Weryfikacja danych uzyskanych z odpowiedzi respondentów na pytania o wysokość zarobków
Streszczenie
120 manual workers in six industrial plants in the city of Lodz were interviewed and one of the questions asked related to their average monthly wages. The wages given in interviews were compared with the actual ones, i.e. those obtained from the files in their work establishments. In addition, 30 interviews about an interview were conducted with other respondents asked about their salaries and wages; in those interviews the respondents were asked probing questions about how did they answer the question concerning salaries or wages, what components did they take into account, whether the question was difficult or sensitive and the like. It was found that: 1° — wages given in interviews differed considerably from the actual ones; 2° — the average absolute difference between the declared and actual wages was 296 zlotys; 3° — 50% of the respondents committed errors exceeding 10% of their actual wages, and nearly 25% — errors exceeding 20% of their actual wages; 4° — while the respondents were strongly diversified in respect of the direction and magnitude of errors committed, as many as 63% of them were understating their wages by over 100 zlotys which means that the errors were systematic; 5° — the magnitude and direction of errors depended on the level of actual wages, the work place, and the kind of work. Respondents with higher wages and those having a more irregular work were understating, their wages more frequently and to a greater extent than were those with lower wages.
Despite these errors, the value of the Pearson-Bravais coefficient of correlation between actual and declared wages was 0,83 which means that the structure of the declared wages resembled that of the actual ones. Errors did not influence appreciably the strength of relationships between selected variables: C — coefficients between the answers to several selected questions and the declared as well as the actual wages were computed and turned out to be very similar.
Errors may be committed by respondents knowingly or unknowingly. Various categories of causes of errors were distinguished, the most important being those connected with: 1° — ambiguity of the question (as when it is not explained clearly to respondents how the question should be answered); 2° — general sensitivity of questions concerning income, wages earnings etc. This sensitivity may be enhanced by such social factors as fearful traditional attitudes and habits that underlie the answering such questions as well as by individual personality traits of respondents.
Collections