On 03-Sep-09 12:21:54, Alex Scotton wrote:
When your writing a reply below the "To:" box next to the formating buttons is a link "« Plain Text" click it and it should just send plain text emails. At least thats been my understanding.
On the web, yes.. however on the GMail app for the Google phone, I can't find the option and in fact now I'm pretty sure it doesn't exist at all.. So yet again forced to hit my PC to mail you guys...
Could we do this the lazy way and enable HTML emails on the list ha ha.. why out of interest are they disabled anyway?..
For the following reason:
<head><style>body{font-family: Geneva,Arial,Helvetica,sans-serif;font-size:9pt;background-color: #ffffff;color: black;}p{margin:0px}</style></head><body id="compText">Helena Oakey wrote<br><<<<br><o:p></o:p>
<p class="MsoNormal">There is some evidence from other research that a treatment examined in one of our clinical trials has a differential sex effect. I am in the process of re-examining the data for several outcomes and have fitted a model as follows: <o:p></o:p></p>
<p class="MsoNormal">Outcome=treat+sex+treat*sex<o:p></o:p></p>
<p class="MsoNormal">Where sex=male/female , treatment=placebo/treatment<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">I would like some advice on the best way to present the results in a paper. <o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">I have read the paper by Wang et al (07) which gives guidelines for how to present the results for a subgroup analysis-but it does differ to how I initially planned to present them. I was going to present the table of probabilities (p-values for interaction and main effects) and as there were no significant interactions -the tables of main effects one for each of sex and treatment which included the RR and 95%CI or mean difference and 95%CI (dependant on the outcome). <o:p></o:p></p>
<p class="MsoNormal">I notice in the wang et al (2007) nejm paper that they suggest that the interaction table should be presented- presumably regardless of whether an interaction is present? <o:p></o:p></p>
<p class="MsoNormal">I would have no qualms presenting the interaction table, if the interaction was significant, but given the potential to misinterpret results from a subgroup analysis I feel a bit uncomfortable about doing this if the interaction is not significant. I am wondering what the reason behind the guidelines is. Is the suggestion to present an interaction table because in clinical trials or health science data there is more of a concern about statistical vs clinical significance than say agricultural science (which is where most of my experience lies) or is it because of a concern of lack of power or because using the title from Hubbard and Lindsay paper âp-values are not a useful measure of evidence in statistical significance testingâ and so we should present the interaction results regardless? <o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">In addition to whether I should or shouldnât present the interaction table I can see several options on how to calculate the RR/CI and am wondering whether I include contrast p-values in the interaction table- <o:p></o:p></p>
<p class="MsoNormal">Should I present the interaction table from the full model with contrast p-values (i.e. p-values for treatment vs placebo within each sex)? <o:p></o:p></p>
<p class="MsoNormal">Should I present the interaction table from the full model without contrast p-values? <o:p></o:p></p>
<p class="MsoNormal">Should I present an interaction table but use the estimates from the model with just main effects present to determine the RR?<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Any advice, comments or examples of other papers which deal appropriately with this situation would be much appreciated. <o:p></o:p></p>
<p class="MsoNormal">I hope Iâm not being too pedantic or maybe even stupid (or both) about what I should be presenting! The only time I have presented an interaction table when it wasnât significant was when I had was investigating a factorial model with 3 terms of interest. All the three two-way interactions were significant but not the three way interaction. I presented a three-way interaction table (based on the model where I fitted the three two- way interactions) for ease of interpretations - so feel I am now in unchartered territory. <o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">There is a question that follows on from this. What about the non- subgroup analysis situation where an interaction is investigated. I am looking at another health science data set with maternal and infant outcomes where weâre interested in looking at an interaction between two terms (not a subgroup analysis situation). Is the general consensus that I should also be presenting the interaction table here again regardless of whether the interaction is significant? <o:p></o:p></p>
