Understanding Society User Support: Issueshttps://iserredex.essex.ac.uk/support/https://iserredex.essex.ac.uk/support/support/favicon.ico?15995719382024-01-22T12:50:29ZUnderstanding Society User Support
Redmine Understanding Society User Support - Support #2036 (Feedback): Understanding Society - weightshttps://iserredex.essex.ac.uk/support/issues/20362024-01-22T12:50:29ZValentina Di Iasio
<p>Good morning,</p>
<p>After reading the user guide and watch the short YouTube video, I am still confused on which are the correct weights I should select for my pooled cross-sectional analysis using Understanding Society.</p>
<p>I am using waves 6 and 9 for a pooled cross-section analysis. I would therefore being inclined in using the cross-sectional weights. However, when reading the user guide it says that cross-sectional weights should only be used when the analysis includes one wave only. I also read the paragraph on re-scaling the weights to use more waves to conduct cross-sectional analysis. However, I am not sure whether the described procedure would apply to my case since I don't have a year overlapping over the two waves (wave 6 goes from January 2014 to May 2016 while wave 9 goes from January 2017 to May 2019). Therefore I am not sure whether I should simply use cross-sectional weights, re-scale the cross-sectional weights somehow (maybe for the first 6 months of 2016 and 2019 only?), or exclude the first 6 months of the years 2016 and 2019. Or, if I am missing something and I should use longitudinal weights (in that case, since I am doing a pooled cross-section analysis, how should I deal with 0 weights?)</p>
<p>Thank you in advance</p>
<p>Valentina Di Iasio</p> Understanding Society User Support - Support #1894 (Resolved): Weight for unbalanced and merged U...https://iserredex.essex.ac.uk/support/issues/18942023-04-21T14:30:20ZYanan Zhangzhangyanan0918@gmail.com
<p>Dear Sir/Madam,</p>
<p>I hope this message finds you in good health and high spirits.</p>
<p>I am currently working with individual-level data from the merged Waves 1-18 of the BHPS and Waves 1-8 of the UKHLS datasets. I have a couple of questions regarding the use of weights in my analysis. I would appreciate any guidance you could provide.</p>
<p>1. In my study, I am employing fixed effects estimates to analyze the relationship between two variables, x and y. Given this approach, is it necessary to apply weights to the analysis?</p>
<p>2. I have followed the guidelines and used the longitudinal weight provided in Wave 8 of the UKHLS. However, I understand that this weight is applicable only to those who have participated in all waves. Since many individuals have only participated in parts of the waves, I am unsure how to generate weights for these participants. Could you please advise on the appropriate way to handle this situation?</p>
<p>Thanks for your time!</p> Understanding Society User Support - Support #1872 (Resolved): Pooling waves for longitudinal ana...https://iserredex.essex.ac.uk/support/issues/18722023-03-02T16:13:58ZSpencer Thompson
<p>Hello,</p>
<p>I am doing some analysis on informal care. I want to track people's incomes for several years after they start doing care work.</p>
<p>So far I have used a single wave to set the cohort, e.g. I look at people who start care in wave 4 (t) and see what happens to their income in wave 5 (t+1), wave 6 (t+2), etc. However, I'd like to work with a larger sample and not rely on a single year. For example, I would pool people who start care in waves 4, 5, or 6, and treat all of these as t. For these respective cohorts, t+1 would then be waves 5, 6, 7, and so on.</p>
<p>I have seen the documentation on adjusting weights in order to pool waves for cross-sectional analysis. How can I do this for my longitudinal analysis?</p>
<p>Many thanks.</p> Understanding Society User Support - Support #1868 (Resolved): Use of weights for analysing job q...https://iserredex.essex.ac.uk/support/issues/18682023-02-25T16:00:04ZThomas Stephenst.c.stephens@lse.ac.uk
<p>Good afternoon,</p>
<p>I have a few questions about the weights to use for some analysis of job quality which I'm carrying out using Understanding Society. I have read another very useful support response on this (see: <a class="external" href="https://iserredex.essex.ac.uk/support/issues/1739">https://iserredex.essex.ac.uk/support/issues/1739</a>), but this still gives rise to some further questions.</p>
I'm planning on carrying out two distinct types of analysis for my research. Although working conditions data is available in every other wave, note that I exclude wave 2 from my analysis, for reasons I expand on below:
<ul>
<li><strong>Descriptive statistics of changes over time across every other wave,</strong> ie comparing Wave 4 vs. 6 vs. 8 vs. 10...;</li>
</ul>
<ul>
<li><strong>Analysis of pooled data from these waves</strong>, to understand the relationship between job quality and various other individual and household characteristics across all of Waves 4, 6, 8, 10.</li>
</ul>
<p>I have the following questions about which weights to use, and the weighting process in general:</p>
