

The application of our various formulas concerning the bases of party support to an actual data set in the parties project is illustrated in Table 5.3, which reproduces a page of computer output from the special program that calculates our various measures. The three data matrices in Table 5.3 conform to three ways of calculating percentages for the basic sample survey data, here taken from a 1959 survey in West Germany and pertaining to education as a cultural differentiator. This is data source 15 for West Germany, mentioned earlier in the context of our discussion of reliability. The first matrix in the table computes percentages by columns and is used in computing the party attraction scores which are located on the upper lefthand side of the printout. Thus we see that the CDU with a score of .89 tends to be more attractive of educational groupings than the SPD with a score of .74. Although the effective sample size for this analysis is 628 cases, the numbers of cases involved in percentages for the columns are relatively small in the instances of secondary and university educationwith the latter group represented by only 18 respondents. Thus the associated AC codecalculated according to the procedures specified in Table 5.2is only 6 for each of the attraction measures. The second matrix in Table 5.3 computes percentages by rows and is used in computing the party concentration scores, which are located in the middle of the lefthand side of the printout. The SPD's concentration score of .77 reflects its high proportion of supporters with only primary education, making it more concentrative of educational groupings than the CDU, with a score of .52. Note that the associated AC codes are 8 for both the CDU and SPD but that the AC is only 4 for the FDP, whose proportions are based on a sample size of only 30 and are thus less reliable. As explained in Table 5.2, the adequacyconfidence codes for concentration measures are calculated differently from those for attraction and depend on the number of cases in the row totals rather than the column totals. The third matrix in Table 5.3 computes percentages by the total effective number of cases to exhaust the three methods of calculating percentages within a table. The bottom margin, which shows the proportional distribution of subgroups in the society, is matched against the party composition proportions in the second matrix in computing the party reflection scores, which are located at the bottom of the lefthand side. of the printout. The reflection score for the CDU (.90) is somewhat higher than that for the SPD (.82), because the distribution of CDU supporters by educational grouping is closer to the educational composition of the West German societyas computed in the column totals of the 12W E S T
GERMANY ANALYSIS OF SUPPORTERS FOR
CULTURAL DIFFERENTIATOR  EDUCATION FOR BOTH PARTS OF
PERIOD ATTRACTION MATRIX PERCENT OF
SUPPORT GIVEN TO PARTY FROM EACH SOCIAL SUBGROUP
[Percents are of column
totals] POLITICAL PARTY CDUCSU SPD FDPOVP OTHER CONCENTRATION VAR.
4.06 CONCENTRATION MATRIX PERCENT
OF TOTAL PARTY STRENGTH CONTRIBUTED BY EACH
SUBGROUP [Percents are of
row totals] CDUCSU SPD FDPOVP OTHER PARTY CLEAVAGE MATRIX PERCENTS
BASED ON THE TOTAL EFFECTIVE NUMBER OF CASES
[All percents are of total
cases (628)] CDUCSU SPD FDPOVP OTHER (Total Effective N.
of Cases)
DATA SOURCE: 15 CODED BY KENNETH
JANDA The survey was conducted in 1959
and reported in Gabriel Almond and Sidney Verba,
The Civic Culture. the data were obtained from the
ICPSR and processed at the University of
Pennsylvania by Farid Lavipour. The original sample
was 955. The party identification item is based on
recall of the vote in the 1957 election with DK/NA
and 'didn't vote' eliminated from this analysis.
DK/NA responses to the education question were also
eliminated. 
