Understanding Society User Support: Issueshttps://iserredex.essex.ac.uk/support/https://iserredex.essex.ac.uk/support/support/favicon.ico?15995719382024-01-18T13:10:31ZUnderstanding Society User Support
Redmine Support #2032 (Resolved): hidp missing from hhresp but present in income, egoalt, indresp fileshttps://iserredex.essex.ac.uk/support/issues/20322024-01-18T13:10:31ZChris Grollman
<p>Hi there,</p>
<p>I am taking values of <em>ficode</em> from the <strong>income</strong> file, and want to attach them to the associated household file <strong>hhresp</strong> using a join by <em>hidp</em>.</p>
<p>Across waves f to l I have 4111 unique values of <em>hidp</em> from the <strong>income</strong> file. When I link to <strong>hhresp</strong> using <em>hidp</em> I find that 119/4111 (2.9%) do not link: that is, the <em>hidp</em> in <strong>income</strong> doesn't occur in the <strong>hhresp</strong> file. The proportion where the <em>hidp</em> is not present in <strong>hhresp</strong> varies across the seven waves I am looking at.</p>
<p>I have looked in detail for one <em>hidp</em> in wave j. That <em>hidp</em> occurs in the <strong>income</strong>, <strong>egoalt</strong> and <strong>indresp</strong> files, with a total of four <em>pidps</em> living at that household. But the <em>hidp</em> doesn't appear in <strong>hhresp</strong>.</p>
<p>What is going on?!</p>
<p>Thank you!<br />Chris</p> Support #2016 (Resolved): Family connectionshttps://iserredex.essex.ac.uk/support/issues/20162023-12-19T09:56:11ZStav Federman
<p>Hello,</p>
<p>I have a few questions regarding family connections:</p>
<p>1. I read the user guide for the family matrix (xhhrel) and want to ensure I understand how family connections are formed. As I understand it, information is collected for core sample members and their household members. So if I see a connection between an adult child and a parent, it means that they had to be cohabitating in one of the waves of the survey. Hence, if a person did not cohabit with their parents at any point, I wouldn't have any information regarding this family connection. Is that true? <br />2. Suppose that the core sample member lived alone in waves 1-3, and then cohabited with their adult child in wave 4, and then the child moved out (from wave 5 onwards). Then, is the child being followed for the rest of the waves (4 and onwards), or is he being sampled just for the wave in which he cohabits with the core sample member (wave 4)? <br />3. Does the family matrix contain connections made in the British Household Panel Survey as well, or only connections of family members cohabitating in understanding society?</p>
<p>Thank you</p> Support #2006 (Resolved): Longitudinal analysis using calendar year?https://iserredex.essex.ac.uk/support/issues/20062023-12-12T13:52:21ZMarina Kousta
<p>Hello,</p>
<p>I am reaching out to kindly request help on how to conduct longitudinal analysis using calendar year datasets.<br />1) Although online you state the published calendar year data are meant to be used for cross-sectional analysis, does that also stand for when we create our own calendar year datasets? Or is it meant to be a guidance only for when you release the pre-made calendar year data? If that is the case regardless, is there some way for us to still conduct longitudinal analysis after creating our own calendar year data?<br />2) Although you recommend using the w_month (sample month) to create calendar year data, would it still be ok to instead use the interview date instead, when the exact date is of great importance to the research question itself (i.e. when testing the introduction or removal of a social policy).</p>
<p>Many thanks in advance for your time and consideration.</p>
<p>Best wishes,<br />Marina</p> Support #1999 (Resolved): treatment of HMOshttps://iserredex.essex.ac.uk/support/issues/19992023-12-05T14:30:43ZMarika Cioffi
<p>Hi, I am interested in how USOC treats full-time students and HMOs in terms of households composition. More specifically,<br />Are full-time students living alone or with other people (not family) considered as being one-person households or are they part of the household of their parents?<br />Are they the “reference person” of their own household or are they just a member of the household where one of their parents is the reference person?</p> Support #1966 (Resolved): cross-sectional analysishttps://iserredex.essex.ac.uk/support/issues/19662023-09-05T18:24:43ZZahra Khan
