t Survey Theory

By pjain      Published June 10, 2020, 10:24 p.m. in blog Lambda7   

Statistical Analysis of Survey Data

Survey panel

Consumers who respond to a survey are called the survey panel. In statistical terms, the larger and more targeted the survey panel, the more accurate the results are going to be. If you know that your target customer demographic is women between the ages of 25 and 50, you want to survey as many women in that demographic as possible and rule out everybody else. Only then will you get an accurate assessment of what women in that age group generally think about your product.

Means are important

There are many ways to find survey panels. Phone calls are a popular method ("Is there a man in the house between the ages of 18 and 25?"), as is something called the "mall intercept," where people walking through a mall are stopped and asked if they would participate in a survey [source: American Marketing Association]. Postal mail is another way to distribute surveys.

E-mail has a higher incidence rate (%responders) than other survey methods unlike phone surveys. E-mails are relatively painless to fill out and send back, unlike snail mail (need to lick a stamp), or in-person or phone surveys (intrusive).

Website hooks. An advantage of e-mail surveys is that a company can use its own Web site to generate contacts. Most websites have options for customers to sign up to receive e-mail newsletters and other electronic notifications. A company can use that e-mail contact list to send surveys to targeted existing customers.

The problem of "in-company-website" surveys is they are not representative of the far larger populations that never heard of your company. In essence you are "preaching to the choir" when using that segment.

Panel Fraud Fighting

A major disadvantage of online surveys is survey fraud.

The reliability of survey statistics depends on something called panel integrity. There are several security measures that a company can use to decrease survey cheaters and maintain high panel integrity:

  1. Invitation-only. A consumer must receive an e-mail invitation to participate in a survey. This gives the company a chance to verify the address and assure that the same name and contact information isn't already assigned to another e-mail address in the system.

  2. Require mailing address. Ask for a mailing address to send reward points and coupons. Once again, if the same mailing address comes up for multiple e-mail accounts, you might have a cheater.

  3. Minimum time frame. Set a minimum time for completing an online survey. This cuts down on cheaters who fly through the survey just randomly answering questions.

  4. Block "straight line" answers. Have the software automatically purge any surveys that come back with the same letter or number chosen for every question.

  5. Block IP addresses. If anyone is caught cheating, block his or her IP address from submitting any future surveys. It's also wise to limit the amount of surveys that can be submitted from any one IP address to cut down on multiple submissions.

  6. Demographic consistency. Start the survey with straight, demographic questions. The next couple of questions should be demographic-profiling questions that verify a person's sex, age, income level, etc. If the answers to the demographic profiling questions don't jibe with the earlier answers, the user should be blocked.

  7. Open-ended questions. Include at least one short-answer, essay type question that can be analyzed for thoughtfulness. Cheaters won't take the time to answer an open-ended question multiple times with significantly different responses

Survey Question that segment and validate panel

If you need say n=1000, you may need to throw out 2-3x the size to avoid misrepresentative samples. So you should add questions in your survey that segment and validate panel.

Population Analytics and Validation

NLP, Psychologically Influencing via taking Surveys

Subtle even Sublimal Influence

For example, by asking questions about the positive attributes of a product ("How important is Coke's great taste?") you can plant a positive association in the consumer's mind.

Remind Existing Customers - REPETITION

By polling existing customers, you can remind them of a product that they already like, but might have forgotten about - until now, when they were filling in the survey.

Rewards - Coupons, Offers made in excuse of "survey"


There are no comments yet

Add new comment

Similar posts

mr Market Review Sites, Apps

ab Survey Requirements PRD, MRD

uxd Survey Gallery

m Survey Market Review, Trends