Hi both,
Sorry for the delay in replying. Thank you so much for your help and advice - it is greatly appreciated!
Many thanks.
Best wishes,
Laura.
________________________________
From: Research of postgraduate psychologists. <[log in to unmask]> on behalf of Jeremy Miles <[log in to unmask]>
Sent: 21 May 2018 20:54
To: [log in to unmask]
Subject: Re: Purpose of this JISC list - stats query!
I agree with Thom about how to analyze this (I think the last time I disagreed with Thom was some time around 1997, and that was about whose round it was), but one other thought.
If you're only interested in predicting (which is what the email said, but you might not have meant) you have 2 (victim gender) * 3 (perp gender) * 2 (participant gender) = 12 groups. Just label them 1 through 12 and you have twelve groups, and you don't need to worry about constructing and interpreting interaction effects.
Also, again if you just want to predict, enter your group variable as a categorical variable (with 12 categories) and then run the model without an intercept. The parameter estimates are the predicted values (when covariates = 0) and you have standard errors / confidence intervals for each of these, which saves a fiddly step later on. (But beware - the p-values are meaningless in this case.)
Jeremy
On Sat, 19 May 2018 at 11:02 Baguley, Thomas <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Laura,
Generally one forms the product terms for the interaction in the same way by multiplying the two predictors. If you have a categorical predictor with more than k = 2 values this means (k-1) product terms to capture the interaction effect. Collinearity should not be an issue here. The test of the interaction (the change in model F when you add all the product terms) is unaffected by the collinearity with the main effects.
Where it does have an impact is on the the interpretation of the coefficients for the indicator variables representing the main effects. This interpretation depends on the parameterisation of the model. For example, with dummy coding parameterisation the interpretation of one group becomes the intercept and the others differences from the intercept and the product term coefficients the slope for the intercept and differences in slopes for the other categories. You can change the parameterisation using effect coding (which works a bit like entering in this context) and generally it is always sensible to centre continuous predictors prior to calculating product terms.
However, my advice is generally to focus on interpreting effects looking at the adjusted means or equivalently the model predictors - ideally graphically if it looks like there might be interactions.
If all the interactions are with categorical predictors like gender I would consider effect coding these. This gives you a parameterisation similar to an ANOVA. Some software will more-or-less do that for you if you run the model as an ANCOVA (it isn't quite the same but has a similar interpretation). To do this manually:
dummy effect
male 0 -0.5
female 1 +0.5
Classically effect coding using -1 and +1 but I usually prefer -0.5 and + 0.5. The latter means there is a 1 unit difference between the groups and this the slope represents the difference in groups as it would for dummy coding (not half the difference in groups). In a balanced design the intercept is now the grand mean.
With more than two categories it gets messier but you can extend effect coding like this:
effect 1 effect 2
cat1 -0.5 -0.5
cat2 0 +0.5
cat3 +0.5 0
Depending on software running this as an ANCOVA might be the simplest option (this is effectively a regression where the categorical predictors are parameterised in a particular way). In either case it is probably a very good idea to centre any continuous covariates before you do anything else.
Thom
________________________________
From: Research of postgraduate psychologists. <[log in to unmask]<mailto:[log in to unmask]>> on behalf of Laura Scurlock-Evans <[log in to unmask]<mailto:[log in to unmask]>>
Sent: 19 May 2018 17:47
To: [log in to unmask]<mailto:[log in to unmask]>
Subject: Re: Purpose of this JISC list - stats query!
Hi everyone,
Same here! To prove it too, I have a stats query that I wondered if anyone could help with? I have been looking into the best way to analyse some data for my PhD, but am a bit stuck. I have a continuous outcome variable (measured on a scale of 40-240) which I would like to predict from victim gender (male/female), perpetrator gender (male/female/not specified) and participant gender (male/female), whilst controlling for a number of other variables (e.g. socially desirable responding, social roles endorsement). I am particularly interested in the interaction between victim gender and perpetrator gender, and the three-way interaction with participant gender, but am also interested in the 'main effects' for these variables too.
I understand how to create interaction terms between continuous variables in linear regression, but can't seem to find any guidance which explains creating interaction terms between categorical variables (e.g. how to avoid multicollinearity). I wondered if anyone could offer any advice?
Any help would be gratefully received!
Many thanks and hope you all have a lovely weekend.
Best wishes,
Laura.
________________________________
From: Research of postgraduate psychologists. <[log in to unmask]<mailto:[log in to unmask]>> on behalf of Helen Foster-Collins <[log in to unmask]<mailto:[log in to unmask]>>
Sent: 17 May 2018 12:56
To: [log in to unmask]<mailto:[log in to unmask]>
Subject: Re: Purpose of this JISC list
Ditto! I am not doing stats at the moment, but find it very reassuring to know that support is there if required.
