So there was an article in Ars Technica the other day saying that people who were exposed to banner ads for a particular product had a slightly more positive attitude towards it (the product, not banner ads).
I wont go into detail about how the study was performed (if you want to get a better idea, read the article) but what I will say is that I am always a little sceptical of these types of studies. Why? Because after reading (well listening to the audio book of) Malcolm Gladwell's "Blink" I have come to realise that what people SAY and what people actually THINK are often 2 very different things. Having done one or two market research surveys in my time, I know that when asked specific questions about "how you feel" about a brand the true answer is quite often "I don't FEEL anything about it" yet feeling obliged to answer the question I choose a random number somewhere between 1 (meaning poor) and 10 (meaning excellent).
In my opinion the proof is in the pudding and the statistics that can be gathered from said pudding. Statistical research of how your advertising campaign is extremely important but it tends to be something that only gives you true data after the fact. Asking people how they "feel" about a brand might not result in accurate data, looking at how many people bought your brand after a certain campaign will usually tell you everything you need to know