EpiData Analysis -
Copyrights 2002-2006 Epidata Association www.epidata.dk
Differences from Epi Info Version 6
Features not implemented or replaced
Outputting summary data sets (sumfreq sumtables) is implemented as the command Stattables
- Options for single commands will be different
Update only affects memory use savedata command to store to disk
Other differences
- In Epidata Analysis, use the dash to indicate a list of variables in the order in which they appear in the questionnaire and in the Variables display (<F3>) (Epi Info uses "TO")
- Note no spaces are allowed before or after the dash. Wild cards are also allowed. Here are some examples:
- list v1-v9
- list * List all variables
- List V* List all variables that start with V
- List V?S List all 3 character variables that start with V, end with S regardless of the second character. E.g. variables called V1S, V3S and VQS will be listed
- Please report problems with parsing as given here to the EpiData-list
- Commands can be of any length (not limited to 255 characters)
- Handling of missing values may differ. EpiData's rules for missing values are:
- In unary (e.g. x=-x) or binary operation (x=z+y) a missing value propagates if both operands are integer or real numbers (does not apply to string or boolean)
- Comparison operations: missing data is not handled yet
- Function calls e.g. x=sqrt(y) x will be set to missing if y is missing
- Special date fields
- Statistics differences
- 2x2 epidemiological tables are made with a special table command options
- Confidence Intervals for the mean are based on the T-distribution with N-1 degrees of freedom. Not on z (Normal distribution).
- Note percentiles are weighted for distance btw. observations.
Ref: Bland M.Introduction to Medical Statistics, Oxford, 1995. Second Ed. p 55
And for N<20 a lookup table is used, see http://www.epidata.dk/documentation.php