Menu ENN Search
Language: English Français

Maximum % of FLAG in SMART surveys

This question was posted the Assessment forum area and has 2 replies. You can also reply via email – be sure to leave the subject unchanged.

» Post a reply

Anaïs

Emergency Nutrition Advisor

Normal user

28 Jan 2013, 07:34

Good morning all,

I would like to know what are the maximum % of SMART Flags for WHZ, HAZ and WAZ when analysing the quality of an assessment?
In plausibility check report we're given some WHZ, HAZ and WAZ % but how to interpret them? When do we start saying that we observe too many flags for an index?

Thanks.

Anonymous 730

Nutrition and Food Security Officer

Normal user

28 Jan 2013, 09:52

Hie
Anything less than 5% for missing and flagged values is good and anything below 10% is acceptable.
Regards

Mark Myatt

Consultant Epideomiologist

Frequent user

28 Jan 2013, 15:22


If ...

(1) The data are from a population survey rather than (e.g.) a SAM patient cohort.

(2) Survey staff were well trained in weighing, measuring, and recording.

(3) You have done double-entry, followed by one or more validation, correction, validation cycles.

(4) You have corrected obvious errors (e.g. misplaced decimal points) or censored records with clearly faulty data.

(5) You are using sensible default flagging criteria (software defaults in ENA and EpiInfo could be used

... then you should see very few flagged records.

From my own surveys, I cannot recall this ever being more than about a dozen records in sampling sizes of 700 or more (i.e. less than about 2%) for WHZ. You may see more than this for HAZ or WAZ because it is common to have large errors in age.

In any case, you should check flagged records against paper records as falling may be due to copying errors or miskeying of data.

I hope this helps.

Back to top

» Post a reply