Lazarsfeld et al 1944: The People's Choice

in a nutshell
Ground-breaking study for survey and voting research (e.g. first panel study ever: 7 monthly waves until presidential election of 1940 (Roosevelt vs Willkie), 600 participants, all living in one US county)

With basic descriptive statistics, they show that media messages matter little, because they are primarily consumed by the politically interested who have already formed their opinion and select their media content in line with that opinion. Apart from that, the campaign related through the media seems to merely activate or reinforce voters' already existing predispositions. Thus, much variation in voting behavior can be explained by socio-structural variables (socio-economic status, religion, rural/urban). (but caution: socio-structural cleavages were much stronger in 1940 than today)

In passing, they come up with the theory of the "two-step-flow of communication" (their evidence is weak, but the theory becomes very influential). It contends that some of the politically interested media consumers (opinion leaders) spread their views to the uninterested (who do not consume political media) by personal contact.

Summary
.

social structure as powerful predictor
.
 * socio-structural index consisting of:
 * SES (home, appearance, manners - rated by interviewer!, intracoder reliability 0.8 - not good!, intercoder: 0.7) (1 to 4 points)
 * Catholic vs Protestant (0 vs 2 points)
 * urban vs rural (0 vs 1 point)
 * measures the "predisposition" to vote for one of the parties: If person has high socio-economic status, is Protestant, and lives in rural environment -> prone to vote Republican
 * explains a lot of variance: 74% of those with most republican-prone index score actually vote Republican, but only 17% of those with least Republican-prone index score
 * (also discussed in passing: occupation: whether a person self-identifies as white/blue collar)
 * cross pressures -> make for late deciders (decide upon whom to vote for late in the election campaign)
 * "Conversion": vote intention more often changes towards alignment with "predisposition" score (then away from it), but unclear if this is because of media messages or personal contact. (?)

The less highly exposed to media:
.
 * are less interested
 * decide late whom to vote for
 * associated with: less partisan, poor, rural, young, women
 * stability: the less highly exposed stay less exposed (almost 80%) (but of the more highly exposed, almost 30% are less exposed in the second half of the campaign, so more people lose than gain interest). so there is some stability (but b/c it's so highly aggregated (only 2 time periods), no evidence for a strong effect / stability)
 * little exposure to one medium correlates with little exposure to another medium (newspapers, magazines, radio) (correlations cannot be high, but are not given)

media effects:
.
 * all three media genres (newspapers, radio, magazines) were similarly biased towards Willkie (in this county)
 * but Democrats still favored the radio (explanation: Roosevelt's 'superb radio voice', possibly rather because Democrats are less educated, working class),
 * and towards-Democrat changers mentioned the radio more often as main source of opinion change (doubtful that they really know and honestly report why they changed their opinion)
 * in spite of bias: Willkie gained only 4% from June until elections
 * Several reasons (Kinder 2003, 1998) for weak media effects:​
 * Neutralization of campaign messages​
 * Resistance: Strength of predispositions​
 * Indifference: lack of attention

So who decides the election outcome?
.
 * not the doubters (open to rational argument, deciding late) because they are few
 * rather: the other late deciders, who are poorly informed, have little political interest, and say more often they were persuaded by a personal contact (as opposed to media).
 * but: to assume that the vote of other people is fixed, is only valid when assuming a campaign of at least roughly equal strength by both (all) parties. and even then: the agency of the influencers, not the late deciders themselves, may be crucial.

In passing they come up with the "Two-step-flow" theory of communication:​
 * From the media to the opinion leaders​
 * From the opinion leaders to the less active sections of the population