Calculate the sample number by age and gender

#1
Hello everyone! Beginner question: I used this site to calculate a survey sample:
https://www.surveymonkey.com/mp/sample-size-calculator

I compiled it like this for a survey of italian's habit:
Population size (Italian): 60,000,000
Confidence level: 95%
Margin of error: 5%

The result is 385, but I would like to divide it into males and females and by 10 age groups each.
If I do 385/20 = 19.25 is that correct?

Then administer the questionnaire to 19.25 males between 20 and 25 years, 19.25 females between 20 and 25 years, 19.25 males between 25 and 30 years and so on. it's correct?
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Are you trying to make a representative sample? If so, what is the final purpose using the results? If you try to report stats for these subgroups or make comparisons - you will likely have poor power.

Also, are there 60M people in Italy and are you really randomly selecting from all of the country or just one city, so your sample may not be generalizable to Italy but just the city.
 
#3
Are you trying to make a representative sample? If so, what is the final purpose using the results? If you try to report stats for these subgroups or make comparisons - you will likely have poor power.

Also, are there 60M people in Italy and are you really randomly selecting from all of the country or just one city, so your sample may not be generalizable to Italy but just the city.
Thanks for your answer. I am doing the survey via facebook and from what I see I have uniformly covered the Italian territory. What I would like to discover are habits in free time (cinema, sport, reading, drawing, etc.) divided by sex and age groups
 
Last edited:
#4
It does not work this way. Sampling is done on the basis of the universe, i.e. proportion of population. If the population of Italy is 60 million, then you must first determine the male: female ratio. For instance, if it is 1000: 800, then your sample size will be 210 males and 175 females (out of 385). Similarly for the age groups.
Ratio is the key, not equality.
 
#5
It does not work this way. Sampling is done on the basis of the universe, i.e. proportion of population. If the population of Italy is 60 million, then you must first determine the male: female ratio. For instance, if it is 1000: 800, then your sample size will be 210 males and 175 females (out of 385). Similarly for the age groups.
Ratio is the key, not equality.
A thousand thanks! Looking at this page I came to these results.
https://dati.istat.it/Index.aspx?QueryId=42869

Cattura.PNG

I realized that in some cases I already have more completed surveys than required, while in some cases the number coincides.
For example: Females between 20 and 25 must be 10,27190343 polls, but I have 16.
What should I do? Do I have to remove 6 random surveys or the more surveys there are, the better?
Thank you
 
Last edited:

Karabiner

TS Contributor
#6
Thanks for your answer. I am doing the survey via facebook and from what I see I have uniformly covered the Italian territory. What I would like to discover are habits in free time (cinema, sport, reading, drawing, etc.) divided by sex and age groups
You won't be able to construct a sample which is in any way representative for the Italian population,
for example because Italian Facebook users and non-users certainly are systematically different. So
you will just be able to answer research questions about a subgroup within the group of Italian
Facebook, i.e. Facebook users who by chance notice your survey and who have time and interest
to answer your survey. What exactely is the topic of your research, what are the research question
you intend to answer? It is not yet clear why your convenience sample should resemble the
Italian population with respect to age and gender, in order to answer your research questions.

With kind regards

Karabiner
 
#7
You won't be able to construct a sample which is in any way representative for the Italian population,
for example because Italian Facebook users and non-users certainly are systematically different. So
you will just be able to answer research questions about a subgroup within the group of Italian
Facebook, i.e. Facebook users who by chance notice your survey and who have time and interest
to answer your survey. What exactely is the topic of your research, what are the research question
you intend to answer? It is not yet clear why your convenience sample should resemble the
Italian population with respect to age and gender, in order to answer your research questions.

With kind regards

Karabiner
Thank you very much for your reply.
So do I have to use this data to find the sample?
https://napoleoncat.com/stats/facebook-users-in-italy/2021/10/

So for example, Facebook users are 41780100, the sample always remains 385, so for males between 25 and 34 years (12.2%) there are 46.2 polls.
So would that be right?
The survey is about: how Italians spend their free time.Thanks
 
#8
In research, there are different types of sampling. No matter which sampling type you choose, you must have a strong justification for the sample size and type. In your case, out of a population of several million females, if your sample size is only 16 then i do think it is too less. I advise you to increase it. It should be enough to represent the universe of the population. Moreover, having such a small sample restricts the kind of statistical tests you can apply on the data.
 
#9
In research, there are different types of sampling. No matter which sampling type you choose, you must have a strong justification for the sample size and type. In your case, out of a population of several million females, if your sample size is only 16 then i do think it is too less. I advise you to increase it. It should be enough to represent the universe of the population. Moreover, having such a small sample restricts the kind of statistical tests you can apply on the data.
in accordance with this link, I have increased the ranges of the age groups. so could it be representative sencod you? Thank you
 

Attachments