Consequences of Data Heaping in the British Religious Census of 1851

Abstract

Abstract. The 1851 Census of Religious Worship contains an excess of "round" numbers reported for congregation sizes, which indicates that estimates (as opposed to exact head counts) were often returned. The authors estimate the proportion of rounded estimates returned by the Church of England and the dissenting denominations. They develop a structural equation model to estimate for the mean degree of inflation (i. e., overstimation) that resulted from rounding. Anglican figures were most likely inflated by approximately 11 percent, whereas the corresponding figure for dissent was approximately 5.5 percent. The authors examine what sort of counting behavior could have led to such margins of inflation, and why denominational differences exist.