I hate to be the bearer of bad news, nutribelle, but your research methods, vis-a-vis an online survey, are fundamentally meaningless and if you use the results in your dissertation don't be surprised to get royally slammed by your committee.

This type of survey suffers from an extreme and unquantifiable degree of self-selection bias. The only people participating are the tiny fraction of those that have potentially had the opportunity to participate, which is itself a highly-skewed sampling of your target population, and that have gone out of their way to do so. That means that they are very much NOT representative of ANY identifiable population and, in fact, the people that ARE representative are the ones that chose NOT to participate.

Furthermore, the people that are members of this forum are NOT representative of people on low carb diets -- the people here have chosen to join this forum, for a variety of reasons, but the overwhelming majority of people on low carb diets have NOT made such a decision and what separates the two groups almost certainly affects their opinions regarding low carb diets. More to the point, it is very poor research practice to assume that it doesn't and you simply have no basis to even estimate the degree to which it does or does not.

Surveys such as yours represent negative knowledge because they tend to lead you into thinking that you have learned something about your target population when you actually haven't, and that is less knowledge than knowing that you don't know anything about them.

To get the kind of survey results that are worthy of inclusion in a dissertation you need to do a few things.

First, define your target population. You've stated that yours is people that have been on or are currently on a low carb diet. These are two very different populations and are likely to have very different viewpoints. In fact, separating them out and comparing the responses from people that are no longer on a low carb diet to those that are is very possibly a very valid line of research that you might consider. But lumping them all into one group is likely to really muddle your conclusions.

Second, you need to devise a means of approaching a representative sample of your target population. This is not easy to do. The most tested and true approaches are to randomly phone people or to randomly approach people at places that most people in the target population would be expected to go with an even distribution. Shopping malls are a very common location that produce valid samplings for most target populations.

Third, you need to track how many people were approached and what fraction of those that meet the inclusion criteria actually participated. The rule of thumb is that at least 75% need to participate otherwise your data is subject to too-high a likelihood of suffering self-selection bias to be reliable.

If done correctly, you actually don't need that many respondents to get good data. But if done poorly, like you are doing, then there simply is NO amount of data that can produce meaningful results.