<p>I am little confused since I have read a lot about cross-sectional research using the understanding society data: since understanding society is a longitudinal survey- how is a valid cross-sectional analysis a possibility? <br />I am interested in looking at hours spent on doing housework for South Asians. A single wave does not allow me sufficient data for this analysis. So, if I pool different waves- will this be considered a true cross-secitonal analysis whereby the respondents are different in each wave?</p> Support #1945 (Resolved): Using the individual questionnairehttps://iserredex.essex.ac.uk/support/issues/19452023-08-01T12:11:05ZJoshua Parry
<p>When doing analysis at the individual level. Is it valid to do research using multiple members of the same household or would you just one person within each household.. For instance, if i was trying to assess the prevalence of volunteering in individuals. Would you use one individual response per household or the response of all individuals from each household?</p> Support #1937 (Resolved): "Inapplicable" responses for dweltyphttps://iserredex.essex.ac.uk/support/issues/19372023-07-10T14:08:37ZVikram Patil
<p>Hello, <br />My question is about the variable "dweltyp" in the hhsamp files. <br />- Why are so many households reported as "-8" or "inapplicable" from Wave 7 onwards in the Understanding Society (USoc) data? <br />- How is "dweltyp" different from the "hstype" variable that was included in the BHPS?<br />- Why are there no inapplicable responses prior to Wave 7? <br />Thanks!</p> Support #1843 (Resolved): missing data on parental educationhttps://iserredex.essex.ac.uk/support/issues/18432023-01-19T19:03:27ZYaojun LiYaojun.Li@manchester.ac.uk
<p>Dear Sir/Madam,</p>
<p>Sorry to trouble you again.</p>
<p>I wish to ask a question on parental education in the USoc. I understand that the BHPS data were 'rolled-in' at W2 of the USoc and parental education was asked at W13 of the BHPS. Information of parental education was also collected at w1 of the USoc. So, in the USoc, parental education was a combination of the BHPS W13 and USoc W1 data. Looking at the xwdat at Wave 12 of the Usoc, I could see that both paedqf and maedqf contain both BHPS and USoc sources information.</p>
<p>. ta xwdat_dv paedqf</p>
<pre><code>Study enumerated in: | Father's educational qualifications<br /> UKHLS, BHPS or both | missing don't kno he did no he left s he left s he gained he gained other | Total<br />----------------------+----------------------------------------------------------------------------------------+----------<br />in UKHLS xwavedat onl | 68,481 0 1,381 15,084 8,756 8,596 4,584 239 | 107,121<br />in BHPS xwavedat only | 21,326 664 44 2,496 804 1,202 332 2 | 26,870<br /> in both | 5,976 781 93 4,659 1,765 2,390 716 22 | 16,402<br />----------------------+----------------------------------------------------------------------------------------+----------<br /> Total | 95,783 1,445 1,518 22,239 11,325 12,188 5,632 263 | 150,393</code></pre>
<p>. ta xwdat_dv maedqf</p>
<pre><code>Study enumerated in: | Mother's educational qualifications<br /> UKHLS, BHPS or both | missing don't kno she did n she left she left she gaine she gaine other | Total<br />----------------------+----------------------------------------------------------------------------------------+----------<br />in UKHLS xwavedat onl | 66,753 0 2,323 17,359 11,305 6,151 3,075 155 | 107,121<br />in BHPS xwavedat only | 21,326 503 45 2,855 1,156 758 226 1 | 26,870<br /> in both | 5,917 529 108 5,179 2,554 1,563 536 16 | 16,402<br />----------------------+----------------------------------------------------------------------------------------+----------<br /> Total | 93,996 1,032 2,476 25,393 15,015 8,472 3,837 172 | 150,393</code></pre>
<p>My question is: why is there so much missing on parental education? 92902/150393 = 61.8%</p>
<p>. ta paedqf maedqf</p>
<pre><code>Father's educational | Mother's educational qualifications<br /> qualifications | <del>9. missi -1. don't 1. she di 2. she le 3. she le 4. she ga 5. she ga 97. other | Total<br />---------------------</del><ins>----------------------------------------------------------------------------------------</ins>----------<br /> <del>9. missing | 92,902 0 148 1,275 819 481 147 11 | 95,783 <br /> -1. don't know | 0 711 5 384 194 134 17 0 | 1,445 <br />1. he did not go to s | 12 6 1,182 189 95 25 6 3 | 1,518 <br />2. he left school wit | 327 126 465 17,291 2,723 1,069 227 11 | 22,239 <br />3. he left school wit | 271 69 408 2,429 6,260 1,384 500 4 | 11,325 <br />4. he gained post sch | 387 112 150 3,271 3,656 3,879 702 31 | 12,188 <br />5. he gained a univer | 79 8 102 494 1,221 1,477 2,229 22 | 5,632 <br /> 97. other | 18 0 16 60 47 23 9 90 | 263 <br />---------------------</del><ins>----------------------------------------------------------------------------------------</ins>----------<br /> Total | 93,996 1,032 2,476 25,393 15,015 8,472 3,837 172 | 150,393</code></pre>
<p>I would be most grateful for your explanation and guidance.</p>
<p>Best wishes and many thanks in advance.</p>
<p>Yaojun</p> Support #1811 (Resolved): Longitudinal weight selectionhttps://iserredex.essex.ac.uk/support/issues/18112022-11-21T16:02:56ZRebeka Balogh
<p>Hello,</p>
<p>I'd have (another) question on longitudinal weight selection. <br />I've conducted a latent class analysis in wave 4 and I'm using the latent class membership as an explanatory variable in subsequent analyses. However, another important explanatory variable I've derived is the length of unemployment respondents had been exposed to prior to wave 4. This I've derived from the employment status history modules and the annual event history modules. The employment status history module was asked to some households in wave 1 and to others in wave 5. For those that were asked in wave 5, I've 'backdated' their employment spells. My question concerns the correct weight to use. Originally I was planning on fitting regressions on the outcome in wave 5 in which case I would have used the correct wave 5 longitudinal weight. However, now I'm considering looking at an outcome which is only available in wave 4. However, because of the unemployment variable, would it be methodologically sound to still apply the wave 5 longitudinal weight (given that some of the information on prior unemployment has been derived from wave 5?) even when I'm looking at an outcome at wave 4? <br />Thanks very much for the help.</p>
<p>Best wishes<br />Rebeka</p> Support #1770 (Resolved): Making best use of the Ethnic Minority Boost Samplehttps://iserredex.essex.ac.uk/support/issues/17702022-10-02T14:18:46ZLaurence Rowley-Abel
<p>Hi Understanding Society team,<br />I am conducting a basic analysis of health outcomes across combinations of ethnic groups and age groups using the Wave 10 data. My issue is that when I produce things like cross-tabulations, once I apply weights, and break down the sample by both broad ethnic group and age group, I end up with a small N in many cells (ie: < 10) and in some cases 0 counts for a cell and so I am ending up with very large confidence intervals. Having read the general User Guide and the Ethnicity and Immigration User Guide, my understanding is that one of the reasons the ethnic minority boost samples were included was to try to tackle this issue of small sample sizes in marginalised subgroups, so I wanted to check that I am not missing something. I think my main issue is that once I apply a weight (such as j_indinui_xw), the N for the ethnic minorities is being scaled down in order account for the oversampling of these groups, and since the weighted N (rather than the unweighted N) is being used to calculate confidence intervals by R, I end up with very large confidence intervals.</p>
<p>As a quick illustration: I'm using the j_indresp.dta file and have recoded the j_ethn_dv variable into Asian, Black, Other, White. In the Black category, there are 1314 respondents, but when I apply the weight j_indinui_xw and tabulate by ethnicity using the svytable function in R, this is scaled down to only 458.3841. If I then break this down by 10-year age group and my health outcome variable, I end up with very small Ns (or zeros) which means when using a function such as svyciprop in R, I get very large confidence intervals.</p>
<p>This may simply be an unavoidable problem, but given that Understanding Society has put lots of effort into including these extra ethnic minority samples, I wanted to make sure I was making best use of them. And just to double check - I can simply use the normal indresp file in order to draw on this ethnic minority boost sample?</p>
<p>Many thanks for your help and the amazing resources you provide!</p>
<p>Best wishes,<br />Laurence Rowley-Abel</p> Support #1731 (Resolved): Interview Time for Main Survey Wave 10https://iserredex.essex.ac.uk/support/issues/17312022-07-22T10:51:56ZClaire Wu
<p>Dear support team,</p>
<p>I am conducting a study comparing the timing of housework before and after the Covid-19 outbreak, and I intend to run two separate models for these two points in time. For the data prior to the Covid-19 outbreak, I intend to choose wave 10 of the main survey. However, I am confused about the timing of the fieldwork for wave 10 - does it end before the Covid-19 outbreak or until May 2021? I checked the survey timeline webpage: <a class="external" href="https://www.understandingsociety.ac.uk/documentation/mainstage/survey-timeline">https://www.understandingsociety.ac.uk/documentation/mainstage/survey-timeline</a> which shows wave 10 between January 2018 and May 2020, but when I check the user guide it seems it already ends in 2019?</p>
