| Sumario: | The errors recorded by most public opinion polls in reporting the recent election clearly indicate the need for a reexamination of procedures. Accurate measurements of public opinion on any issue depend on several factors: preparation of a valid cross-section of the universe to be studied; the best possible fieldwork to insure that the cross-section is actually obtained; and correct timing. Of these three factors, careful fieldwork is the most difficult to secure and too often has received the least attention. In this connection the New Jersey Poll interviewing methods may highlight the importance of fieldwork in political forecasting and, for that matter, in all opinion research. The New Jersey Poll made predictions covering six candidates in New Jersey. All predictions were correct as to winners and standings. Definite reasons lie behind the employment of these techniques adopted for the opinion polls conducted in New Jersey. Each attacks a potential source of error which in most cases is likely to give a Republican bias on election forecasts. The irreducible statistical error is already of sufficient size to make predictions of close elections somewhat hazardous; therefore, every effort must be made to reduce the error subject to control.
|
|---|