Results: Internet Access Conditioning Among Newly Recruited "Non-Internet" KnowledgePanel® Members

By Patricia Graham

One of the core elements of the Knowledge Networks (KN) advantage is our nationwide online panel, known as KnowledgePanel®, which incorporates the views and opinions of the entire U.S. population – including those who did not have access to the Internet at home prior to their panel membership.

Since KN is a firm that believes in transparent research on research, we wanted to share our research conclusions about an important topic in survey research, that being: Did we change these new previously non-Internet KnowledgePanel members – and, if so, how?

We are happy to report the answer is "no" based upon a measurement of within-person change on the study items from baseline (Wave 1) to follow-up (Wave2)

Let me lay out the details of this conclusion. The summary of results is as follows:

• We found no concerning evidence of behavior change for non-Internet households who were given PCs or MSN WebTVs in the following aspects:
  • Media usage
  • Online behavior
  • Attitude towards technology
• In the few instances where we measured significant change, the direction of change is showing lower media usage at wave 2 in comparison to the baseline in wave 1
• When we measured an increase of media usage, the magnitude is very small

You can access the results among Latinos given PC's and non-Latinos given WebTVs by clicking here.

As you will see, given the high numbers of significance tests performed (120), we should expect that 5% of them will be statistically significant just by chance (6 tests). That is indeed what we found and no more.

Data Analysis and Statistical Testing

Because of the dependency in the data (same respondents answering the same questions), we could not use traditional statistical testing. For ordinal variables we used

• Sign test (a.k.a. binominal sign test)

For categorical or nominal variables we used

• Pairwise t-test

Results are presented using deviation analysis

• Wave 1 is set as baseline and wave 2 shows percent change

The Research on Research Study Design

Knowledge Networks used a longitudinal study design of recently recruited KnowledgePanel members who reported that they had no Internet access from home at the time of their recruitment and were subsequently provided Internet access by Knowledge Networks. There are two different devices for accessing the Internet given to these "non-Internet" panel members: either a MSN WebTV (device given in the past) or a laptop computer (device currently provided). Both the WebTV group and the laptop group access an ISP through a conventional telephone line and a dial-up modem.

• We executed a repeat measurement of the extent of change in attitudes and behaviors for areas such as technology and media usage, when new "non-Internet" panel members are given Internet access. We also measured differences among Spanish-speaking panel members between Internet and non-Internet groups. (Note: the study questionnaire is also administered in Spanish.) The questionnaire was 15 minutes in length, with data analysis performed on 794 members who answered both wave 1 and wave 2.
  

Longitudinal Design Specifics: Administer the same study questionnaire to each group at baseline and at two 3-month follow-up periods.

Month 0 Baseline Feb 2 - Feb 24 '09

Month 3 1st Follow-up May 1 - Jun 8 '09

Month 6 2nd Follow-up Field began Aug 09

Study Subjects: Panel members recruited within the past 90 days were considered "new" recruits eligible for this study. Only new panel members were sampled for Month 0 and then re-sampled for Months 3 and 6. Note that the Hispanic population was oversampled.

The following table represents the number of panel members available to be assigned a study questionnaire as well as the ending sample.

Available Members and Members Responding to the Panel Conditioning Survey

• Free Web Device Provision
  • Non-Hispanic panel members: MSN TV WebTV and Dial-up ISP, n=62 interviews
  • Hispanic panel members: Laptop PCs & Dial-up ISP, n=150 interviews

Group	Total Size including Spanish speakers	Spanish-speaking Group Size	English-speaking Group Size
Non-Internet	676	458	218
WebTV	176	0	176
Laptop	500	458	42

Survey completion rates were 69.6% for the baseline wave and 74.6% in the follow-up wave

Conclusion

By combining true probability sampling and the Internet's many advantages as a research medium, KnowledgePanel® provides a uniquely accurate resource for commercial, academic and government researchers alike. Our research on research has shown that giving home Internet access to people who did not have it before does not significantly change their behavior.

From time to time we will continue to report our research on research in A:I/R.

Patricia Graham is Chief Marketing Officer and Executive Vice President of Knowledge Networks. Over the past 32 years she has evaluated the marketing plan performance, advertising and communication plan effectiveness of hundreds of new and established brands.

Photo: © Glo5-l-Dreamstime.com