<p><strong>1.</strong> Will I have to use two different weights for these two types of analysis? My understanding is that the existing indinub_xw weight (ie just removing the wave prefix) would suffice for the first type of analysis (as per my reading of the Harmonised BHPS user guide, p. 25, <br /><a class="external" href="https://www.understandingsociety.ac.uk/sites/default/files/downloads/documentation/mainstage/user-guides/bhps-harmonised-user-guide.pdf">https://www.understandingsociety.ac.uk/sites/default/files/downloads/documentation/mainstage/user-guides/bhps-harmonised-user-guide.pdf</a>), but that I will have to do weight rescaling for the pooled analysis to avoid under-representing respondents from later waves. Is this correct?</p>
<p><strong>2.</strong> Although I only analyse at every other wave, I have created some new indicators by looking back at data from the wave immediately prior to the respondent's wave (eg I use Wave 3 data to establish whether respondents in Wave 4 have been continuously employed for >1 wave or >2 waves, wave 5 for wave 6, wave 7 for wave 8, etc...). Does this have any bearing on the weight I should rescale to for the pooled analysis?</p>
<p><strong>3.</strong> For the above reason, I exclude wave 2 data, as the relevant questions weren't asked in wave 2. Am I correct in assuming that if my pool starts at Wave 4, this means I need to re-scale to Wave 4 rather than Wave 2, using a variation (albeit in R rather than Stata, as that's what I'm using...) of the code you give here: <a class="external" href="https://iserredex.essex.ac.uk/support/issues/1739">https://iserredex.essex.ac.uk/support/issues/1739</a>? Are there any other issues I need to be aware of?</p>
<p><strong>4.</strong> I won't be analysing changes based on calendar years; I'll be keeping respondents in their waves. My reading is that I therefore don't have to carry out the adjustments you outline in p. 10 of your weighting FAQs: <a class="external" href="https://www.understandingsociety.ac.uk/sites/default/files/downloads/documentation/user-guides/mainstage/weighting_faqs.pdf">https://www.understandingsociety.ac.uk/sites/default/files/downloads/documentation/user-guides/mainstage/weighting_faqs.pdf</a>. Is this correct?</p>
<p><strong>5.</strong> I haven't seen any discussion of seasonality in the user forum or weighting FAQs. Ie if one wave happens to over-represent people interviewed in later seasons where labour market statistics might be different. Could I check whether your weights account for this?</p>
<p>Many thanks in anticipation.</p>
<p>Best wishes,</p>
<p>Tom</p> Understanding Society User Support - Support #1865 (Resolved): Changes to USOC wave data download...https://iserredex.essex.ac.uk/support/issues/18652023-02-23T16:42:31ZWilliam Shufflebottom
<p>Hi,</p>
<p>QUESTIONS</p>
<p>Q1: indscub_xw weight from wave 6 of USOC is present in our historical download of the wave 6 data but appears to be missing in the version of wave 6 we downloaded from UKData Service a few months ago and is also not listed as being in wave 6 on the USOC variable search page - can we confirm why only the indscui_xw weight is in the latest Wave 6 version, confirm it was in the original release, and if/when (and if so why) it was removed?</p>
<p>Q2: Our estimates run on the latest download of wave 1 to 12 of USOC are producing different numbers from the estimates we ran at the time of the previous wave's releases. Has there been a change to the data or weights (beyond wave 6 having a different weight) or how the weights work that could explain the difference we are seeing for all waves (bar wave 1 and wave 12) in a recent download of the data from all the waves. We are using the same weight (bar wave 6) and the same variable (sclfsat_7 in this case - but we use a range of USOC variables in our analysis).</p>
<p>BACKGROUND</p>
<p>We are producing estimates for the OECD and just discovered some differences for the estimates and CIs for the sclfsat7 variable when we re-ran historical estimates for all USOC waves 1 to 12. We run breakdowns for this variable (and others) by various domains when we update our publications and a new USOC wave has been released so we have the estimates from previous runs made at the time of USOC wave data release. We only ran the sclfsat7 variable again recently so there may be other changes.</p>
<p>We have a document for the weights to use for each variable which states that the indscub_xw weight is the correct weight to use for the sclfsat_7 variable in wave 6 but we noticed it was "missing" in the wave 6 data we downloaded around November from UK Data service (instead indscui_xw is present). As we are getting differences in our estimates and CIs for all waves (bar wave 1 and 12), this has prompted us to check with you if there have been changes made to the versions of the USOC main study wave data currently on the UK Data Service compared to what would have been available at the time each wave's data was released which could explain the differences we are seeing.</p>
<p>Your help is greatly appreciated as this has the potential to impact a lot of our publications and the current ad hoc we are working on</p> Understanding Society User Support - Support #1864 (Resolved): Using survey weights in longitudin...https://iserredex.essex.ac.uk/support/issues/18642023-02-23T15:34:05ZTanya Braune
<p>I am using Waves 7, 9 and 11 to conduct longitudinal analyses on changes in fruit and vegetable intake with age. I am using a multilevel (mixed effects) model where the measurement wave is level 1, individual (pidp) is level 2 and psu is level 3. I have looked through all the documentation and the other support cases (particularly Support <a class="issue tracker-3 status-3 priority-5 priority-high2" title="Support: Weighting in multilevel modelling in R Studio (Resolved)" href="https://iserredex.essex.ac.uk/support/issues/1572">#1572</a> ) but I have not found a solution on how to apply the survey weights to my model. I was advised to use the 'g_indscui_lw' weight for this analysis. The online documentation advises to use the 'survey' package in R but I am struggling to properly apply this to my multilevel model as the resources provided do not go into detail about this.</p>
<p>Any recommendations or advice you have on this would be greatly appreciated. <br />Many thanks<br />Tanya</p> Understanding Society User Support - Support #1852 (Resolved): Select the correct weighting valueshttps://iserredex.essex.ac.uk/support/issues/18522023-02-07T17:53:18ZYushi Bai
<p>Dear colleagues,</p>
<p>I'm a post-doc research associate at the University of Manchester. We're currently planning an analysis investigating how mental health problems spread within a family network using your data (thank you for providing such an excellent dataset!). However, we're confused about how to create the correct weighting on our data even after reading all the tutorial materials. So I sincerely hope we can have your support for our analysis. I will first brief you on our initial analytical plan:</p>
<p>1. Formulate an initial participant pool consisting of all data in waves 1, 3, 5, 7, 9, and 11, because the Strengths and Difficulties Questionnaire (SDQ) data are available in those waves.<br />2. Within this initial pool, compare the data quality for each family across the waves (e.g. compare the quality of SDQ data for family A in waves 1, 3, 5, 7, 9, and 11).<br />3. Select a particular dataset for each family if the dataset has the fewest missing values across the waves, and formulate a large cross-sectional dataset. For example, if SDQ data have the fewest missing values for family A in wave 1, and for family B in wave 3, we use data for family A from wave 1, and data for family B from wave 3 to formulate a cross-sectional dataset.</p>
<p>By doing so, we hope we can boost our sample size and the quality of the data. This is because our analytical approach (network analysis) requires highly on data quality. However, we're aware that this participant selection approach may introduce bias. Therefore, we're wondering whether you can suggest whether our participant selection plan is reasonable in the light of your research design, and if so, what materials we can use to create the correct weighting values for our data?</p>
<p>Thank you in advance for your time and help, and we're looking forward to hearing from you.</p>
<p>Kind regards,<br />Yushi</p> Understanding Society User Support - Support #1715 (Resolved): Longitudinal Weighting of Non-Move...https://iserredex.essex.ac.uk/support/issues/17152022-06-13T11:06:10ZSue Easton
<p>Hi, I have searched and can't find these key words in any posts.</p>
<p>Due to limitations of time I need to limit my analysis to individuals who have not changed location since they entered the survey in Wave 1 (UKHLS sample and any others in from Wave 1 with more than 1 wave).</p>
<p>This means some people's data will be right censored due to household moves.</p>
<p>How will this affect weighting?</p>
<p>Will I need to calculate new weights? As variables such as age are highly likely to be correlated with the "risk" of moving home.</p>
<p>Thanks.</p>
<p>Sue EAston</p> Understanding Society User Support - Support #1705 (Resolved): Longitudinal Weighting of UKHLS Da...https://iserredex.essex.ac.uk/support/issues/17052022-05-24T13:02:37ZSue Easton
<p>Hi,</p>
<p>1. Are there any reference documents on how/where to enter the UKHLS variables: PSU, strata and k_indinus_lw in order to weight the survey data for multilevel modelling in MLwiN please?</p>
<p>A link to a simple "how to" guide with examples would be really useful if possible please (due to limitations of time). Thanks.</p>
<p>Also: <br />2. If the longitudinal weight k_indinus_lw excludes GPS individuals who have missed waves over time in the series a-k (1-11), is there any point trying to "backfill"/impute data for their missed waves?</p>
<p>Thanks!</p>
<p>Sue Easton</p> Understanding Society User Support - Support #1698 (Resolved): Weights for Waves 1-11 (CS and Lon...https://iserredex.essex.ac.uk/support/issues/16982022-05-11T16:08:39ZRowan Isaaks
<p>Hi Support Team,</p>
