Occupation Coding During the Interview

Abstract

Summary. Currently, most surveys ask for occupation with open-ended questions. The verbal responses are coded afterwards, which is error prone and expensive. We present an alternative approach that allows occupation coding during the interview. Our new technique uses a supervised learning algorithm to predict candidate job categories. These suggestions are presented to the respondent, who in turn can choose the most appropriate occupation. 72.4% of the respondents selected an occupation when the new instrument was tested in a telephone survey, entailing potential cost savings. To aid further improvements, we identify some factors for how to increase quality and to reduce interview duration.

Natives and Migrants in Home Production: The Case of Germany

“…have a child younger than 1 in the household; while it induces a 0 043 percentage point increase for the one aged 36-45 (Panel B, Col 5 and 6) It should be noticed that the share of female immigrants is measured at time t as well as the presence of a child aged 0 1 in the household Clearly, a woman decides to become pregnant at time t 1 For robustness, in unreported regressions, we regress the share of female migrants at time t 1 on the probability to have a child aged 0 1 in the household at time t and we obtain very similar results (note that the uncoditional correlation between the share of female migrants at time t and the one at time t 1 is 0 99) Female immigrants can induce native women not only to have more children, but also to work more hours and devote less time to home production activities Table 5 shows that the share of female immigrants increases the probability for a native woman to work more than a certain amount of hours per week 15 Consistently with the results that we have found on fertility decisions, when we divide the sample by skill level, we fnd a positive and statistically signifcant efect only for the medium skilled native women For example, the estimate suggests that a 1 percent increase in the share of female migrants induces a 0 10 percentage point increase in the probability for a medium skilled native woman to work more than 40 hours per week (Panel A, Col 8) 16 Again, if we consider separately the efect for the medium skilled women aged 22-35 and 36-45, the results hold for both groups, with a stronger statistically efect for the frst one In particular, a 10 percent increase in the main explanatory variable raises by 1 29, 1 33, and 1 14 percentage points the probability a (medium skilled) woman aged 22-35 to work more than 30, 35, and 40 hours per week, respectively (Panel B, Col 2, 3, and 4) The results seem to indicate that immigration helps medium skilled native women to better reconcile work and family responsibilities, especially in their early career stage These fndings are further supported by the results reported in Table 6, which presents the estimates of the impact of the share of female immigrants on the number of weekly hours (logs) that native women devote to housework and childcare 17 As in the previous cases, we split the estimation sample by three skill levels We fnd that migration decreases the time that medium skilled native women allocate to housework and childcare Considering the efect by age groups, it is clear that the results hold especially for the youngest ones: a 1 percent increase in the share of female migrants decreases by 0 4 percent the total amount of hours spent by medium skilled women aged 22-35 in both housework and childcare activities (Panel C, Col 5) Taken all together, our results suggest that medium skilled women, aged 22-35, are more likely to respond to changes in the availability of household services (in this case due to immigration) working more hours, having a child and devoting less hours to home production 1 We report only the estimated coefcient of the variable of interest. Estimated coefcients for control variables and FE/IV results are provided in Tables A.. and A.5 of the Appendix A.…”

“…Results are available upon request. 17 We report only the estimated coefcient of the variable of interest and IV results. Estimated coefcients for control variables are provided in Tables A.6 of the Appendix A.…”

Natives and migrants in home production: the case of Germany

“…14 The correlation is also weekly signifcant for the high skilled women, but with a negative coefcient. have a child younger than 1 in the household; while it induces a 0 043 percentage point increase for the one aged 36-45 (Panel B, Col 5 and 6) It should be noticed that the share of female immigrants is measured at time t as well as the presence of a child aged 0 1 in the household Clearly, a woman decides to become pregnant at time t 1 For robustness, in unreported regressions, we regress the share of female migrants at time t 1 on the probability to have a child aged 0 1 in the household at time t and we obtain very similar results (note that the uncoditional correlation between the share of female migrants at time t and the one at time t 1 is 0 99) Female immigrants can induce native women not only to have more children, but also to work more hours and devote less time to home production activities Table 5 shows that the share of female immigrants increases the probability for a native woman to work more than a certain amount of hours per week 15 Consistently with the results that we have found on fertility decisions, when we divide the sample by skill level, we fnd a positive and statistically signifcant efect only for the medium skilled native women For example, the estimate suggests that a 1 percent increase in the share of female migrants induces a 0 10 percentage point increase in the probability for a medium skilled native woman to work more than 40 hours per week (Panel A, Col 8) 16 Again, if we consider separately the efect for the medium skilled women aged 22-35 and 36-45, the results hold for both groups, with a stronger statistically efect for the frst one In particular, a 10 percent increase in the main explanatory variable raises by 1 29, 1 33, and 1 14 percentage points the probability a (medium skilled) woman aged 22-35 to work more than 30, 35, and 40 hours per week, respectively (Panel B, Col 2, 3, and 4) The results seem to indicate that immigration helps medium skilled native women to better reconcile work and family responsibilities, especially in their early career stage These fndings are further supported by the results reported in Table 6, which presents the estimates of the impact of the share of female immigrants on the number of weekly hours (logs) that native women devote to housework and childcare 17 As in the previous cases, we split the estimation sample by three skill levels We fnd that migration decreases the time that medium skilled native women allocate to housework and childcare Considering the efect by age groups, it is clear that the results hold especially for the youngest ones: a 1 percent increase in the share of female migrants decreases by 0 4 percent the total amount of hours spent by medium skilled women aged 22-35 in both housework and childcare activities (Panel C, Col 5) Taken all together, our results suggest that medium skilled women, aged 22-35, are more likely to respond to changes in the availability of household services (in this case due to immigration) working more hours, having a child and devoting less hours to home production 1 We report only the estimated coefcient of the variable of interest. Estimated coefcients for control variables and FE/IV results are provided in Tables A.. and A.5 of the Appendix A.…”

Transforming Naturally Occurring Text Data Into Economic Statistics: The Case of Online Job Vacancy Postings

“…Official classifications can be applied inconsistently. Schierholz et al (2016) surveys disagreements amongst those who code job titles into occupational classes, finding that the agreement overlap between coders is around 90% at the first-digit of the code (the highest level, for instance "Managers, Directors and Senior Officials") but reduces to 70-80% at the 3-digit level that we work with for SOC codes (for instance, "Managers and proprietors in agriculture related services"). Automated approaches which use job title alone have even lower levels of agreement; Belloni et al (2014) showed that algorithms which use job title alone agree on only 60% of records even at the top, 1-digit level of the International Standard Classification of Occupations.…”