properly
> structured. By doing data entry through "forms" you can make entry easier
> and more robust. For example, instead of typing in a species name you could
> have the user select it from a drop down (no more typos) and if a value is
> logically bounded between 0 and 100 you could prevent it from taking other
> values.
>
> _______________________________________________
> Discuss mailing list
> [email protected]
> http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

_______________________________________________
Discuss mailing list
[email protected]
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Reply via email to