<p class="MsoNormal"><o:p> >>>></o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>The key question is NOT whether the interaction is significant, but a) whether it is large and b) whether it is important. Since you say that other research found differences, it is important for you to present your findings about those differences, regardless of their statistical significance. How big was your interaction? Is this similar to what other research found, or is it very different? <br></o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>I would therefore present the full model, with estimates of the effect sizes, and I would use the full model to calculate everything. You are only adding one variable to the equation.<br></o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>I would love to suggest that you NOT include p-values, but editors may object to that ... people have been (mis)trained to look at p-values rather than the things that matter. <br></o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>And the question is neither pedantic nor stupid (there are no stupid questions, right?). It's actually critical to the whole statistical enterprise.</o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>More generally than the above points - it gets at the question of how analysis should be performed. All too often, we let the statistics drag the substance along. In my view, this is silly. Statistics should be driven by substantive concerns, not vice versa. Here, this means that, since the interaction is important, it should be included.</o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>HTH</o:p></p><p class="MsoNormal"><br><o:p></o:p></p><p class="MsoNormal"><o:p>Peter<br></o:p></p><br><br><br>-----Original Message----- <br>From: Helena Oakey <br><blockquote style="border-left: 2px solid rgb(0, 0, 255); padding-left: 5px; margin-left: 0px;">helena.oakey@adelaide.edu.auSent: Sep 3, 2009 1:56 AM <br>To: medstats@googlegroups.com <br>Subject: {MEDSTATS} subgroup analysis
<br><br><zzzhtml xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<zzzhead> <zzzmeta http-equiv="Content-Type" content="text/html; charset=us-ascii"> <zzzmeta name="Generator" content="Microsoft Word 12 (filtered medium)"> <style> <ZZZ!-- /* Font Definitions */ @font-face {font-family:"Cambria Math"; panose-1:2 4 5 3 5 4 6 3 2 4;} @font-face {font-family:Calibri; panose-1:2 15 5 2 2 2 4 3 2 4;} /* Style Definitions */ p.MsoNormal, li.MsoNormal, div.MsoNormal {margin:0cm; margin-bottom:.0001pt; font-size:12.0pt; font-family:"Times New Roman","serif";} a:ZZZlink, span.MsoHyperlink {mso-style-priority:99; color:blue; text-decoration:underline;} a:visited, span.MsoHyperlinkFollowed {mso-style-priority:99; color:purple; text-decoration:underline;} span.gmailquote {mso-style-name:gmail_quote;} span.EmailStyle18 {mso-style-type:personal-reply; font-family:"Calibri","sans-serif"; color:#1F497D;} .MsoChpDefault {mso-style-type:export-only;} @page Section1 {size:612.0pt 792.0pt; margin:72.0pt 72.0pt 72.0pt 72.0pt;} div.Section1 {page:Section1;} --> </style> <zzz!--[if gte="" mso="" 9]=""><xml> <o:shapedefaults v:ext="edit" spidmax="1026"> </o:shapedefaults><zzz![endif]--><zzz!--[if gte="" mso="" 9]=""><xml> <o:shapelayout v:ext="edit"> <o:idmap v:ext="edit" data="1"> </o:idmap></o:shapelayout><zzz![endif]--> </zzz![endif]-->
<zzzbody link="blue" vlink="purple" lang="EN-AU">
<div class="Section1">
<p class="MsoNormal">Hi all</p><p class="MsoNormal">Thanks in advance<o:p></o:p></p>
<p class="MsoNormal">Helena<o:p></o:p></p>
<p class="MsoNormal"><span style="font-size: 11pt; font-family: "Calibri","sans-serif"; color: rgb(31, 73, 125);"><o:p> </o:p></span></p>
</div>
<br> </zzzbody>
</xml> <br> </zzz!--[if></zzz![endif]--></xml></zzz!--[if></zzzmeta></zzzmeta></zzzhea d></zz zhtml>/helena.oakey@adelaide.edu.au</blockquote><br> --~--~---------~--~----~------------~-------~--~----~<br> To post a new thread to MedStats, send email to MedStats@googlegroups.com . <br> MedStats' home page is http://groups.google.com/group/MedStats . <br> Rules: http://groups.google.com/group/MedStats/web/medstats-rules<br> -~----------~----~----~----~------~----~------~--~---<br> </body><pre>
Peter L. Flom, PhD Statistical Consultant www DOT peterflomconsulting DOT com I write some at:http://www.associatedcontent.com/user/582880/peter_flom.html I trade books at http://www.bookmooch.com/</pre> <br> ========================== Ted.
For information: The email with the above was received with ONLY an HTML attachment, no plain-text equivalent; so my client (which does plain-text only) showed me the above. I have the option to open an HTML attachment in an independent browser (an old version of KDEhelp), which doesn't succeed in parsing it, so I get the same anyway ...
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@manchester.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 03-Sep-09 Time: 22:19:27 ------------------------------ XFMail ------------------------------