Data entry in Excel - Dana Hince PhD

Data entry guidelines

using Excel

Dr Dana Hince and Professor Max Bulsara

Institute for Health Research, The University of Notre Dame, Fremantle

1
 Data entry in Excel - Dana Hince PhD

Data entry guidelines using Excel

Good data entry is a process that:
• Minimizes data entry error
• Minimizes the risk of misinterpretation
• Minimizes the amount of data manipulation required prior to
commencing analysis

Although there are specialised programs available for data entry, Excel is still
the most commonly used program we see used for this purpose. Most of the
analysis of the data, however, is conducted in specialised statistical programs
(e.g. SPSS, Stata or SAS). You will save yourself/your data analyst A LOT of time
if your Excel data spreadsheet is constructed in such a way that the statistical
program can “understand” it.

The more common mistakes we see in spreadsheet construction are a

consequence of thinking like a human and not like a computer. For example, 9
the number and 9 the character have the same meaning for people, as do the
words male and Male. We can interpret what we see within the context of the
presentation – computers aren’t so good at that. So when it comes to setting
up a spreadsheet, your life will be easier and you will get your results quicker if
you follow the “computer’s rules”.

By the end of this manual you will be able to engage in “good data entry” (as
defined at the top of this page) by achieving the following objectives:
1. Structure an Excel spreadsheet that allows easy importation into various
statistical packages.
2. Choose appropriate variable names, variable labels, and value codes and
labels.
3. Construct a codebook/data dictionary.
4. Format Excel cells.
5. Limit Excel cells to valid ranges or prepare drop down options.
6. Know what not to include in your spreadsheet.

2
 Data entry in Excel - Dana Hince PhD

1. Set up your spreadsheet.

a) Data spreadsheets can be wide (Fig1) or long (Fig2).

In both formats
• the columns represent variables
• the first column is a unique identifier for each participant
• the rows represent cases/events
• the rows must be uniquely identifiable

The difference between the two is how many variables are needed to
uniquely identify a row. Wide format needs only one identifier, whereas
long format needs two or more.

For example, compare Fig1 and Fig2. In wide format (Fig1) knowing the id
and variable name (id=1, variable=t1_bp) leaves only one choice of value
(120), whereas in long format (Fig2) knowing the id and variable name
(id=1, variable=bp) leaves you with two options (120 or 110). In this case,
we also need to know the value of time to limit the number of possible
rows to one.

Long format is often the format required for the analysis of longitudinal or
repeated measures study designs. It is possible to use statistical software
to switch between wide and long format (called restructuring or reshaping
the data). If your data requires restructuring, please DO NOT CUT AND
PASTE IN EXCEL. It is too error prone!

CAUTION! Regardless of format, make sure your unique id

matches any case report forms or other hard copy or

MAKE A START!
1. What is your spreadsheet format?
2. How is each row uniquely identified?
3. What are your id values?

3
 Data entry in Excel - Dana Hince PhD

Figure 1. Basic ‘wide’ spreadsheet format.

Figure 2. Basic ‘long’ spreadsheet format.

4
 Data entry in Excel - Dana Hince PhD

b) The first row is used for variable names.

Variables names
• are meaningful and short
• contain all relevant information
• have no spaces between words
• start with letters only (or _)
• only take one row (the first one!) in the spreadsheet
• are consistently named (e.g. heart_rate1, heart_rate2 NOT HR_1,
hrate2) THIS IS PARTICULARLY IMPORTANT IF ANY DATA
RESTRUCTURING IS REQUIRED (see page 3)

TIP! Using lower case letters only for variable names

helps reduce typos if you are using command driven
software

MAKE A START!
4. Test your knowledge on the next page then
5. Enter your variable names into a blank Excel

5
6
Answers.
a) Time1 and T1 refer to the same time. Needs to be consistently labelled.
b) Beginning variable names with numbers upsets many statistical programmes.
c) Variable names need to be contained in one cell only.
Figure 2a), b) and c). Examples of troublesome variable names.
c)
b)
a)
POP QUIZ 1: what is wrong with these variable names?
Data entry in Excel - Dana Hince PhD
 Data entry in Excel - Dana Hince PhD

c) Decide on codes for

• categorical variables i.e. use numbers to denote the categories
So gender could be 0 = female and 1 = male and 99 = missing data
• Although you don’t need to code continuous variables given the
measurement is already numeric (e.g. height, weight, blood pressure)
you will need a missing value code that is
i. NOT within the range of expected or probable values for
that variable (e.g. 99 would not be a good missing code for
diastolic blood pressure because it could be a true value).
ii. NOT text! If you use text (e.g. N/A) in a numeric variable,
some statistical packages assume that the whole variable is
text (string), and won’t be able to perform the calculations
you need it to do.

d) For free text variables (e.g. answers to open ended questions or

important notes from data collection).
• The number of variables required for each free text question depends
on the number of answers given.
For example:
i. If the question is “List 3 breeds of dog you would consider
as a pet” then you need to have 3 variables (e.g. breed1,
breed2, breed3).
ii. If the question is “List all the breeds of dogs you would
consider as a pet”, you need enough variables so that the
maximum number of breeds listed as an answer can be
accommodated with one breed per variable.
• Enter free text answers exactly as they appear on the survey.

e) Be careful with date variables.

To people, 14 May 2009 is the same as 12 May 09 is the same is 14/5/09 is
the same as 14/05/09. To your statistical package, they are potentially all
different. Always enter your dates into excel in the same format, and set
the date value cells to date format by
• Right click in the cell (or selected group of cells)
• Choose Format Cells in the drop down option (Fig. 3a)

7
 Data entry in Excel - Dana Hince PhD

• In the pop up box that follows, choose Date, and the format you
would like and click OK (Fig. 3b: I like the 3rd option down because it
includes the 4 digit year).

Figure 3a. Formatting cells

Figure 3b. Formatting cells containing date values

8
 Data entry in Excel - Dana Hince PhD

Make sure dates are entered in a consistent format, with 4 digit

years, because inconsistent/incomplete dates can equal hours of
data manipulation

MAKE A START!
6. Format all the cells that will contain date values now

2. Reduce the risk of data entry error or misinterpretation

a) Set up a codebook or data dictionary

The codebook is how you communicate what each variable represents, and
what to expect to find in it, to other members of your research team.

Figure 4. An example of a codebook/data dictionary.

• List your variable names with labels/descriptions in sufficient detail so

that someone else can understand what they are.

• Include the codes and value labels for any categorical variables.

• Include ranges and/or units of measurement for continuous variables

(e.g mmHg, ng/L) as this may not be obvious to others.
9
 Data entry in Excel - Dana Hince PhD

• Include codes for missing values.

MAKE A START!
7. Set up your data dictionary on the next sheet in your Excel
file, following Fig4 as an example

b) Excel Data Validation function

Now you have your data dictionary sorted, you can use this information to
reduce the risk of data entry error using Excel’s data validation tool.

This function allows you to specify allowable ranges/values for variables,

provide data entry prompts and/or data entry error messages. There are
many options available – let’s take two very useful as an example.

Firstly, let’s set up a drop down list with an error message for the variable
gender. Go to the Data ribbon on the top of the Excel window (1), and click
Data Validation (2).

1
2

The window below will pop up. Choose List from the Allow: drop down
menu (3).

10
 Data entry in Excel - Dana Hince PhD

Click on the button next to the Source: field (4) this will open another
pop up box.

Now choose the cells that have the values you want to allow as valid
input from your data dictionary. In this case, they are the codes for
gender (5: see also Fig 4). Now click the button marked 6.

11
 Data entry in Excel - Dana Hince PhD

You should be back at the Data Validation pop up, with the cell
reference now in the Source: field (7). Now click the Error Alert tab (8).

Enter in the error message that you would like to appear if someone
makes a mistake in the error message box (9), and click OK.

12
 Data entry in Excel - Dana Hince PhD

When you place the cursor in the cell/s you just set up, a little arrow will
appear next to the cell. Click this to access the allowable values.

Now try putting in an invalid number. Your error message will appear
when you move to the next cell.

13
 Data entry in Excel - Dana Hince PhD

There are other options in the Allow: drop down list in the Data
Validation window. Let’s suppose that our id values can only be between
100 and 200. Choose Whole Number (1), and in the Data: drop down
menu that appears (2) choose between and then enter the maximum
and minimum of the allowable range for that variable (3).

14
 Data entry in Excel - Dana Hince PhD

Again, you could set up an error alert that will appear if someone tries to
enter an invalid id value.

A bit of extra time at the beginning setting up these Data Validation

rules will result in reduced risk of data entry error, and therefore a lot
less double checking later on when you come to “clean” the data.

For further “how to” see:

https://support.office.com/en-us/article/Apply-data-validation-to-
cells-29fecbcc-d1b9-42c1-9d76-eff3ce5f7249

https://support.office.com/en-us/article/Create-a-drop-down-list-
7693307a-59ef-400a-b769-c5402dce407b?ui=en-US&rs=en-

MAKE A START!
8. Consider your data and how you might use Excel data
validation to reduce data entry error
9. Set up your variables accordingly!

3. Enter away!
• Take lots of breaks if entering “in bulk”. Fatigue is your enemy when it
comes to accurate entry.

• Try and enter data in real time if at all possible as this might give you
the chance to collect missing data values before it is too late….

• If you have multiple people entering data, make sure that they have
access to the data dictionary and please use the data validation tool!

15
 Data entry in Excel - Dana Hince PhD

4. What NOT to include in a spreadsheet.

• Column or row means, standard deviations or any other summary
statistic calculated in Excel (if you have used them to do preliminary
data checks, just remove them before importing into statistical
package/sending to the statistician).

• Empty/blank rows or columns.

• Text and numeric values in the same variable. If you find yourself doing
this you need two separate variables.

• Colour coding – this is fine if it helps you with ensuring accurate data
entry, but all the relevant information needs to be included in the
variables.

• Identifying information (e.g. names address etc). Keep a separate,

password protected spreadsheet which links the ID number with the
participant identifying information.

16
 Data entry in Excel - Dana Hince PhD

POP QUIZ 2: What is wrong with this spreadsheet? Can you find the errors? How would you
fix them?
Hypothetical data comparing meditation vs no meditation treatment effects on blood pressure, body temperature and 3
anxiety questionnaires, pre (time 1) and post (time 2) intervention.
id test 1
sbp temp anx_q1 anx_q2 anx_q3 2_SBP t2_temp t2_anx_1 t2_anx_2 t2_anx_3 gender
1 120 37 3 3 4 121 37 3 6 5 M
2 124 37 5 2 6 124 36.9 5 6 6 m
3 125 37.1 6 0 5 123 37 6 6 9 f
4 126 36.9 9 2 135 36.9 8 8 9 fem
5 135 36.8 10 9 9 110 36.8 5 9 3 male
6 101 37 2 7 8 112 37 8 6 6 male
7 26 36.6 4 6 3 120 37 5 6 8 female

8 180 37.2 6 6 8 121 9 3 7 f

9 125 37.5 8 3 5 117 37.1 6 6 8 f
10 115 37 7 5 6 160 36.1 5 9 4 F
11 116 37.4 6 8 8 138 37.4 3 9 5 f
12 138 36.8 3 4 9 124 36.9 1 5 6 Male
13 117 36.9 9 9 2 120 36.8 2 8 9 Male
14 106 37 4 3 9 115 37 5 4 1 F
mean 118.1429 61.12143 124.2857 61.90769

Meditation
No
meditation

17
18
P.O Box 1225, Fremantle, WA 6959
19 Mouat Street,
University of Notre Dame, Fremantle
Institute for Health Research,
[email protected]
Chair of Biostatistics
Professor Max Bulsara
or
[email protected]
Research Biostatistician
Dana Hince PhD
For further information please contact:
Answers.
1. Two rows used for the variable names; these should be in one row.
2. Variable name starts with a number; instead put the number at the end.
3. Variable names are inconsistent; choose one format and stick to it
4. Gender is not numerically coded; assign numbers to male, female and
missing and add codes into the data dictionary. NB it is possible to ‘encode’
text values in statistical packages, but it requires extra steps and it isn’t easy
if the text coding is not consistent as is the case here!
5. Row means are included; delete them.
6. Colour coding used to define treatment groups; create another variable that
assigns numbers to the groups and add into the data dictionary.
7. Missing values are not coded (in anx_q2 and T2_temp); decide on an
appropriate code, enter into spreadsheet and into data dictionary.
8. Where is the dictionary?
Data entry in Excel - Dana Hince PhD