Understanding Society User Support: Issueshttps://iserredex.essex.ac.uk/support/https://iserredex.essex.ac.uk/support/support/favicon.ico?15995719382023-02-23T12:30:19ZUnderstanding Society User Support
Redmine Understanding Society User Support - Support #1863 (Resolved): coding health conditionshttps://iserredex.essex.ac.uk/support/issues/18632023-02-23T12:30:19ZKaren Arulsamy
<p>Hi there,<br />I am analyzing data from Waves 7,9 and 11 of UKHLS and would like to create a variable to denote if someone has a health condition. I tried the following, assuming for now all the missing values are given a value of 0. However, the values for wave 11 are odd. This may be due to the changes in the questions on health conditions in Wave 10. Which variables should I use to adequately capture individuals who have a health condition across these waves? If this is not possible, then which variables can I use to capture individuals who ever had a health condition across these waves?</p>
<p>gen hcond_main = 0</p>
<p>replace hcond_main = 1 if hconds01 1 | hcondn1 1 <br />replace hcond_main = 1 if hconds02 1 | hcondn2 1 <br />replace hcond_main = 1 if hconds03 1 | hcondn3 1 <br />replace hcond_main = 1 if hconds04 1 | hcondn4 1 <br />replace hcond_main = 1 if hconds05 1 | hcondn5 1 <br />replace hcond_main = 1 if hconds06 1 | hcondn6 1 <br />replace hcond_main = 1 if hconds07 1 | hcondn7 1 <br />replace hcond_main = 1 if hconds08 1 | hcondn8 1 <br />replace hcond_main = 1 if hconds09 1 | hcondn9 1 <br />replace hcond_main = 1 if hconds10 1 | hcondn10 1 <br />replace hcond_main = 1 if hconds11 1 | hcondn11 1 <br />replace hcond_main = 1 if hconds12 1 | hcondn12 1 <br />replace hcond_main = 1 if hconds13 1 | hcondn13 1 <br />replace hcond_main = 1 if hconds14 1 | hcondn14 1 <br />replace hcond_main = 1 if hconds15 1 | hcondn15 1 <br />replace hcond_main = 1 if hconds16 1 | hcondn16 1</p>
<p>by pidp (wave), sort: replace hcond_main = hcond_main[_n-1] if missing(hcond_main)<br />by pidp (wave), sort: replace hcond_main = hcond_main[_n+1] if missing(hcond_main)</p>
<p>Thanks in advance for your help.</p>
<p>Best wishes<br />Karen</p> Understanding Society User Support - Support #1807 (Resolved): hadsymp and longcovidhttps://iserredex.essex.ac.uk/support/issues/18072022-11-17T13:28:32ZKaren Arulsamy
<p>Hi there, <br />I am trying to create a variable based on longcovid that matches the number of observations in hadsymp excluding individuals residing in Wales who responded in Welsh.<br />My understanding is that everyone who reported having symptoms of covid would have answered the longcovid question. However, based on the dictionary, the question is asked "...if previously reported having covid symptoms, and not reported having symptoms in current survey". Are individuals asked this question only once if they reported symptoms in the last survey or are they followed up in every survey? My goal is to create a new variable for longcovid where 0 covers individuals who never reported having covid symptoms and 1 if they report long-term symptoms. I attempted a version below (not excluding Welsh responses yet but I doubt this would account for a substantial number of observations) and would appreciate advice on how to properly do this.</p>
<p>gen covid = . <br />replace covid = 1 if hadsymp 1<br />replace covid = 0 if hadsymp 2</p>
<p>tab covid</p>
<pre><code>covid | Freq. Percent Cum.<br />------------+-----------------------------------<br /> 0 | 115,704 94.03 94.03<br /> 1 | 7,343 5.97 100.00<br />------------+-----------------------------------<br /> Total | 123,047 100.00</code></pre>
<p>egen anycovid = max(covid 1), by(pidp)<br />replace anycovid = . if covid .</p>
<p>tab anycovid </p>
<pre><code>anycovid | Freq. Percent Cum.<br />------------+-----------------------------------<br /> 0 | 89,451 72.70 72.70<br /> 1 | 33,596 27.30 100.00<br />------------+-----------------------------------<br /> Total | 123,047 100.00</code></pre>
<p>*1 is the number of observations that takes the value 1 at least once i.e, the individual reports covid in at least <br />*one wave</p>
<p>gen longcovid_main = . <br />replace longcovid_main = 1 if longcovid 1 <br />replace longcovid_main = 0 if longcovid 2<br />replace longcovid_main = 0 if anycovid == 0</p>
<p>tab longcovid_main</p>
<p>longcovid_m |<br /> ain | Freq. Percent Cum.<br />------------+-----------------------------------<br /> 0 | 90,113 92.34 92.34<br /> 1 | 7,474 7.66 100.00<br />------------+-----------------------------------<br /> Total | 97,587 100.00</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1754 (Resolved): advice on creating a variable for ...https://iserredex.essex.ac.uk/support/issues/17542022-08-19T11:57:47ZKaren Arulsamy
<p>Hi there,<br />Can I please get some advice on how best to create a variable to denote the number of children in a household that is consistent across the main and covid surveys? In the Covid-19 surveys, the number of children variables (num1217, num418 and num511) are provided by child age-groups but these do not cover children below age 5. <br />If this is too complicated, presence of children would also be helpful.</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1746 (Resolved): reconciling household income betwe...https://iserredex.essex.ac.uk/support/issues/17462022-08-10T13:10:22ZKaren Arulsamy
<p>Hi there,</p>
<p>Can I please get some advice on how to accurately compare household income in the main survey vs. the covid-19 survey?</p>
<p>As I understand, household income is collected differently in the main survey vs. the covid-19 survey. As household income is reported based on different periods in the covid-19 survey, I've first accounted for these periods to obtain a monthly value as follows:</p>
<p>gen covid_hhincome = .<br />replace covid_hhincome = hhincome_amount if hhincome_period 3<br />replace covid_hhincome = hhincome_amount*4 if hhincome_period 1<br />replace covid_hhincome = hhincome_amount*2 if hhincome_period 2<br />replace covid_hhincome = hhincome_amount/12 if hhincome_period 4<br />replace covid_hhincome = hhincome_amount*4 if hhincome_period 5<br />replace covid_hhincome = . if hhincome_amount -9 | hhincome_amount -8 | hhincome_amount -7 | hhincome_amount -2 | hhincome_amount -1</p>
<p>I then use fihhmnnet1_dv from the main survey combined with the above variable to form a total household income variable. Am I using the right income variable from the main survey?</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1745 (Resolved): the equivalent of jbstat in the co...https://iserredex.essex.ac.uk/support/issues/17452022-08-09T13:16:32ZKaren Arulsamy
<p>Hi there,<br />Is there an equivalent of the variable 'jbstat' (in the main survey) in the covid survey? I've looked through the Covid survey data and questionnaires but can't seem to find it. Can I please get some advice on how to put a variable like a jbstat together based on variables from the Covid survey? Importantly, what would be the best way to identify cases of unemployment and those out of the labour force?</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1741 (Resolved): Weights - Main Survey Wave 9 & Cov...https://iserredex.essex.ac.uk/support/issues/17412022-08-04T14:38:51ZKaren Arulsamy
<p>Hi there,</p>
<p>I've combined the main & covid-19 waves in long-format. My goal is to understand changes over time comparing data from Wave 9, July 2020 (Wave 15) and January 2021 (Wave 18). The outcome variables of interest are only captured in these Covid-19 Waves.</p>
<p>Is it appropriate to use the following weights to look at trends over time? This is still quite early stage but the goal finally is to use the change in these outcomes, say from W9 to July 2020, as predictors of change in another outcome and including individual fixed effects. However, using betaindin_lw drops the Wave 9 observations. Based on the User Guide, the Covid-19 survey weights are modeled on Wave 10 but incorporates weight sharing to apply to respondents without Wave 10 weights. Do I need to create a tailored longitudinal weight if I'm using Wave 9 as the 'baseline' wave?</p>
