The key words here are "for the most part" and "I would say...".
You are in effect, setting up an hypothesis that the poll accurately measures the sentiment among, to use your characterization in the thread title, "Army" personnel. The problem that you have in doing so is that you don't have enough information to judge the accuracy of your hypothesis. It can't be accepted because you have no idea about the strength (or lack thereof) of the population that responded to the poll with the "Army". Moreover, that strength of association (or lack thereof) is highly likely to impact the respondent's choice of preferences, e.g., an Army wife left at home after her husband's repeat deployment to Iraq might well respond differently than one whose husband has yet to be deployed to Iraq.
Further, no information is furnished as to the demographics of the respondents; demographics in the case of such a poll referring to the number of respondents that are active duty, the number that are retired, the number that are in the 30 - 40 age group, the number that are in the 40 - 50 age group, etc., and the number that are officers, the number that are enlisted, etc, etc, etc.
Without this kind of info, one simply cannot accept this hypothesis. Perhaps this info will subsequently be furnished and a better assessement can be made, though the last time I looked, it wasn't available.
You "would say that your claim is accurate" is indeed an accurate characterization of the strength of your claim: you would say, but you don't know. Acceptance that this poll accurately reflects the attitudes of "Army" personnel requires acceptance of unwarranted assumptions as truth.
Accept it if you wish, but as it stands, more careful observers will properly note it as interesting - but not very.