HARTFORD — If anyone needed any more evidence about being cautious of political polling these days, all they have to do is look at the results of two very different opinion surveys released in the past few days.

On Tuesday, a new-style, online poll on Connecticut's gubernatorial race came out with the somewhat startling result that Republican Tom Foley was leading Democratic incumbent Dannel Malloy by nine percentage points.

One reason it got so much play — despite all the questions about how it was done — was that it was commissioned by the New York Times and CBS.

On Thursday, a slightly more traditional opinion survey was released, this time by a GOP-leaning group called Vox Populi Polling. It found Malloy ahead of Foley by a single percentage point. A third-party candidate, Jonathan Pelto, came in with 3 percent support among those polled. With the margin of error included, the only conclusion from that poll is that the race is too close to call.

So, which one are we supposed to believe? The NYT-CBS sponsored survey that used methods that some called questionable? Or the Vox Populi Polling results that also involved some methods (such as automated "robo-calls" to presumed voters) that traditional pollsters question?

Maybe the answer is to be very wary of both.

The last time a survey on the race was done by the highly respected and very traditional Quinnipiac University Poll was in May, and it also found the race essentially too close to call. The Q Poll does its surveys by telephone, with the interviews conducted by real people rather than computer.

Four years ago, Malloy managed to squeak out one of the narrowest gubernatorial election wins in modern Connecticut history, beating Foley by less than one percentage of the overall vote.

We're still in mid-summer, with Election Day more than three months away. Foley is being challenged by GOP contender in an Aug. 12 Republican primary. Relatively few Connecticut voters are paying much attention to anything political at the moment.

That is why the NYT-CBS poll seems so odd to some political observers. It was done by the non-partisan group YouGov, and was a survey of people who volunteered to take part on the Internet. The initial New York Times report included many caveats, cautions and conditions detailing the questions surrounding the online polling system's accuracy.

Still, just having the New York Times and CBS names associated with the poll gives it (at least in the minds of many people) weight and importance.

The lesson here may be to take all these early polls with a very large dose of salt. By November, things could look rather different.