<p>I am doing two types analysis: One is a repeated cross-section of households and the other is a panel of adults. I want to use waves A-K of UKHLS for both of these. I identified hhdenub_xw and psnenus_lw as the respective weight variables I should be using, but these are not available for wave A. My intuition is that this is because of the introduction of the BHPS sample in wave 2 (is this correct?). Is there a way to derive values of hhdenub_xw and psnenus_lw for wave A, or are there alternative weights that would allow me to use all 11 waves?</p>
<p>Thanks for your time</p> Understanding Society User Support - Support #1696 (Resolved): random effects logistic regression...https://iserredex.essex.ac.uk/support/issues/16962022-05-09T12:57:51ZZohra Ansari-Thomas
<p>Hello,</p>
<p>I am attempting to run a random effects logistic regression model using waves 1-10 of the UKHLS, and am running into some issues with how to take into account the longitudinal weighting, strata, psu, as well as clustering by PIDP or allowing for random intercepts by PIDP to account for the longitudinal design of the study. I am using Stata</p>
<p>I can svy set my data to account for the longitudinal weights (indinus_lw), the psu, and the strata, but I am not sure how to account for the clustering by PIDP. I am using the svy: melogit command.</p>
<p>Any advice would be much appreciated, thank you!</p> Understanding Society User Support - Support #1667 (Resolved): Youth self completion longitudinal...https://iserredex.essex.ac.uk/support/issues/16672022-03-14T17:20:00Zjennie parnham
<p>Hello,</p>
<p>I was looking to check if I have correctly understood the weight I needed for my analysis .</p>
<p>My analysis is a longitudinal using the youth self-completion data from waves 7-11, following them into young-adults, if applicable. However, I do not need the participant to have participated in every wave between 7-11, I just need them to have participated in at least two waves, it doesn't matter which.</p>
<p>I have read the responses to the following similar queries, and wanted to check if what I have understood is correct.<br /><a class="external" href="https://iserredex.essex.ac.uk/support/issues/1091">https://iserredex.essex.ac.uk/support/issues/1091</a> <br /><a class="external" href="https://iserredex.essex.ac.uk/support/issues/1323">https://iserredex.essex.ac.uk/support/issues/1323</a><br /><a class="external" href="https://iserredex.essex.ac.uk/support/issues/1585">https://iserredex.essex.ac.uk/support/issues/1585</a></p>
<p>Would it be correct to use the last applicable sub-optimal weight for the wave that that individual participated in, and make a weight which is a combination of these?</p>
<p>For example:<br />IF last wave of data collection is Wave 11 (Young adult) then weight = k_indscui_lw<br />IF last wave of data collection is Wave 11 (Youth) then weight = k_psnenui_lw<br />IF last wave of data collection is Wave 10 (Young adult) then weight = j_indscui_lw <br />IF last wave of data collection is Wave 10 (Youth) then weight = j_psnenui_lw <br />IF last wave of data collection is Wave 9 (Young adult) then weight = i_indscui_lw (and so on ...)</p>
<p>So the weight variable that I use in the analysis is a combination of different weights, specific to their age at the last wave of participation. Or is it incorrect to take from different weight variables in this way?</p>
<p>I am concerned that if i just use the longitudinal enumeration weight for the last wave in the analysis (wave 11), I will end up excluding many participants.</p>
<p>Many thanks for your help</p> Understanding Society User Support - Support #503 (Closed): Inconsistencies in self-completion mo...https://iserredex.essex.ac.uk/support/issues/5032016-02-15T14:34:25ZTill Hoffmann
<p>In wave three of the Understanding Society survey, there are six entries for respondents who have refused the self-completion part of the interview but have positive self-completion interview weights. In particular, c_csac is 4 (refused) or 5 (not able to complete) (see <a class="external" href="https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/3/datafile/c_indresp/variable/c_scac">https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/3/datafile/c_indresp/variable/c_scac</a>) but c_indscub_xw is nonzero (see <a class="external" href="https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/3/datafile/c_indresp/variable/c_indscub_xw">https://www.understandingsociety.ac.uk/documentation/mainstage/dataset-documentation/wave/3/datafile/c_indresp/variable/c_indscub_xw</a>).</p>
<p>Similarly, there are three entries indicating that the self-completion part of the interview was refused but the response to the "What is the sex of your first friend?" question contains valid data even though the question is in the self-completion module.</p>
<p>Am I missing something?</p> Understanding Society User Support - Support #448 (Closed): weightshttps://iserredex.essex.ac.uk/support/issues/4482015-11-16T00:17:28ZVernon Hedgevernonhedge@hotmail.co.uk
<p>I am looking at data exclusively at Wave C Understanding Society c_indresp.sav. I am planning to employ model based inference which may (as needs be) incorporate weight, strata and PSU into the model.
