Alternatively worded, "What caller ID messages/labels and phone numbers are most likely to be answered?"
This topic comes up from time to time in my field, but I don't recall ever seeing any solid or interesting findings. Full disclosure, I haven't searched AAPORNet, SRMS proceedings or done any kind of search for this yet.
I work on a survey of California, and we've had a few thoughts about what we COULD do. I don't know if these things are technically possible, but we were thinking about what makes us most likely to answer (or not answer) the phone.
1) Use a local area code (i.e., same as the sampled number). Seems to me that if you can set the label/message that a phone displays, you could make that label a phone number itself. It would probably be best (and least sketchy) if the outgoing call (from the call center) was actually routed to show as coming from a local number (and that calling that number connected back to the call center).
2) Custom label: It seems like a bad idea to say "Survey" anywhere on the label. There also isn't a lot of text space to work with. Our survey is known as "CHIS" and is run out of "UCLA", so we're thinking that "CHIS - UCLA" (or something like that might be okay). We all feel like "UCLA" would probably make people think it's fundraising, or would be meaningless to people without a UCLA affiliation. I also wonder if it's a negative force on response for folks in Nor Cal.
3) I think we all agree that "unlisted" or "blocked" or anything that hides the number generically is a bad idea.
My personal feeling is that people aren't likely to pick up the phone when caller ID is unfamiliar or the number isn't known. So being a local activates some familiarity. Any other ideas?
Have you seen research on this (or done any yourself)? I'll share back what I find elsewhere (here or in a blog post).
Thanks in advance!
I would prefer the short name of my institution. Will you too?
Matt, my unscientific n = 1 (me) judgment is that displaying a local number is the way to go, as it doesn't carry any negative baggage - e.g., an unrecognized, long-distance number is likely a telemarketing type call. Anything that you use to convey actual information, like "UCLA", also tends to carry unpredictable connotations. At the least, a local number MAY be viewed as potentially relevant to a hh member. I don't think there's any free lunch here - the potential respondent may react negatively when he/she expects to know the caller and discover that it's instead a SURVEY (yuck!). But if you can get past the first step, and get them to pick up the phone, that's a start. Tough business, isn't it?
The short answer is 'it depends'. There's a great article which tackled this topic fairly recently:
Callegaro, Mario, Allan L. McCutcheon, and Jack Ludwig. 2010. “Who’s Calling? The Impact of Caller ID on Telephone Survey Response.” Field Methods 22(2):175–91.
The entire discussion is rooted in this question: Do you want people to know that the call is coming from UCLA?
If it might significantly bias responses, the answer should always be 'no'.
If it might positively bias response rates (elicit more answers), the answer is likely 'yes'.
If the general affect of the local residents toward UCLA is positive and the topic of the survey doesn't directly correlate to local issues/sentiments which relate to people's thoughts of UCLA, I'd include a direct identifier in the caller ID label.
The exact wording of the label might take a bit of experimentation. Often including the words 'poll' or 'survey' helps, but sometimes it doesn't.
We run a community research center here at Baylor University and because of the positive local sentiment, our call center is usually set up to display 'Baylor University' as the caller ID. Still, some phone carriers will display the number as 'Restricted/Unknown' depending on their infrastructure and policies, so it's still hit-or-miss.
Thanks for all the comments.
@Jeffery, thanks for the reference. I know those authors and that journal. I guess I just haven't been reviewing it frequently enough. They find "...initial evidence for the hypothesis that the caller ID transmission works as a sort of “condensed survey research organization business card” that can trigger brand awareness, thus legitimating the survey and diminishing suspicions of collector or telemarketing calls." It looks like final response rates were lower when "Gallup" or "Gallup Poll" were mentioned in the Inpatient, Outpatient, and Discount surveys, but that those labels may have helped for the Banks survey and RDD survey. Since we're an RDD survey, that's good news, and maybe we shouldn't be so worried about saying "Survey". In fact, for the RDD survey they used, just saying "Gallup" didn't make a difference, but saying "Gallup Poll" did (2.5-3 %-pt increase in RR). See table 2.
@Gordon, thanks for the thoughts. We're leaning toward local numbers if we can do that. I'm not sure about the logistics.
@Bengt, has Stats Norway ever evaluated this?
For some years ago we changed from lockal numbers to our company number for both decentralized cati/capi and cati. We changed back after a few weeks because of significant decreased contact rates.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS