Hi, Deep,

I recommend that you not worry about your sampling methodology. Convenience sampling is completely acceptable in an area without much prior work. I am not personally familiar with Sikkim tourism, but I can easily imagine that this is a fairly new field of inquiry.

The prerequisite that a sample of human respondents is representative of a population for statistical analyses is always - and I do mean always - violated in reality. There is simply no way to ever know that a sample is truly representative because no researcher knows anything about those not included in the sample. Even when a random sample is selected, non-respondents invalidate the logic of random sampling.

Further, I would suggest that a sample of 1,200 is sufficiently large that any potential for sampling bias is most likely muted. If you have access to any population demographics, a descriptive comparison of your sample demographics with what is known of the population may further underscore the ways in which your sample is "typical" of the population.

Define your research hypothesis and run the appropriate statistical tests. Let other researchers learn from your exploration of this area and potentially improve upon it in later studies with more rigorous sampling methodologies. This is how we make progress as a research community.

Hi, Deep,

I recommend that you not worry about your sampling methodology. Convenience sampling is completely acceptable in an area without much prior work. I am not personally familiar with Sikkim tourism, but I can easily imagine that this is a fairly new field of inquiry.

The prerequisite that a sample of human respondents is representative of a population for statistical analyses is always - and I do mean always - violated in reality. There is simply no way to ever know that a sample is truly representative because no researcher knows anything about those not included in the sample. Even when a random sample is selected, non-respondents invalidate the logic of random sampling.

Further, I would suggest that a sample of 1,200 is sufficiently large that any potential for sampling bias is most likely muted. If you have access to any population demographics, a descriptive comparison of your sample demographics with what is known of the population may further underscore the ways in which your sample is "typical" of the population.

Define your research hypothesis and run the appropriate statistical tests. Let other researchers learn from your exploration of this area and potentially improve upon it in later studies with more rigorous sampling methodologies. This is how we make progress as a research community.

Statistical analysis depends on your research questions rather than the type of design. 

if you post your questions, you will get specific help on the suitable analysis

Your sample size seems to make it representative although you did it through convenience sampling. It will not effect in your statistical analysis.

Peter and Hom, are you two saying that having a large sample size makes "sampling BIAS" (my CAPS) less important and makes the achieved sample more likely to be representative? Sample size reduces the variance of test statistics like the mean so a big sample makes this statistics more reliable (i.e., smaller standard error), but BIAS? If the questioner only sampled in certain months or sampled at one exit point in one week and another in another week, does sample size really address the problems.

Many of the statistics make assumptions about the sampling procedure (not representativeness of the population, since how would you know that), and often the sampling (or allocation) used in research is not what the statistics assume. Sometimes deviations are okay; sometimes not. But, it is important to know more about the sampling procedure to know how this may affect the conclusions. If by convenience it turns out only people at exit point A took the survey in July and at exit point B in January, then the sampling procedure has this bias and regardless of statistics if you compare those at A and B it is confounded by season (and there presumably are week effects due to holidays). 

In short, I would need more information to address this question. I do not believe having a large sample solves this problem. But, it sounds like other responders think different and are not requesting more information, so I might be wrong.  

It is worth adding that increasing sample size can decrease reliability.  Reading my previous response I realized a reader could assume I was saying adding subjects always increased reliability, which is not true. I have seen student projects where they are told to get say 100 subjects and collect 98 students and in desperation to reach 100 choose two people who will likely be outliers which will increase the se of the mean. Big outliers,  even in the hypothesized direction can increase p values.  

Mr Daniel, it is difficult some times to generalize that data collected by using scientific procedure is representative although it follow all the tools of representatives. On the other hand, data collected by convenience sampling which is considered poor tool of data collection,may collect more representative data. Depending on the situation we can use what types survey technique/methodology can be used. I agree with your previous comments.

Mr. Muayyad Ahmed has rightly pointed out that unless we know the research question and objective of the study, it may be difficult to suggest which statistical test is more appropriate in the given situation. As far as convenience sampling is concerned, if it has been done properly means at least given due regard to both the genders, various age groups etc,  it may not affect  your findings  largely

Dear all I need references for using convinience sampling in quantitative research

Don´t worry about the statistical technique. The statistical procedure depend of the purpose of the research.

I agree with Mr. Jorge statistical techniques depend on the purpose of the research. However, it may not be appropriate to use parametric tests based on probability since your sampling technique is not probabilistic