Problem Set 1
Due January 27, 2004
Measures of association for tables in Canadian Ethnic Studies article
In “Minorities and elections in Canada's fourth party system: macro and micro constraints and opportunities,” Canadian Ethnic Studies, Spring 2002, volume 34, pp. 85-111, the authors Livianna S. Tossutti and Tom Pierre Najem (TN) provide several tables that include measures of association such as V and phi.
You can obtain the full article online through the University Library – I have included only a few selections here. See http://infotrac.galegroup.com/itw/infomark/99/965/43689362w6/purl=rc1_ITOF_0_A94550900&dyn=5!xrn_5_0_A94550900?sw_aep=ureginalib
In the questions that follow, you are to reproduce some of the statistics and reanalyze some of the data.
In this article, ethnic is defined as not being from French, English, or aboriginal ancestry. Visible minority is defined as in the federal government definition.
1. Table 2 – distribution of ethnic and non-ethnic candidates by party.
a. For Table 2, the data for the year 2000 can be presented as follows:
For this table, TN report a Cramer’s V of 0.11, with a probability of less than or equal to 0.01. Obtain the chi-square value, its probability, Cramer’s V, phi, and the contingency coefficient for this table. State the null and research hypothesis and your conclusions.
b. Eliminate the Bloc Québecois from the above table and redo the exercise in a.
c. Using the results of a. and b., describe the results in words, including comments on the TN statements about the results in Table 2.
2. Table 7 – distribution of candidates by whether they faced ethnic competition.
Use the data for 2000 in Table 7 for this question.
a. Construct the 2 x 2 cross-classification tables for (i) ethnic and non-ethnic and (ii) visible minority and non-visible minority.
b. Using the tables in a., compute the following measures of association: phi, V, the contingency coefficient, the odds ratio, Yule’s Q, and the Pearson correlation coefficient. State the null and alternative hypotheses and, if possible, obtain the probability of these statistics. What do you conclude about the relationships in the two tables.
c. From the results of a. and b., comment on the statements of the authors and state what you conclude from the tables and statistics.
a. Is anything accomplished by using phi for some tables and V for others?
b. Comment on the statement by TN in footnote 11:
(11.) Social researchers use correlation coefficients to measure the strength of the relationship between two variables. The choice of coefficient (Phi, Cramer's V, Tau-b and so on) is contingent on the level of measurement within the data. There is no universal agreement within political science or across disciplines on interpreting their magnitude, but generally, coefficients with a range of .00 - .19 indicate non-existent to weak relationships; coefficients in the .20 - .30 range are considered moderate; coefficients in the .30 - .40 range indicate a moderately strong relationship; those in the .40 - .50 range would be considered strong, and so on.
c. Do you have any suggestions for TN in order to make their data and statistical analysis more accessible?
Ethnic and Visible Minority Nominations by Party, 1993-2000 (row
percentages/N in parentheses) *
Political Party 1993 (a,d) 1997 (b) 2000 (c)
Liberals 23.7 (70) 25.6 (77) 26.6 (80)
6.4 (19) 6.6 (20) 7.3 (22)
Reform/Alliance 23.3 (48) 23.8 (54) 23.5 (70)
.5 (1) 4 (9) 5 (15)
Progressive Conservatives 17.3 (51) 21.9 (66) 19.2 (56)
3.1 (9) 2.3 (7) 2.4 (7)
New Democrats 22.9 (67) 25.9 (78) 24.7 (74)
5.8 (17) 4.3 (13) 4.3 (13)
Bloc Quebecois 6.7 (5) 4 (3) 6.7 (5)
2.7 (2) 1.3 (1) 2.7 (2)
* Data for ethnic candidates reported in first row of cells; data for
visible minorities reported in second row of cells Party differences in
(a) Cramer's V = .11
(b) Cramer's V = .12
(c) Cramer's V = .11; all p [less than or equal to] .01 (11) Party
differences in visible minority nominations
(d) Cramer's V = .11; p [less than or equal to] .01
Selected comments by TN on data in Table 2:
The Liberals generally fielded more ethnic and visible minority candidates, although the New Democrats (NDP) and the Reform/Canadian Alliance (in 1997 and 2000) were not far behind their rival (Table 2). The patterns for the Liberals and the NDP are in keeping with the 1988 study, but what is unexpected is that the Reform/Alliance, despite their anti-immigrant images, have fielded a more diverse contingent of candidates than their more established competitor on the political right.
