Mudanças entre as edições de "Usuário:Abdo/MMMNS/en"

Edição das 07h36min de 3 de junho de 2009

Ni!

Usuário:Abdo/MMMNS/RDS/en

Questions

For each interviewee, the questions below shall be asked for every name, there being 24 names total.

Only the order of names will be sampled, according to the Instructions for the deck of names, and must be registered.

Question-2 is asked only if the answer to Question-1 was not zero.

Question 1: How many people do you know named __________[name] ?

Question 2: How many of those know you __________[condition of the study, ex: had sex with another man in the last six months]?

Segmentation with 4 questions (the 2 above + 2 extra)

Definitions

For a given NAME:

t = "total number of alters named NAME that the respondent knows"

k = "has name NAME and knows respondent uses drugs"

u = "has name NAME and respondent knows he uses drugs"

xy = x and y

~z = not z

Example

~k~u: alters who do not use drugs and have no knowledge of the respondent's use

Equations

ku + k~u + ~ku + ~k~u = t

ku + k~u = k

~ku + ~k~u = ~k

ku + ~ku = u

k~u + ~k~u = ~u

Questions

?t = T

?k = K

?u = U

?ku = KU

Examples

~k~u = ~u - k~u = t - u + ku - k = T - K - U + KU

k~u/~u = (k - ku) / (t - u) = (K - KU) / (T - U)

Segmentation with no questions

Let me propose that we ask NO QUESTIONS.

Instead, we present the respondent with tokens and a sheet of paper divided in four labeled quadrants:

1. knows you use drugs & uses drugs
2. knows you use drugs & does not use drugs
3. does not know you use drugs & uses drugs
4. does not know you use drugs & does not use drugs

The respondent is asked to distribute the tokens into the quadrants, one for each person he knows with that name.

The interviewer counts the tokens and writes down the four numbers, then proceeds to the next name.

This gives us all relevant information directly, and as a bonus makes the interview a bit more fun and perhaps quicker.