Why was the referendum lost?

More analysis on the weekend's referendum. A number of hypotheses:

1. Voters judged that the referendum would worsen the position of economically disadvantaged regions relative to economically advantaged regions; we should expect the referendum to be lost most heavily in the most economically disadvantaged regions
2. Voters judged that the referendum would worsen the position of the South and Islands relative to the Centre and the North, and would worsen the position of the Centre relative to the North; we should expect the referendum to be lost most heavily in the South and Islands, less so in the Centre, and least of all in the North
3. Voters followed party cues (1); voters who supported parties of the centre-right should vote for the proposal
4. Voters followed party cues (2); voters who supported Forza Italia and the Lega Nord should vote for the proposal

The following table presents the regions in order of their geographic affiliation (North/Centre/South&Islands), support for the referendum proposal, GDP per capita in 2003 euros, and total centre-right support in the 2006 elections to the Senate. Lower numbers indicate more support, more money, and more centre-right support.

	Referendum	Economic	Centre-right
Piemonte	3	5	7
Valle d'Aosta	5	2	-
Lombardia	2	1	4
Veneto	1	6	3
Liguria	4	7	12
Emilia Romagna	8	3	16
Toscana	12	8	18
Umbria	10	10	14
Marche	7	9	9
Lazio	6	4	8
Abruzzi	9	11	5
Molise	13	12	10
Campania	16	16	15
Puglia	15	17	6
Basilicata	17	14	17
Calabria	18	18	11
Sicilia	11	15	2
Sardegna	14	13	13
(Valle d'Aosta)	(-)	(-)	1

The use of ordinals is a quick-and-easy compensation - we can try and forgive a little fuzziness caused by a big difference in ordinal rankings caused by minimal cardinal rankings.

Hypothesis 1: is broadly confirmed; the difference between ordinal referendum support and ordinal economic standing is less than or equal to two places with the exception of Sicilia, the Veneto, Basilicata, Toscana, and Emilia Romagna. The first two were more likely to support the referendum proposal than economic position would suggest; the remaining three less likely to support the referendum proposal than economic position would suggest. Sicilia and the Veneto are regions in which Forza Italia has considerable strength; Toscana and Emilia Romagna are regions in which the left has considerable strength. Basilicata has been won by the centre-left in the last two regional elections, but is not noted for centre-left strength.

Hypothesis 2: Hypothesis 2 is similar to Hypothesis 1 (since regions in the South tend to be poorer), but is less fine-grained and is subject to similar counter-examples, particularly Sicilia and Emilia Romagna.

Hypothesis 3: Hypothesis 3 tends to fail because the differences in centre-left/centre-right support tend to be much les strong (and so consequent to much greater ordinal confusion) than differences in referendum support.

Hypothesis 4: I haven't yet the disaggregated data to test Hypothesis 4 - but it would have to improve on Hypothesis 3 to be of much interest.

So, perhaps the conclusion should be this: the relative economic position of the regions was a relatively good predictor of their relative support for the referendum proposal, with more economically advanted regions supporting the proposal, except in party strongholds of the White and Red Belt, and in Sicilia.