Can personal cell phone contact list be considered as a sampling frame for a consumer behaviour research?
The research is on innovation resistance and car owners are my samples.
What should be my sampling frame?
Thanks,
Hi Azhar,
What are your concerns? I assume under-coverage will be negligible. Over-coverage will be present but you can just screen out quickly those who don't own a car. Well, I believe you will not be able to poststratify to account for unit non-response, because you will have no information on the population, even the sampling frame, other than simply the number.
One interesting aspect that emerges could be that you could reach not only the 'head of the household', the actual owner of the car, but also household members who had an influence on choosing the car but who are not the legal owners. Maybe two separate questionnaires?
If by personal phone list you mean your own contacts, your sample will most probably be very skewed unless you have a wide coverage of the community you are researching in your personal list.
Survey monkey is an alternative. For the respondents, there is probably very little alternative than to go through the sampling process. One example is to use the subscriber list of the phone companies in the area(s) your are covering.
Thank you for your view.
Its not just my cell phone contacts but many. For instance, I identify 30 contacts from my contact list who would lead me to the respondants. I could ask them to forward my questionnaire link (Google Forms) to every 3rd person in ther contact list who owns a car. This person can sent to 20 people in his contact list who have cars. Here, Identifying the first 30 is convenience, and then the next 20 from each of those 30 would be systematic random sampling. It is possible?
I m working on a topic which would be for a whole population of my state in India. A very large population indeed. Considering the time factor for my PhD, what sampling techniques can I use?
OK, I thought you have a long list of telephone numbers which would be a sampling frame. But now I understand that you want to do some form of snowball sampling. That means that you don't have a sampling frame at all.
There are several issues with this approach
- it is clearly onerous, and I don't know how many of your contacts will be happy to help, or whether they know that their contacts have or don't have a car
- some contacts would be reached more than once
- you will not reach contacts that are outside the extended network of your friends
This is clearly non-probabilistic sampling, and you are risking a huge amount of bias. Some bias is probably not a big issue for a PhD thesis, but it needs to be understood and controlled.
Are there mobile phone directories in your state? This looked relevant to me but I have obviously no idea of telephone directories in India: http://www.findandtrace.com/mobile-phone-number-database
Dear Mr. Mate Peter Vincze
There is only one telephone directory published by the govt office only for landline telephones. There are no such telephone directories by mobile phone service providers. And I would not know who owns a car using a printed telephone directory.
The challenge infront of me is, Can I use snowball sampling for my research where I am intending to use SEM? Can I conduct a parametric test on a non probabilistic sampling technique? I have seen many articles published in known journal where quota sampling and snowball sampling technique is used to collect data and done parametric tests.
I am looking for more support in literature.
Thanks.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics