Last week, I leveraged my not-inconsiderate internet resources and begged anyone who had any sympathy for me to take a survey for my consumer behaviour class. I promised everyone I’d write up the results after everything was done, and after having spent my entire weekend (I was stuck in a meeting room with my group for a total of 12 hours) doing data clean-up and analysis, I figured I might as well make a blog post out of it.

The Set-Up

“Priming” involves subtly providing someone with cues in order to influence their later attitude or behaviour. Malcolm Gladwell’s Blink catapulted this concept into the popular consciousness by describing a series of experiments by psychologist John Bargh. He primed different groups of people using sets of scrambled words, some of which had more words that connotated politeness, and others that were more rude. Afterwards, he found that the participants actually were more rude or polite, depending on what they were primed with.

We wanted to try to replicate the effect, but with consumer behaviour. We sat down and came up with a whole bunch of attributes that related to a digital camera – things like optical zoom, pre-set shooting modes, LCD screen size, viewfinder, durability, design, etc. We then divided these attributes into a “technical” set, involving attributes that affect the quality of the picture being taken, an “auxiliary” set, involving attributes that deal more with aesthetics and personal preference, and a “neutral” set, involving features that almost all cameras on the market carry, like a manufacturer’s warranty or an instruction manual.

The Experiment

People who took our survey were randomly divided by the survey software into one of three groups. If you were slotted into the “technical” group, you were given a list of 15 technical attributes and 3 neutral attributes asked to evaluate how important each attribute was to your decision-making process when making a purchase. If you were in the “auxiliary” group, you were given 15 auxiliary attributes and 3 neutral attributes, and the control group got a list of 6 of each type of attribute.

After this, you were shown three different cameras, one coded to each type of attribute, and you were asked to rank them in order of preference.

Canadian readers will recognize this as being a screencap from Future Shop, Canada’s answer to Best Buy. We even made up a brand for it called “Sharpshoot” to prevent any negative or positive associations with existing brand. Gotta love my professionally embossed logo on the camera, eh?

After you picked your camera, you were asked why you picked the camera you did, as well as what you thought the experiment was about, just so we could discard any responses where people figured out what we were up to. (There was another question that asked you to compare the camera you picked against a very similar camera from Sony, to kind of throw people off our trail.)

The Results

The initial data was kind of disappointing. I created a weighted summary score of everybody’s preferences in the different test groups, and there was basically no effect of priming. The technical camera was universally preferred, which isn’t really surprising since technical attributes are (IMO) inherently more attractive. The differences we found in the other two groups weren’t statistically significant enough to draw any sort of conclusion.

There was a slight interaction effect when you took out the ranking, and just looked at whether or not someone in a primed group picked the camera they were primed for.

Unfortunately, some more detailed statistical analysis in SPSS demonstrated that this is basically attributable to chance. Boo.

Thankfully, we weren’t being evaluated on whether or not our experiment succeeded, so it wasn’t a big loss to me. The actual designs of the camera had an unfortunate effect as well. Even though I just grabbed any random camera I found online and modified them, the specific ones I picked ended up influencing consumer decisions more than we wanted. The technical camera, for example, was bright red, which had a very polarizing effect; people either loved it or hated it.

But I was bored, so I decided to take another look at the post-purchase rationalizations, and that’s where the interesting stuff really came into play.

(click for a bigger version)

Even though people in all three groups saw the same three cameras and made more or less the same decisions, they evidently had very different reasons for making those decisions…. or at least, they thought they did. Zoom, for example, is mentioned the most in the technical group but almost not at all in the auxiliary group, and the frequency shoots up again in the neutral group because it had evidently been one of the technical attributes they saw in the neutral test.

This effect is even more pronounced when you aggregate the data.

People in the tech group mentioned technical attributes in their post-purchase rationalization, people in the auxiliary group mentioned auxiliary attributes, and people in the neutral group were pretty evenly split across the board.

Pretty exciting stuff. Basically, even though we couldn’t influence consumers’ decisions, we managed to trick them into thinking they made their decision for different reasons. Of course, there are all sorts of limitations with this study – the fact that tech vs. auxiliary is inherently biased, for example–so I’m not really sure what the implications of this is, but I had fun playing around with the numbers anyway.

Thanks to everyone who helped me out with this! You seriously rock. Our group had 143 respondents from across the world, which totally blew the other groups’ response rates out of the water.