Hey everyone, this is my first time using this site and I'm hoping you can help me.

I'm working on a project that has the following specs and i'm not sure how to interlock the quotas. I would very much appreciate all the help I can get.

a. Total sample size of n=150 per cell for a total sample size of n=300) – One Test cell, One Control cell
b. Region: Ontario
c. Age: 25-40 (monitor 25-32 / 33-40)
d. 55% Male and 45% Female
e. Interlocked quotas cell x gender
f. Tequila Engagement: non-rejecters of Tequila who have consumed tequila in the past 6 months
g. Key Subgroups: a minimum readable base (n=50 per cell) of Sauza and of Cuervo users is necessary
h. Incidence: 10%

[/color]


Thanks!!

OK, so a few of the quotas is no problem at all, normal quota, you just build your quota normal.

a. Total sample size of n=150 per cell for a total sample size of n=300) – One Test cell, One Control cell

Limesurvey doesn't have a feature to close a survey if you have reached a certain total of people. But here it isn't relevant anyway, as you do not need n=300 in total, but n=150 per cell. Even this isn't relevant, because you have also an interlocked quota of cell x gender, so we apply the quota on this level only.

b. Region: Ontario

Just set zero quotas on all other options than Ontario. Done.

c. Age: 25-40 (monitor 25-32 / 33-40)

Depends on how you ask this question. If you use age brackets and those age brackets are exactly like your quotas here, then you can just create zero quotas on ages "0-24" and "41 or above". Monitoring needs to be done "manually" if you don't have any further rules here.

d. 55% Male and 45% Female

Irrelevant as a standalone, as you need to guarantee the quota per cell.

e. Interlocked quotas cell x gender

OK, here it gets a little tricky and you'll need to create a hidden single choice question with 4 options, and the answer to this hidden questions will be given via a equation question that you also need to create and hide.
The single choice question will have these for answer options (and quota):
1. Test cell - male (n=82)
2. Test cell - female (n=68)
3. Control cell - male (n=82)
4. Control cell- female (n=68)

As your quota of 55% vs 44% based on 150 completes is basically impossible (it would require half respondents), I went for 82 instead of 82,5 and 68 instead of 67,5. In general, as you usually always need some overrecruit because you will usually need to delete a few responses for quality reasons, you migh actually go for something like 85 male and 70 femal for each cell, rather than go with the exact quotas anyway (this is true for other quotas as well.

Before this question you will create an equation type question that checks the answers of the gender question and the the test cell (not sure how how determine who belongs to the test cell and who belongs to the control cell, but I assume you'll plan on creating a question question for this as well with a random number between 1 and 2.

So then you check, if gender is male and test cell you apply something like this in the equation: This code has not been tested and is just to give you a rough idea. Obviously I don't have your question codes and answer codes to give you exact information either.

What does this do? It checks if the gender is male and the cell is test (assuming that 1 means test), then it basically gives the question Qx (which would be our single choice question for the quota) the value 1. You would have to do the same for all 4 combinations from above.

f. Tequila Engagement: non-rejecters of Tequila who have consumed tequila in the past 6 months

Simple quotas. Here you probably need two, depending on your questionnaire. You would check if they reject Tequila and if so, you kick them out with a zero quota. Otherwise they can continue to the next question, where you ask when they drank tequila the last time. Then you set a quota on 6 months and above. If this includes more answer options you will have to use a hidden proxy question just like I showed above.

g. Key Subgroups: a minimum readable base (n=50 per cell) of Sauza and of Cuervo users is necessary

This will be very difficult to control. You can stop people to participate when they are above a quota, but quotas can't do anything to guarantee a minimum, because it depends purely on your respondents. It depends very much on the incidence of these two brands within your general population.

h. Incidence: 10%

The incidence is irrelevant for quotas, so I'll ignore this. Also, incidence based on what? People that are between 25 and 40, that do not reject Tequila and have consumed it in the past 6 months in general population? I find this estimated incidence rate pretty high, but hey, I don't know Ontario well.

Hi,

I din't look totally, but sometimes when adding restriction agaion and again and agin …
Sometimes is not possible …

I have it one time, when we need 2000 respondant,
- By sex
- By age
- By region
- By city size
- By Socio-Professional Category
- And some other …

Finally : it's too complex to have it, because it make one man in this town, one women in this other one etc …

Thanks Holch.

For the readable base, it'd be a minimum of 50 completes/drink correct?

For the readable base, it'd be a minimum of 50 completes/drink correct?

I don't see how to control this via the questionnaire and quotas. Because you are not asking for a maximum there, which you can control with quotas. But you want a minimum and there quotas can't do much.

If there are just not enough consumers of a certain brand in the sample, you can't do anything from within the questionnaire.

You would have to monitor the distribution of "Sauza", "Corvo" and "other". If one of the two brands falls much behind the other brand all "others" you would have to see if you don't have to set quotas for the brands that are used more frequent so that you still have the chance to fill the minimum quotas.

But the whole thing is rather something that needs to be discussed with the provider of the sample.

Now, I what I would not do is to set a quota of n=50 for Sauza and for Corvo, because this means that you might loose valuable and valid respondents if you have for example such a distribution at some point during the field time:

Sauza n=45
Corvo n=50
Other n=10

If you set the quota for Corvo and Sauza to n=50, Corvo drinkers could not participate anymore, but given your brief they could, becaus you need n=50 as a minimum, but n=60 would be perfectly fine as well, as long as you get n=50 for Sauza, which given the scenario above look perfectly achievable at this point.

So you actually need to monitor the distribution of these 2 brands closely to understand how likely it is that you reach or pass n=50 for Sauza and Corvo.

E.g. if you see that the distribution at one point is something like:
Sauza n=55
Corvo n=5
Other n=40

Then you will need to do something, because it looks like there are a lot more consumers of Sauza and Other, than Corvo and once you reach the total sample of n=300 you might not have enough "Corvo" drinkers if you don't do something.

this post has been deleted.