<p>Also, what's the best way to incorporate survey weights if I restrict the analyses to Scotland?</p>
<p>svyset [pweight = indinui_lw]<br />svy: tab XX if wave == 9, obs</p>
<p>svyset [pweight = betaindin_lw]<br />svy: tab XX if wave 15, obs</p>
<p>svy: tab XX if wave 18, obs</p>
<p>Thanks in advance for your help and please let me know if you need more information to support me with this.</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1439 (Resolved): jbpenm in BHPS & UKHLShttps://iserredex.essex.ac.uk/support/issues/14392020-11-16T11:21:37ZKaren Arulsamy
<p>Hi,</p>
<p>I am working on a project to potentially combine BHPS + UKHLS datasets with a specific interest in the variable jbpenm.</p>
<p>In the BHPS dataset, there is quite a large decline in sample size for the years 1992-1994 for those who respond "yes" which does not track with overall pension participation trends in the UK. It also doesn't track with the data 1994 onwards - this type of sudden decline is uncharacteristic of pension participation which tends to be stable over time albeit declining slowly over time beginning in the 2000s. I have checked the questionnaires to see if there was change in wording or routing which doesn't seem to be the case. I attach a document contain tabulation results for jbpen as a whole and by specific filters.</p>
<p>I was wondering if you might have some insights into what happened with this particular variable during these years?</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1319 (Resolved): clarification on jbpen variable in...https://iserredex.essex.ac.uk/support/issues/13192020-03-02T17:10:42ZKaren Arulsamy
<p>Hi,</p>
<p>The question for this variable jbpen was asked to those who satisfied jbsemp = 1 (currently an employee of a firm).<br />There are 57,712 responses coded as inapplicable in the dataset. Initially, I assumed that these were for individuals who were self-employed (ie. those who do not satisfy jbsemp = 1) but the numbers don't quite add up. There are only 25,482 responses for those who report being self-employed.</p>
<p>Can I check what these 57,712 responses capture exactly? Would they also include people who are currently unemployed? Which would be the right unemployment variable for me to use to check if this is the case?</p>
<p>Many thanks<br />Karen</p> Understanding Society User Support - Support #1223 (Resolved): Department for Work and Pensions (...https://iserredex.essex.ac.uk/support/issues/12232019-07-31T14:21:18ZKaren Arulsamy
<p>Hi, <br />Can I check when the above dataset (as mentioned here <a class="external" href="https://www.understandingsociety.ac.uk/documentation/linked-data/data-linkage-phases">https://www.understandingsociety.ac.uk/documentation/linked-data/data-linkage-phases</a>) is scheduled for release and if there are any specific access requirements? <br />Previous linked datasets require the researcher to be in a UK institution and I wonder if the same applies to this linked dataset. Any further information on the specific scope/variables that will be linked in will be much appreciated. Thanks in advance for your help!</p> Understanding Society User Support - Support #1222 (Resolved): Self-reported employer size variab...https://iserredex.essex.ac.uk/support/issues/12222019-07-31T14:09:54ZKaren Arulsamy
<p>Hi, <br />I'm running analyses using the above variable in Understanding Society (US). However, I am seeing discrepancies in the distribution of employer size in US data as opposed to employer provided sources such as the Business Population Estimates (BEIS). It seems that US data reports a higher proportion of individuals in the smaller companies (those with less than 50 employees). It could be that individuals who work for branches of larger companies report the number of employees in those specific branches rather than the company as a whole. I wonder if this is true based on previous experience of researchers using this variable? Is there any linked datasets with US that I could access that would provide the employer provided number of employees in the company? Any pointers would be much appreciated.</p>