landline, schmandline.

September 17, 2004 7:24 AM

Hold the phone. You probably already know that many Americans are ditching their land lines in favor of cell phones:

It is part of a generational shift to wireless, says Leap's chief executive, Harvey White. "Our demographic is younger, and when people start a household today they simply never bother to get a land line."

But were you aware that pollsters don't call cell phones? Media saturation, changes in communications technology, and missed demographics (americans abroad, for example) seem to make polls increasingly irrelevant. (first link via e-v.com)
posted by whatnot (21 comments total)

My brother did that recently. He said only salesmen called his home phone, so he dumped it.

On the other point, a guy I used to know back in the States who worked for one of big polling companies that proivided political polling data to the TV news programs (this was about 1985) told me that it was commonplace to make up answers in order to pad out the poll data.
posted by planetkyoto at 7:53 AM on September 17, 2004

One of the studies I'm involved in is a neighborhood prevalence study of epilepsy, done by cold-calling. My prof thinks it's a pretty good idea because he had some success with a similar protocol in Harlem in the early '80s.

I've tried to tell him that I think differently - for the reasons noted above - but he doesn't really hear it. I think there's really no reason to hear it about a grant that's already been funded :/
posted by ikkyu2 at 7:58 AM on September 17, 2004

I see people mentioning this within the context of the upcoming election as a reason not to lose heart. I'm not so much concerned with the polls themselves (which I've never put much stock in regardless of the methodology) but I have to wonder: Even if the results are skewed, couldn't it easily become a self-fulfilling prophesy? Hypothetically: If you were John Kerry, would you rather be genuinely ahead in the polls by 5 points, or would you rather it widely reported that you were 5 points ahead when you actually weren't? We have a fickle contituency in this country who vote less on their beliefs than on who they think will win.
posted by RavinDave at 8:16 AM on September 17, 2004

Don't forget that land lines are designed to remain operating in the case of electrical blackout. Furthermore, even if you have a landline, if you have a corless landline phone, it's useless in the event of a blackout. Always keep an old school wired landline phone in a closet somewhere just in case!

Not sure if the "still up during blackout" thing is true with cell towers... anyone know?
posted by afx114 at 8:25 AM on September 17, 2004

More importantly, which way does this cut, politically? Is no longer having a "land line" more democratic or liberal? My first impression is "neither," but let me think about it...
posted by ParisParamus at 8:36 AM on September 17, 2004

Polls are only relevant insofar as they affect opinion, which they do (see RavinDave's example above, which feels like the pollsters' dirty little secet). Part of the job of a polling organization is to stay on top of these advances in technology and how they're used. If they're not doing that, then they risk irrelevance.

And I haven't had a land line in about 5 years. Between the cel and a cable modem, it's about the same cost, with vastly less hassle. (afx114 has a point, but blackouts don't happen often enough to justify the extra price.)
posted by chicobangs at 8:39 AM on September 17, 2004

Vonage numbers aren't published either. I've heard the blackout argument against it before, but it doesn't seem to matter much. I still have a cell, and a 1000VA batter backup for my server, cable modem, router, and phone, and that takes me for at least a half an hour or so.

The last blackout that I had didn't result in me having a desire to make a phone call, anyhow. Except one to the 'lectric co. to say wtf?, and that didn't take long.
posted by adampsyche at 8:44 AM on September 17, 2004

Sure you may not need to make a phone call durring a regular old blackout, but what about a blackout caused by a big fucking hurricane? Or a big fucking terrorist attack?
posted by afx114 at 9:03 AM on September 17, 2004

The only reason landlines still work in a blackout is because there's low voltage power coming in over the phone lines. That power has to come from somewhere -- either a phone switching station that still has power or one with a backup generator.

In a big fucking hurricane or a big fucking terrorist attack, either the lines are going to get knocked down or the phone switching station is going to get destroyed, in which case you're still screwed making a phone call.

But who ya gonna call anyway?
posted by crawl at 9:10 AM on September 17, 2004

More importantly, which way does this cut, politically? Is no longer having a "land line" more democratic or liberal?

I think the only thing anyone can say for sure is that those who choose to forego land lines skew younger. Which doesn't necessarily correlate with "democratic or liberal."
posted by whatnot at 9:14 AM on September 17, 2004

My cell phone worked fine during last year's blackout so the cell towers probably had backup generators.
posted by gyc at 9:23 AM on September 17, 2004

I have a land line only because I use DSL for high-speed Internet bandwidth. There's a phone plugged in, but if anyone's tried to call, I wouldn't know, because I keep the ringer turned off.
posted by alumshubby at 9:25 AM on September 17, 2004

Oh, and did I mention I'm over 40, in case anybody cares?
posted by alumshubby at 9:26 AM on September 17, 2004

yeah, that crap about demographics overlooks those of us who hated the Bellocracy for decades and severed the landline as soon as practical. not that the Cellucrats are any fucking better.
posted by quonsar at 10:23 AM on September 17, 2004

Pollsters have been dealing with differing reachability of demographic groups by norming for decades (e.g. over 65 = highly reachable, under 30 = hard to reach). Declining wireline penetration in certain demographics can be easily handled by further norming unless within a certain demographic, those with wirelines are not representative of those who are wireline only.

A lot of pollsters think that the wireless-only issue is not so much of a problem for political polls because they assume that wireless only tracks a variety of social factors also correlating with low voter participation (renter, student, etc.) It is a huge problem with consumer and other non-political surveys, because students, renters, minorities, etc., are very important participants in the market, if not electorate.

In terms of the election, I also wonder whether wireless-only in swing states resemble wireless-only in the nationwide sample. New York hipsters <> Jacksonville African Americans <> Ohio State fratboys.
posted by MattD at 10:29 AM on September 17, 2004

In the 80's I participated in a bunch of academic survey research and at the time the gold standard for good data was the "Random Digit Dial" method which in our protocol further randomized things by asking to speak to the person with the "most recent birthday" (so that you didn't introduce some other bias because of who would pick up the phone).

At the time, we'd be thrilled with a response rate of, say, 60%. I think that the rise of telemarketing, caller ID, and mobile phones has taken a big bite out of the effectiveness of this approach. Is there a political skew? Dunno. I'd guess there's probably a socio-economic skew (which corellates with political affiliation) since better off people are probably more likely to screen calls, be on more telemarketing lists, etc. Of course they are probably more likley to also have a landline and mobile phone so maybe they'd be more reachable? I dunno.

In terms of the validity of survey research, much of it comes down to (a) the sample size and (b) whether the sample maps to the demographics of the population you're trying to make an assessment of . . . if you have a big enough sample and it "looks like" the population, then your data should be fairly solid [what MattD refers to above as "norming"]. A very common mistake is to have a sample size that's adequate for making statemeents about the general population but too small to make statements about a subpopulation like "white women over 40." OK, a bit of a digression from the phone discussion.

Some academic resources if you want to know more at the Inter-University Consortium for Political and Social Research and from the APSA's Society for Political Methodology resource page
posted by donovan at 10:37 AM on September 17, 2004

Anyone else reminded of the Literary Digest poll? I am. I doubt the situation is quite as extreme, but it would be interesting if it turned out that we were seeing a variation on this today.

RavinDave makes an astute point, although I think he's asking it the wrong way: If you were John Kerry, would you rather be 5 points up and have that accurately reported, or would you rather be 5 points up but have it reported that you were trailing? The landline/cellphone divide probably results in poorer polling for Dems, although the self-fulfilling prophecy aspect probably does come into play no matter what.
posted by adamrice at 10:39 AM on September 17, 2004

The other thing is not just cell lines replacing landlines, but do pollsters pay attention to young people living at home?

Granted, I know young people don't turn out in huge numbers, but this year (even moreso than previously) there is a lot going on trying to mobilize young voters. If either of my parents are polled on our home number, they will tell the pollster they're voting Republican -- my brothers and I, however (who are still at home), will be voting Democrat. Is this weighed somehow?
posted by rafter at 10:43 AM on September 17, 2004

But who ya gonna call anyway?

Exactly.
posted by adampsyche at 11:36 AM on September 17, 2004

In the event of a blackout (hurricane/tornado/other natural disaster) a cell phone is by far your best bet for having phone service as quickly as possible. Cell towers all have battery backup which lasts for several hours (how long depending on company and how important that hub is), generators are brought out if the power isn't back on by the time they run out, and if the towers are damaged, COWS (cells on wheels) are set up post-haste ... and all of this is done long before regular phone lines (or electrical lines) can be repaired, usually within the first 6-12 hours of a disaster.

About the cell phone/survey topic: I hadn't actually noticed it, but yeah ... I don't get polling or survey calls anymore now that I only have a cell phone. I suppose it could make certain demographics not show in polls. I hadn't thought about that before. The older generation of my family all have landlines. My generation has a mixture of both (if not actually both at the same time), and the younger ones only have cell phones. It would seem to me that polling only people with landlines would skew the results toward the older set.
posted by Orb at 1:00 PM on September 17, 2004

NPR's "Day to Day" discusses the topic of political phone polling on today's program (RealAudio, 2min into the broadcast)
posted by donovan at 1:44 PM on September 17, 2004