<p>I am having difficulty finding out how the weights were computed. I was hoping to use include the variables by which the weights were calculated within the model and specify PSU as level 2 random effects. I just cannot seem to find how the weights were calculated from Understanding Society documentation.</p>
</p>
<p>All the variables are from the c_indresp file. The 12 are listed here as name, “label”, [position number in variable view of c_indresp.sav]</p>
<p>c_sex_cr “sex (corrected)” [2292],<br />c_age_cr “age (corrected)” [2294],<br />c_birthy “year of birth” [2771], <br />c_big5c_dv “Conscientiousness” [2896],<br />c_big5o_dv “Openness” [2899],<br />c_hiqual_dv “Highest qualification” [2904], <br />c_gwri_dv “Cognitive ability: Immediate word recall: Number of correct items” [2915], <br />c_cgvfc_dv “Cognitive ability: Verbal fluency: Count of correct answers” [2932],<br />c_cgna_dv “Cognitive ability: Numeric ability: Count of items answered correctly”[2935], <br />c_jbnssec8_dv “Current job: Eight Class NS-SEC” [2947],</p>
<p>I am also having difficulty identifying which weight variable would be most appropriate to my analysis according to the w_xxxyyzz_aa scheme (p67 of the User Manual).</p>
<p>I can fill in this much c_indyyzz_xw – i.e., I know I am dealing with wave c only (so c_ and xw) and only with adult (16+) respondents (so ind).</p>
<p>I have identified 4 weight variables relevant to a cross-sectional design in the c_indresp file,</p>
<p>1. c_indpxub_xw “combined cross-sectional adult main or proxy interview weight” [3002], <br />2. c_indinub_xw “combined cross-sectional adult main interview weight” [3003], <br />3. c_indscub_xw “combined cross-sectional adult self-completion interview weight” [3004], <br />4. c_ind5mus_xw “cross-sectional extra 5 minute interview person weight” [3005].</p>
<p>The yy component must be either px, in, sc, or 5m. I think I can exclude 5m, as none of the variables on my list is on the list on Table 25 (p56) of the User Manual. Likewise, viewing Table 24 (p53), I think sc can be excluded.</p>
<p>As for the zz component it is tempting to just use "us" (for “understanding society”?). The user guide advises me that the "us" designation refers to “GPS [General Population Sample] and EMB samples” – is this what is meant by “Mainstage”?</p>
<p>Looking at the “Levels of Analysis” in Table 28 (p62), I think I can exclude level 4 “Adult or youth self-completion”. I cannot, however seem to find information on whether the c_indresp variables I am using are level 3 “Adult proxy and main interview” or level 2 “Adult main interview only (no proxy)”. Using the Understanding Society website to search each variable name they all return “Mainstage Variable”. I cannot tell from this which level of 1 to 4 is the most appropriate to select a weighting variable.</p>
<p>So the two problems I have are 1) identifying which variables were used to calculate survey weights and 2) identifying which ”xw“ survey weight variable is most appropriate to my analysis.</p>
<p>I would be enormously grateful for any clarification.</p> Understanding Society User Support - Support #253 (Closed): general population samplehttps://iserredex.essex.ac.uk/support/issues/2532014-03-28T16:09:03Zpeter tammes
<p>Dear sir /madam,<br />We would like to use only the General Population Comparison sample. Which of the weight variables should we use in our analysis?<br />Thank you <br />Peter</p>