You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
HH reported with high food consumption score, but expenditure is below minimum expenditure threshold
HH reported with low food consumption score, but expenditure is above maximum expenditure threshold
Number of interviews with total food or non-food consumption expenditures of 0. Observing a household with 0 food or NF consumption expenditures is very unlikely.
Check values that do not make economic sense, e.g., food expenditures of 1 or 2 units when a unit of bread costs 50 local currency. If spotted, investigate the extent to which these values occur, and the potential reasons (e.g., misunderstanding with currencies or anomalies with a specific enumerator). These checks should be done with original values and, optionally, also on per capita values.
Check for outliers. Use different statistics related to individual expenditure items (and/or aggregates of various kind) by enumerator/admin, to spot incorrect application of the module. These checks can be done in different ways, but they should all be done in per capita terms and taking into account potential large difference in prices across survey areas. Examples of possible checks include:
o Comparing mean/median expenditure by enumerator to identify enumerators who might systematically overreport/underreport expenditure
o Identify outliers using a statistical procedure like that described in the “Cleaning” section and compute occurrences of outliers by enumerator
o Box plots can be a handy way of comparing median, minimum, maximum and extreme expenditure values by enumerator to detect possible issues with the administration of the module
o While some outliers are likely on the high-end (e.g., excessive health expenditures), whereby a household has anomalously high expenditures, enumerators might also have included extra zeros by mistake. So, it is best to verify with the field team, where possible.
Check for atypical values like ‘33’ ‘77’ ‘88’ ‘99’ or other overly specific, out of the ordinary figures to flag to enumerators and check if they meant not applicable (which is a practice that should not be applied in the expenditure module).
The text was updated successfully, but these errors were encountered:
Number of interviews with total food or non-food consumption expenditures of 0. Observing a household with 0 food or NF consumption expenditures is very unlikely.
Check values that do not make economic sense, e.g., food expenditures of 1 or 2 units when a unit of bread costs 50 local currency. If spotted, investigate the extent to which these values occur, and the potential reasons (e.g., misunderstanding with currencies or anomalies with a specific enumerator). These checks should be done with original values and, optionally, also on per capita values.
Check for outliers. Use different statistics related to individual expenditure items (and/or aggregates of various kind) by enumerator/admin, to spot incorrect application of the module. These checks can be done in different ways, but they should all be done in per capita terms and taking into account potential large difference in prices across survey areas. Examples of possible checks include:
o Comparing mean/median expenditure by enumerator to identify enumerators who might systematically overreport/underreport expenditure
o Identify outliers using a statistical procedure like that described in the “Cleaning” section and compute occurrences of outliers by enumerator
o Box plots can be a handy way of comparing median, minimum, maximum and extreme expenditure values by enumerator to detect possible issues with the administration of the module
o While some outliers are likely on the high-end (e.g., excessive health expenditures), whereby a household has anomalously high expenditures, enumerators might also have included extra zeros by mistake. So, it is best to verify with the field team, where possible.
Check for atypical values like ‘33’ ‘77’ ‘88’ ‘99’ or other overly specific, out of the ordinary figures to flag to enumerators and check if they meant not applicable (which is a practice that should not be applied in the expenditure module).
The text was updated successfully, but these errors were encountered: