A widely cited paper from rural India, Robert Jensen and Emily Oster’s “The Power of TV: Cable Television and Women’s Status in India” (Quarterly Journal of Economics, 2009), reports a striking finding. When cable television arrived in a village, women there reported lower son preference, more autonomy in household decisions, and less acceptance of intimate partner violence. Cable was the treatment and the outcomes were scales of gender attitudes. The paper is now twenty years old.
The paper’s finding holds for its time. Cable television did carry that signal, at that moment, in those villages, when watching cable showed that a household had chosen to let the outside world in.
With a co-author, in a small triadic pilot, we wanted to test whether cable still carries that signal or whether it has moved to the phone in the young woman’s hand.
The design
The pilot was triadic. In 115 rural households we interviewed three women separately during the same visit: the grandmother, the mother, and the eldest unmarried daughter. All three answered the same core instrument on gender attitudes, covering son preference, decision autonomy, acceptance of wife-beating, and daily media use.
Interviewing them separately matters. A woman asked about son preference in front of her mother-in-law will give the community’s usual answer, and a daughter answering in front of her mother will adjust what she says. An enumerator matched on caste and language administered the instrument to one woman at a time, in a screened corner where possible.
We were interested in three things: the raw difference in gender attitudes between generations, the media each generation uses in a day, and whether the media a woman uses is associated with the attitudes she holds.
What the numbers said
Daily social media use was eighty percent among daughters and two percent among grandmothers. Mothers were in between at forty-six percent, with a wider spread by household income than either of the other groups.
For the first time in one of these instruments, television did not point in one direction on the attitude scales. It correlated positively with some outcomes and negatively with others, and had no significant effect on most. Cable now works as background and no longer acts as a treatment.
Daughters who used social media daily reported much lower son preference, six points lower on the scale, with a confidence interval that stayed well clear of zero. For mothers, each additional platform was associated with about eight more points of say in household economic decisions, and the relationship was linear across zero to four platforms.
The six-point finding is the one people will quote. A different finding troubled us more.
Nothing moved acceptance of wife-beating, in any generation, on any medium, at any level of daily use. Son preference and decision autonomy responded to the smartphone in the way the paper suggests they once responded to cable. Acceptance of wife-beating did not. Shifting that attitude appears to take more than exposure to a new medium.
What the pilot can and cannot show
The pilot does not show that the smartphone causes lower son preference. Triadic pilots are not designed for causal identification. They generate the priors and the sampling frame for a larger, longer study with a different design.
Cable has also kept some relevance. What has ended is its role as the marker of an outward-looking household. It is now one more source of background information, competing with the phone for the same attention. In a village where a fourteen-year-old girl has a Reels feed personalised to her, whatever cable adds has to work much harder to be new to her.
Why we are writing this up
We have two reasons.
The first is that the twenty-year-old finding has shaped policy design for some time. Programmes built on the assumption that television is the medium of change in gender attitudes will be aiming at the wrong screen if the phone now carries the signal. The intervention depends on getting the medium right.
The second is that the wife-beating result is a puzzle that needs stating. If the medium that shifts gender attitudes has moved, and attitudes to intimate partner violence have not moved with it, then either the two attitudes form in different ways or whatever shifts the second is missing from our instrument. We do not know which, and the paper we are writing for EPW says so plainly.
If you know a village where the pilot’s finding would fail, such as a place where cable still marks a household out or where the phone has not arrived, please write in. Finding contradictions is the main reason to write up a pilot before the full trial.