Showing posts with label survey. Show all posts
Showing posts with label survey. Show all posts

Thursday, 1 November 2012

Food Shopping Survey Results


The results are back from the survey that I carried out last week (after analysis on Stata).

(For a more technical critique please see here)

Headline result:

  • The more often people shop for food, the more likely they are to think that they buy items of food out of habit. I think the most likely explanation for this is that people who shop more often have more opportunity to create habits - the reinforcement mechanism is stronger if you buy food twice a week compared to fortnightly. However, it was only significant at the 10% level and so it is not conclusive evidence.

Other notable results:

  • There is no connection between whether people consider themselves a 'creature of habit' and whether they think they buy items of food out of habit. I find this surprising. I suggest that this might be because creature of habit could be seen as a positive character trait, while buying food our of habit could be seen as lazy. Thus the result could be because people are trying to present themselves in a positive light (even if only to themselves).
  • There is no connection between the extent to which people think they consider the price of food and whether they think they buy items of food out of habit. I find this surprising. I expected a negative relationship. At the very least I expected that asking people about whether they looked at prices might prompt them into saying they bought less out of habit, but no significant effect was found.
  • There is no systematic difference between male and female respondents.

In conclusion, none of my hypothesised variables affected whether people think they buy items of food out of habit (apart from how often they go shopping). Obviously, I find this surprising.

Food Shooping Survey Critique

Let me start this critique with a disclaimer: If I read these survey results in an academic journal I would not treat them as highly robust. I am ok with this. It was not intended to be a highly robust investigation into consumer perceptions, rather I undertook the survey to get a flavour for consumer thinking and research. I have learnt a lot, both about consumers and about how to go about consumer research.

Reasons why my survey is not highly robust:
  • There were no incentives for accurate answers.
  • The sample size was small (72)
  • The sample was found through facebook and GuruHogg and as such is not representative of the whole population (although it was never intended to be).
  • Internet surveys are only ever going to sample those who use the internet.
  • 10% significance is not enough. 5% significance is required. The result that increasing how often you go shopping by one shop a week increases perceptions of purchasing habit by 17% is not highly robust.
What I would do different for more robust results:
  • Offer incentives.
  • Recruit a larger sample.
  • Recruit the sample from a more diverse section of society.
  • Ask better worded questions: fewer dummy variables.
  • Use an experiment rather than a survey!

Regarding my opening foray into consumer research:

SurveyMonkey is a great piece of kit, but has it's flaws too. Notably, if you only use their free service then it is time consuming to input the results into Excel. Also, you can only ask 10 questions using their free service which is highly restrictive. Stata, however, is a highly powerful and useful statistics software that was able to do everything I asked of it and more.

Wednesday, 24 October 2012

Survey

I have decided to collect some consumer data...

Please take the survey! It is only 10 questions and you will be contributing to behavioural economics:

http://www.surveymonkey.com/s/7HT99VR

I hope to post the results up here soon...