SoWL 2018 Data Collection

#General information ###Target and survey population The target population for the SoWL 2018 is the non-institutionalised population 15 years and over who usually live in Zealand and were employed in the Household Labour Force Survey (HLFS) reference week. ‘Employed people' includes all respondents who worked for pay or profit, or worked without pay in a family business, or who had a job or business that they were away from because of sickness, holidays, or any other reason. The survey population is the target population with the following exclusions: residents of islands other than the North Island, South Island, and Waiheke Island; people residing in non-private dwellings and people not living in permanent dwellings. Also excluded from the survey population are long-term residents of retirement homes; hospitals and psychiatric institutions; inmates of penal institutions; and people who are not expecting to be resident in New Zealand for more than 12 months. ###Reference period The SoWL 2018 was a supplement to the HLFS and was carried out in the December 2018 quarter. All eligible responding individuals (15 years and over and employed in the reference week) were asked to participate in the SoWL. The SoWL questionnaire was due to be conducted from Sunday the 7th October 2018 and finish on Saturday 5th January 2018 in conjunction with the HLFS. However, due to the affect SoWL was having on the HLFS achieved sample rate, it was decided to terminate interviewing of SoWL prematurely on the 18th December 2018. ###Achieved sample size There was a total of 9,395 usable records for SoWL 2018. ###Achieved sample rate The achieved sample rate target for the SoWL was 80 percent. The achieved sample rate is calculated as the weighted number of eligible individuals (employed people) that responded to SoWL divided by the total weighted number of individuals sampled within the HLFS. Essentially, it tells you what percentage of the sample responded to the survey. Due to the early termination of SoWL, not all individuals who were eligible for SoWL were interviewed, and the final achieved sample rate was 50.2%. Even though the achieved sample rate was low, there was minimal bias present. The non-response adjustment and calibration to benchmarks reduced the impact of this bias. Non-response was partly due to the increased burden of it being a supplement to the HLFS, and partly because proxy responses were not accepted. A proxy response is a response by one member of a household on behalf of another. Benchmarks are generated from the employed dataset of HLFS. ###Data collection methods SoWL was collected by computer-assisted interviewing (CAI). 30 percent of the data was collected for households by computer-assisted personal interviewing (CAPI). The remaining households were surveyed by centralised computer-assisted telephone interviewing (CATI). #Comparability to other SoWL surveys Due to various reasons stated below, Stats NZ advises caution with direct comparisons between the three (2008, 2012 and 2018) SoWL surveys. ###Questionnaire changes Due to the redevelopment of the HLFS in the June quarter 2016, some of the variables within the SoWL questionnaire were deemed of high importance were included within the HLFS. This made it possible to include new information into the SoWL for 2018. New information that was collected includes types of self-employment, skill mismatch, workplace relationships, physical and dangerous work, and employer size and institutional sector. There have also been changes and additions made to study/training, work from home and workplace health and safety, an expanded collection on information around ‘difficulties’ caused by working under certain conditions (non-standard hours, short notice to work, expectation to be available for work, and changing working time arrangements). Adding and removing questions, adjusting the sequence and slightly adjusting the wording of questions - can all potentially have an effect on estimates. ###Household Labour Force Survey (HLFS) redevelopment The HLFS redevelopment in 2016 saw key SoWL variables moved into the HLFS. This included employment relationship, union membership, employment agreements and certain work preference questions. Estimates for these variables may differ between 2008/2012 when they were collected in SoWL, and in 2018 when they were collected within the HLFS. It should be noted that possible differences in estimates for these variables may be caused by the fact SoWL does not allow for proxy responses, whereas the HLFS does. The redevelopment also brought in changes to the target population which now includes overseas diplomats and armed force members if living in New Zealand for 12 months or more, as well as including New Zealand armed forces members if they are living in private dwellings. With the 2016 redevelopment there was also a redesign of the employment questions within the HLFS. This did cause a level shift in the measurement of some employment statuses although this was mainly seen with the unemployed and not in the labour force groups, which does not affect SoWL. Under the redevelopment the identification of self-employed people was also improved which also saw a decline in the number of employees. ###Sample size Due to the early termination of the 2018 SoWL only 9,395 employed individuals responded. This sample size is a lot smaller than that achieved for the 2012 SoWL when 14,335 employed individuals responded, and for the 2008 SoWL when 14,510 employed individuals responded. The smaller sample size for 2018 has led to estimates having slightly larger sampling errors recorded for them. ###Seasonality The SoWL 2018 was run in the December quarter – because working times and hours during the December quarter are likely to be more typical of those over a full year. Interviewing during the December quarter would also enable a good coverage of seasonal workers. There are seasonal differences seen between quarters for the HLFS. For example, the December quarter has shown a higher number of people in the retail trade industry – due to the holiday shopping period – compared with other quarters during the year. These differences can also affect SoWL. It is not recommended to directly compare the December 2018 and 2012 SoWL to the March 2008 SoWL – this could lead to inaccurate conclusions. ###Australian and New