Yet this explanation does not account for why other parties have kept pace with the Liberals' recruitment of ethnic candidates. The NDP's success may be attributed to its programmatic appeal and/or to the incentives it has implemented to boost visible minority candidacies, but the Reform/Alliance's comparable performance is less intuitive, given the party's opposition to Official Multiculturalism and its negative image in many ethno-cultural communities. We propose that while Reform/Alliance has achieved comparable numerical representation of minorities, the neoliberal individualistic principles it expounds may attract individuals who reject or downplay a close identification with their ethnic collectives. The validity of this hypothesis will be ascertained during the elite interviews which were referred to earlier. The Conservatives' image as a white, Anglo-Saxon party remains unchanged in the fourth party system. Despite their long history, they
continue to lag behind more recently-established parties such as the NDP and Reform/Alliance in the area of minority candidate recruitment. Reasons for this cannot be attributed to the lack of appeal which the Conservative program might have for ethnic minorities, but are more likely linked to its historical failure to convey an image of openness to these communities through its organizational structure. Despite the overtures of the Mulroney government toward ethno-cultural communities, including the creation of the first-ever Department of Multiculturalism in 1988 and official apologies for the internment and property expropriation of members of the Japanese and Italian-Canadian communities during World War II, these efforts have not diversified the party's appeal under the subsequent leadership of Campbell, Charest or Clark.
As hypothesized, few minority candidates competed for the separatist BQ. Non-francophones tended to support the federalist Liberals (Crete & Lachapelle, 1996, p. 426), and overwhelmingly rejected the sovereignty option in the 1995 Quebec referendum.
Selected comments by TN on data in Table 7:
The data noted in Table 7 challenges another assumption about the
opportunities available to minority politicians. Neither ethnic nor visible minority politicians competed at a financial disadvantage when compared to politicians from founding groups. This can likely be attributed to legislated limits on candidate spending (Carty, Cross & Young) and to their ability to raise and spend as much money as Charter group candidates. Furthermore, unlike Pelletier's study, which found that in 75 percent of federal ridings, there was only one ethnic candidacy among the major parties, most ethnic and visible minority candidates faced competition from members of other non-founding groups. Between 63 and 82 percent of ethnic or visible minority candidates faced minority competition, compared to between 38.7 and 44.3 percent of Charter group politicians (Table 7).
Our expectations concerning the tendency of ethnic and visible minority candidates to compete in culturally heterogeneous ridings were confirmed. On average, ethnic and visible minority candidates ran for
office in constituencies where between 32 and 43 percent of the population traced their ancestries to non-founding groups (Table 7). In contrast, members of founding groups and non-visible minorities tended to run in relatively homogeneous areas, where about a fifth of the population reported origins that were neither British, nor French nor Canadian.
Select parts of Table 7
Local Conditions for Ethnic, Visible Minority and Non-Ethnic Candidates, 1993-2000
% Candidates Facing Ethnic Competition (row percentages/ N in
Ethnic Candidates 63.1 (152) (h) 65.1 (181) (i)
Non-Ethnic Candidates 38.7 (357) 44.3 (411)
Visible Minority 58.3 (28) (k) 82 (41) (1)
N.-V. Minority 43.1 (481) 47.7 (551)
Ethnic Candidates 64.6 (184) (j)
Non-Ethnic Candidates 42.8 (419)
Visible Minority 61 (36) (m)
N.-V. Minority 47 (567)
% of Minorities in Riding (N in parentheses)
Ethnic Candidates 31.7 (241) (b) 33.0 (278) (c)
Non-Ethnic Candidates 19.3 (923) 17.9 (927)
Visible Minority 31.7 (48) (e) 43 (50) (f)
N.-V. Minority 21.4 (1116) 20.5 (1155)
Ethnic Candidates 32 (285) (d)
Non-Ethnic Candidates 17.5 (980)
Visible Minority 37.3 (59) (g)
N.-V. Minority 20 (1206)
F-scores: a) 6.3, p [less than or equal to] .01; b) 111.86; c) 172.70;
d) 160.13; e) 17.1; f) 80.08; g) 54.3; b thru g, p [less than or equal
Phi: h) .20; i) .18; j) .18; k) .06, p [less than or equal to] .05; l) .14; m).06, p [less than or equal to] .05; h,i,j,l, p [less than or equal to] .001
This study identifies the factors associated with the nomination and election prospects of ethnic and visible minority candidates in three federal elections held between 1993 and 2000. We conducted a statistical analysis of how party affiliation, the geographic location of a riding, incumbency, local party competitiveness, riding cultural heterogeneity, campaign spending, minority competition, and a candidate's racial or ethnic background influenced electoral outcomes for 3,634 candidates. We found that ethnic and visible minority nomination and election rates did not slip from their climb in the third party system, but have stalled. Furthermore, macro factors such as party affiliation and geography did not generally account for differential election rates between Charter group and non-Charter group politicians. Instead, local party competitiveness and campaign spending were better predictors of the
likelihood of a minority victory at the ballot box. Visible minorities
continue to be underrepresented in candi dacies and in the parliamentary
ranks, but there was no evidence to show they competed in unfavourable local contexts. The key to electing more visible minorities lies in recruiting more of these individuals to run for public office.
URL for article from University Library: