Editor’s note: Stefan Althoff is marketing research manager at Lufthansa Technik, Hamburg, Germany. This article is a summary of presentations given at the GOR06 General Online Research conference in March 2006 and at the European Research Event in November 2006.

In the summer of 2004 the market research department of Hamburg, Germany-based aviation services firm Lufthansa Technik undertook an online customer survey. At that time, the market researchers had just eight months of experience in constructing online research projects. For whatever reason - poor questionnaire structure, the length of the survey, the season, the sample or the theme - the response rate was relatively low.

After one week in the field, a reminder was sent out - by a trainee named Julia. The response rate spiked to 30 percent. Could it be that, within a male-dominated business like the aviation industry, it is easier to convince potential respondents to take part in a survey if the sender of an invitation or a reminder mail is female?

For some later online surveys the researchers of Lufthansa Technik used female senders to dispatch the e-mail invitations and reminders. No methodical study was run but anecdotally there seemed to be evidence that e-mails from female senders were read earlier and the surveys were started faster. This assumed effect, dubbed the Anita effect, was named for the last female sender before the General Online Research GOR05 conference in Zurich in March 2005. (The conference is a joint effort of the division of Social and Business Psychology at the University of Zürich and the German Society for Online Research.)

A talk on the Anita effect was presented by the author at the GOR05 for the first time. In Zurich I spoke about the Anita effect with Professor Bernad Batinic from the Johannes Keppler University in Linz , Austria. Batinic is one of the pioneers of online research in Europe . He is one of the founders of the company Globalpark, a Germany-based online survey software. Together we agreed to analyze the Anita effect in a more detailed manner.

Three months later, Keppler University student Victoria Greif decided to write her diploma thesis on determinants of response rates of online surveys and Batinic became her tutor. The gender influence of invitation and reminder mails respectively on response rates was one point she focused on.

Greif developed several versions of invitation mails for a online survey with Batinic’s help. She varied among others the sender (male/female/neutral), the salutation (personal/impersonal) and the topic (“Web site test” or just “test”).

At this stage Bodo Griel, a friend of the author’s and project manager at German research firm Webfrager, heard about Greif’s diploma thesis. Webfrager is a joint venture of the Cologne-based market research institute Psychonomics and Foerster & Thelen of Bochum , Germany, a field service. Griel suggested to run Greif’s survey within the Webfrager panel.

Web site test

In autumn 2005, using a test design created by Greif and Batinic, the survey started. Webfrager panelists, 1,800 in total (18 groups of 100 test persons), were invited to a simple Web site test.

The results of the survey were not as clear as expected. The team had assumed that the combination “female sender,” “personal approach” and “clear, defined survey subject” would have had the highest response rates. But that was not the case. In all groups the response rate was around 50 percent, which was good. Even taking into consideration that the panelists were relatively new (no longtime-registered panel members were included), it must be assumed that panelists would behave and react in a different way:

  • It must be assumed that most panelists have a Web-based e-mail account (e.g., yahoo.com). These accounts work without preview windows, therefore only the sender and the topic could have any impact. There is no additional information displayed.
  • In addition, private persons have less e-mail traffic than businesspeople. The invitation mail does not need to win over other mails.
  • Panelists know the e-mail address of their panel company. They do not pay attention to the complete address, only to the last part (behind the @ sign).

Serious hints

Even if the results of the panel survey didn’t support the existence of the Anita effect, the team didn’t want to give up. In the meantime it got some serious hints from other colleagues that the assumed effect really exists.

The author and Griel wanted to start another survey to further investigate the Anita effect. In January 2006 both had the idea to run a test using an employee survey at Lufthansa Technik about intranet usage. We assumed that the topic was moderately interesting, therefore ideal for testing the impact of the Anita effect.

At Lufthansa Technik it is quite usual to use e-mails for internal communication. Most employees use Microsoft Outlook, which has a preview window. The test design was relatively simple:

  • two groups of 105 men;
  • a similar structure for both groups;
  • one group got an invitation mail from a female sender, the other one from a male sender.

This relatively small survey had a response rate of nearly 80 percent. In the end the group with female sender had a higher response rate (83.9 percent versus 74.3 percent) but the result was not statistically significant.

At the beginning of March 2006 - just before the GOR06 - Greif started a survey with registered users of UNIpark, a Globalpark survey offering aimed at the academic market. In total 460 users - split in four groups - were asked to rate the UNIpark home page and the usability of Globalpark online survey software. Greif developed different invitation mails for each group in which she varied the gender of the sender as well.

In this survey a significant result was seen: The response rate was higher from the group that received an invitation mail from a female sender.

More research necessary

The response rates of online surveys depend on many different interrelated determinants. Obviously, more research is still necessary to find out if the Anita effect is real.

Some questions that deserve further study:

  • What effect does the sender’s possible country of origin have? That is, what if the sender appears to come from Eastern Europe or from China?
  • What about the name itself - are there surnames which could be understood everywhere?
  • Has Julia a better chance than Andrea? (Andrea is also a male first name in Italy.)
  • What about the combination of first name and surname?
  • What about a male sender within a female environment?

The team agrees on one point: The cooperation in this project between academia and business was really interesting and informative and could serve as a model for future efforts.