Zealand Standard Industry Classifications The 2018 and 2012 SoWL industry statistics are based on the Australian and New Zealand Standard Industrial Classification 2006 (ANZSIC06) which is the latest edition of the industry classification. The 1996 version (ANZSIC96) was used in the 2008 SoWL release. Industry outputs defined using ANZSIC06 are not comparable with those based on ANZSIC96. #Information about the data ###Sampling errors Two types of errors are possible in estimates based on a sample survey: sampling error and non-sampling error. Sampling error can be measured and quantifies the variability that occurs by chance because a sample rather than an entire population is surveyed. Non-sampling errors are all errors that are not sampling errors and are not quantifiable. Sampling errors have been estimated using a jack-knife method, which is based on the variation between estimates, based on different subsamples taken from the whole sample. This is an attempt to see how estimates would vary if we were to repeat the survey with new samples of individuals. For example, the estimated total number of employees that are permanent employees is 1,961,900. This estimate is subject to a sampling error of plus or minus 33,000 or 1.7 percent (measured at the 95 percent confidence level). This means that there is a 95 percent chance that the true number of permanent employees lies between 1,928,900 and 1,994,900. Smaller estimates, such as the total number of casual workers who are on a collective agreement is 14,000, are subject to larger relative sampling errors than larger estimates. This estimate is subject to a sampling error of plus or minus 4,600 or 32.9 percent (measured at the 95 percent confidence level). This means that there is a 95 percent chance that the true number of casual workers who are on a collective agreement lies between 9,400 and 18,600. Non-sampling errors are all errors that are not sampling errors and are not quantifiable. Non-sampling errors include unintentional mistakes by respondents when answering questions, variation in the respondent's and interviewer's interpretation of the questions asked, and errors in recording and coding data. Stats NZ endeavours to minimise the impact of these errors through applying best survey practices and by monitoring known indicators (e.g. non-response). ###Ethnicity classification SoWL uses the total response output method for classifying ethnicity. Using this method, people who reported more than one ethnic group are counted once in each group reported. This means the total number of responses for all ethnic groups can be greater than the total number of people who stated their ethnicities, and calculated percentages can add up to be greater than 100 percent. ###Suppressed estimates in this release Some estimates are suppressed (replaced by 'S' in the tables) for reliability reasons. These suppressed estimates have a weighted value of less than 1,000 and reflect a low number of responses which are subject to larger relative sampling errors. ###Editing A minimal approach to editing was implemented for the SoWL 2018. With CAI, the computer software runs checks for validity and consistency as responses to questions are captured. If required, the software prompts the interviewer to clarify answers with the respondent at the time of interview. This keeps the number of inconsistent answers low. A further round of validity, logic, and error checks was performed on the data as part of the data processing stage. A small amount of editing was done post collection on income. Editing was also done on a small number of cases where information was collected in the HLFS by proxy, and subsequently recollected in SoWL (from the actual respondent) and was found to be inconsistent. ###Imputation For respondents who belong to eligible responding households and who have missing values for sex, age, ethnicity, usual and actual hours worked in all jobs, income from jobs (main and second) values would have been imputed by the HLFS and used in SoWL. However, hours and income were re-asked in SoWL which restricted the imputation needed for these variables. Non-response to the SoWL occurred either when an eligible individual in the sample did not respond to the SoWL questionnaire or did not provide information to core questions to determine a usable response. Any non-responding employed individuals from the HLFS sample were dealt with by adjusting the weights of the responding SoWL individuals. Item (or partial) non-response could occur within the responses for SoWL. This includes a response of 'don't know' or 'refused' to non-core questions. Since all the core SoWL questions had been answered, the record was deemed to be a full response. No imputation has been applied to any item non-response for variables collected within the SoWL questionnaire. ###Mean (average) and median The mean or average is calculated as the total divided by the number of units in the population. A mean can be sensitive to extreme values. Unusually high or low values will have a large impact on the estimate of the mean. The median is the value at which half the units in the population have lower values and half have higher, when all values are ordered from highest to lowest. It corresponds to the 50th percentile. The median is less sensitive to extreme values than the mean. ###Percentages Percentages in articles about the SoWL are calculated excluding those who did not specify a response i.e. those who said 'don't know' or 'refuse' to a question. ###Rounding All estimates provided in this release are independently rounded to the nearest hundred. For this reason, estimated totals may differ from the sum of individual cells. All percentages are calculated using unrounded figures and are rounded to whole numbers. ###Confidentiality Only people authorised by the Statistics Act 1975 are allowed to see your individual information, and they must use it only for statistical purposes. Your information is combined with similar information from other people or households to prepare summary statistics. ###Liability While all care and diligence has been used in processing, analysing, and extracting data and information in this publication, Stats NZ gives no warranty it is error-free and will not be liable for any loss or damage suffered by the use directly, or indirectly, of the information in this publication. en-NZ