In lieu of an abstract, here is a brief excerpt of the content:

Statistical฀Procedures฀฀฀฀฀฀฀◆฀฀฀฀฀฀฀211 211 APP ENDIX฀ I Statistical฀Procedures A.฀Sampling฀methods฀for฀the฀manuscript฀census฀returns The samples taken from the manuscript census returns were drawn indirectly. From the 1869 census on, the Imperial-Royal Central Statistical Commission drew up census questionnaires for each household and distributed them in cities to the owner of every developed parcel of land. The latter, in turn, gave one to the head of each resident household, who completed it. When the local officials in Prague collected and compiled the returns, they bound together all the questionnaires for each parcel, typically a single structure, and filed them under the number of the parcel.1 Since the census takers prepared no general index of the population or of the heads of households, the only practical means of sampling the census materials is to select the returns for a group of parcels in each district under study. For simplicity and to facilitate time-series analysis, my samples included the same parcels in each selected census. The territorial definitions of the inner-city parishes established in 1870 were used throughout.2 I drew systematic samples for the parish of St. Henry, beginning arbitrarily with the eleventh parcel and taking every twentieth parcel, and in the smaller parish of St. Gall every tenth parcel, beginning with the tenth. For the parishes of St. Peter and St. Nicholas, I renumbered the parcels consecutively and with a table of random numbers drew simple random samples of 10 percent of the parcels. I made no substitutions in any of the samples for parcels for which manuscript returns from one or more censuses were missing from the materials preserved in the Prague City Archive. The systematic samples can be treated as random for purposes of this study. The parcel numbers in each of the five historic districts of the inner city generally run in ascending order in a radial pattern from the center of each district. Since the character of each building’s inhabitants had no apparent relationship to the parcel numbers, the systematic samples can be considered fully equivalent to the random samples.3 Once the manuscript census returns were drawn for the sampled parcels, all the civilian residents listed in the returns were counted. Statisticians term this a single-stage cluster sampling.4 To simplify all calculations, however, I have 212฀฀฀฀฀฀฀◆฀฀฀฀฀฀฀APPENDIX฀฀I treated all the samples as if they were direct and random. Rigorous statistical procedure requires the use of special formulae for calculating confidence intervals and testing for significant differences when one draws clusters rather than sampling a population directly.5 Using strict procedures, one must draw a cluster sample that is larger in absolute terms than an equivalent simple random sample in order to achieve the same confidence interval. If the clusters are heterogeneous in composition, however, as was nearly always the case for the parcels included in the samples, the differences between the results of direct and cluster sampling are minimal. Accordingly, I have used the simpler formulae pertaining to direct random samples for all the significance tests and confidence intervals. A number of simplifications were also made to facilitate work with the manuscript census returns. All the civilian residents, citizen or alien, who were reported in the census as present or only temporarily absent were counted together. Moreover, all individuals in the samples were counted as either Czech-speakers or German-speakers. The number of residents who reported other languages was nearly always less than 2 percent, and in most instances it seemed reasonable to count them with the Czechs or Germans in whose households they lived. The French and English governesses in some of the German households, for example, usually reported their native tongues as their everyday language. Nearly all these governesses also spoke German, and since there were not many of them, it seems reasonable to count them together with the German population. Technical limitations and considerations of time required the tallying of the totals for each parcel directly from the manuscript returns in the Prague City Archive. The tally sheet used the same basic format as tables 1 and 2 in appendix 2. B.฀Occupational฀categories The Austrian Central Statistical Commission used the following general categories in the censuses after 1880 for analyzing the sources of livelihood for the population: Agricultural (Agr.) = cultivation, raising of livestock, forestry, fishing. Manufacture (Manuf.) = manufacture of all goods, mining, printing, construction, handicraft production. Commerce (Com.) = all trade and...

Share