Missing values
The data for all variables of one case (patient, sample) are entered on one row in the spreadsheet. When
for one variable you do not know the value (or entry) for the case, you leave the corresponding cell blank
and do not enter any data in this cell.
As a rule, the program will ignore an entry for a numeric variable when it is unable to interpret this entry as
a number.
When text is entered in a cell for a numeric variable, the program will not take this case into account for
calculations (it will not substitute the text value by a zero).
The following are recognized as numbers:
5.4
LOG(36.5)
HEIGHT/WEIGHT
(when `Height' and `Weight' are correctly defined variables)
HEIGHT/100
The following are not recognized as numbers and are ignored for calculations:
5,8
`4.6
LOG(CONC)
(when `CONC' is not a correctly defined variable or in case the variable
`CONC' has a zero, negative or missing value)
SQRT( 9) (error!)
1/ HEIGHT
(when `HEIGHT' is not a correctly defined variable or in case `HEIGHT
equals zero)
Data checking
After having entered the data, you should carefully check the data to ensure that they have been entered
correctly.
Sometimes erroneous data input will become apparent when looking at the data range in the summary
statistics report (e.g. maximum value of 78 for pH), or when plotting box and whisker plots, dot plots or
scatter diagrams for the different variables. You should check clear outliers since they may indicate
incorrect data entry, or they may result from a technical failure in measurement or from a study protocol
violation. Only for such plausible reason a value may be excluded from further analysis (and not simply
because a value is the smallest or largest). If there is no evidence of such a mistake then the value must
remain unaltered.
You can locate any value in the spreadsheet using the Find procedure (p. 28).
You can exclude outliers from further calculations by using the Exclude command (p. 37).
6
New Page 1