<p>Could someone please help me with this? Any comments would be greatly appreciated.</p>
<p>Claire</p> Support #1700 (Resolved): Interview timinghttps://iserredex.essex.ac.uk/support/issues/17002022-05-13T09:40:34ZNils Braakmann
<p>Dear Understanding Society team,</p>
<p>I was hoping you might be able to help me with a question: We are currently studying the effect of local economic shocks on various individual outcomes. We use the exact timing of the interviews to distinguish between individuals interviewed before and after the shocks. Previous papers (e.g., Powdthavee, N., Plagnol, A.C., Fritjers, P. and Clark, A.E. (2019). Who got the Brexit blues? The effect of Brexit on subjective wellbeing in the UK. Economica 86: 471-494. or Braakmann, N. (2021). Immigration status uncertainty and mental health – Evidence from Brexit. Industrial Relations: A Journal of Economy and Society 60: 521- 548.) have relied on the argument that the timing of the interview is essentially orthogonal to the respective shock (Brexit in the two cases above). I am now currently dealing with a reviewer who has pointed out that "In Germany, there are for instance studies that show that timing of interview correlates with the socio-economic background of the respondents.” This strikes me as a valid point - if there is selectivity on such characteristics and if these characteristics correlate with whatever shock is being looked at, this could indeed cause bias. Before I go and do 20 tests on interview timings and various respondent characteristics, I was wondering whether you have any information about any selectivity or how interviews are timed that you might be willing to share.</p>
<p>All best and thank you in advance,<br />Nils</p> Support #1693 (Resolved): household datahttps://iserredex.essex.ac.uk/support/issues/16932022-05-03T11:33:51ZThanos Verousis
<p>Hello<br />Can you please confirm the following: if a household consists of two members eg a married couple, does the survey contain responses from both members? For example, can we get the cognitive ability measures for both partners?<br />Thanks<br />Thanos</p> Support #1653 (Resolved): Representativeness of lone mothershttps://iserredex.essex.ac.uk/support/issues/16532022-02-07T10:28:37ZTom Waters
<p>Hi,</p>
<p>I have been comparing the UKHLS data to data from the FRS and LFS. For reasons not clear to me, the UKHLS data seem to be very significantly different when looking at lone mothers.</p>
<p>I attach two .do files which calculate the share of lone mothers that are homeowners, and the age that lone mothers finished education, in both the FRS and UKHLS (for 2009-10 data in FRS, and wave A in UKHLS). It should run if you simply change the "path" global at the top to specify where FRS/UKHLS data are saved.</p>
<p>The results are very different. In the FRS, 16% of lone mothers finished full time education aged 19 or older, and 35% own their own home. In UKHLS those figures are 32% and 47% respectively.</p>
I have tried a number of variations:
<ul>
<li>These differences hold in other years/waves too</li>
<li>The differences persist if you do not weight the data</li>
<li>The LFS gives figures very similar to FRS, not UKHLS</li>
<li>If you look at the whole population (rather than just lone mothers), the two surveys are pretty similar for homeownership, but there is still a much higher rate of leaving education at 19 or later in UKHLS</li>
<li>Similar differences seem to hold for lone mothers for other outcomes, including employment (higher in UKHLS) and the take-up of out-of-work benefits (lower in UKHLS)</li>
</ul>
<p>Very grateful for any guidance in what might be going on here.</p> Support #1645 (Resolved): How to Deal with Missingness in Health Variables Over Timehttps://iserredex.essex.ac.uk/support/issues/16452022-01-31T15:50:45ZSue Easton
<p>Hi,</p>
<p>How to Deal with Missingness in Health Variables Over Time</p>
<p>I'm looking t health variables in UKHLS waves 1-11, and may individuals have incomplete data across the full timespan.</p>
<p>I'm wondering whether "missingness" across time/waves in UKHLS is generally considered MCAR/MAR/NMAR, as in this case it may indeed be linked to episodes of illness, and how other researchers have dealt with the issue of missing data?</p>
<p>I am dubious about imputing values using multiple imputation when the values I would use to impute are the same ones I will be entering into a multilevel analysis of health outcome.</p>
<p>Any advice or references on how to deal the missingness across time/waves gratefully received!</p>
<p>Thanks,</p>
<p>Sue E.</p>