:)
On 17/05/2018 11:37, Martin Brennan wrote:
Jeremy,
I have a stats folder in my inbox full of your (and others) helpful responses to stats based queries, and long may it continue being added to!
Martin
Sent from Google Pixel.
On Wed, May 16, 2018 at 17:28, Jeremy Miles
<[log in to unmask]><mailto:[log in to unmask]> wrote:
Hey Everyone,
tl;dr: Is it appropriate for me (a non-postgrad) to be on the list?
I started this list (when it was hosted by Mailbase) in 1993. (Whoa! That's 25 years of psych-postgrads!) It ran independently for a while, and then when I got a real job I stopped being the owner and it merged with PsyPAG (which was a good idea), and has been run by them ever since.
When I started it, I thought that it should be a place for grad students to talk about issues - and that having non-grad students on the list wasn't necessarily a good thing. I don't want to complain about my supervisor or department head if I think that a friend of theirs might be on the list and 'listening in'. (I don't think anyone has ever done much moaning about that sort of thing though.)
I've stayed on the list because I'm still interested in it, and I try to answer people's questions about statistics. Once in a while i send an email where I ask if list members (and owners) think it's appropriate that I'm still on it. This discussion made me realize that it's been a while since I sent one of those emails, and membership has a reasonable turnover - so this is it.
I'm not in the UK, and I don't work in a university (although I do a little bit of teaching, but not in psychology departments, some in the UK, some not). But that doesn't mean that I'm not friends with your supervisor / head of department.
Jeremy
On Wed, 16 May 2018 at 03:36 Talbot, Catherine <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear JISC list members,
Following a number of emails that were sent yesterday about the new BPS branches, PsyPAG would like to clarify that we do not have a position on these branches and would like to encourage everyone to vote in a way they see fit.
We would also like to remind everyone about the purpose of this JISC list. The purpose of this JISC list is for postgraduates and ECRs to send/receive information that is relevant for postgraduates and to support one another. This includes sharing information that may be of interest (e.g. conferences, workshops, training courses), asking questions (e.g. questions about stats), participant recruitment (although we ask that you only send one request to this JISC list), and posting job opportunities. We want this to be a friendly, safe, and supportive forum for postgraduates and ECRs, so we ask that all members use it appropriately.
If anyone does wish to leave this JISC list for whatever reason, please do so by:
1. Going to https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=PSYCH-POSTGRADS&A=1<https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.jiscmail.ac.uk%2Fcgi-bin%2Fwebadmin%3FSUBED1%3DPSYCH-POSTGRADS%26A%3D1&data=01%7C01%7Cthomas.baguley%40NTU.AC.UK%7C3f8cfbd249b74fa6975208d5bda8358b%7C8acbc2c5c8ed42c78169ba438a0dbe2f%7C1&sdata=kFaWqJAl%2F67amifPVVvwavfFfTb3RIi%2FgI166Wka4Lc%3D&reserved=0>
2. Completing the form by inserting your name and the email address that is signed up to the mailing list
3. Clicking unsubscribe
Best wishes,
Catherine (PsyPAG Communications Officer)
Catherine Talbot MBPsS
PhD Student
University of Exeter Medical School
St Lukes Campus, Heavitree Road, College House 1.23, Exeter, EX1 2LU
REACH: The Centre for Research in Ageing and Cognitive Health
School of Psychology, College of Life and Environmental Sciences &
PenCLAHRC, Institute of Health Research, University of Exeter Medical School
http://psychology.exeter.ac.uk/reach/<https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fpsychology.exeter.ac.uk%2Freach%2F&data=01%7C01%7Cthomas.baguley%40NTU.AC.UK%7C3f8cfbd249b74fa6975208d5bda8358b%7C8acbc2c5c8ed42c78169ba438a0dbe2f%7C1&sdata=yk0PhWVepXVMS0eOSUL%2B9lfFi4dOHoixqZE%2Bs8dGvUQ%3D&reserved=0>
PsyPAG Communications Officer
PsyPAG South West of England branch representative
Twitter: @Catherinetalb
[http://www.exeter.ac.uk/codebox/email-sig/staff-sig.gif]
This email and any attachment may contain information that is confidential, privileged, or subject to copyright, and which may be exempt from disclosure under applicable legislation. It is intended for the addressee only. If you received this message in error, please let me know and delete the email and any attachments immediately. The University will not accept responsibility for the accuracy/completeness of this email and its attachments.
--
--
My employer has nothing to do with this email.
[https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif]<https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.avast.com%2Fsig-email%3Futm_medium%3Demail%26utm_source%3Dlink%26utm_campaign%3Dsig-email%26utm_content%3Demailclient&data=01%7C01%7Cthomas.baguley%40NTU.AC.UK%7C3f8cfbd249b74fa6975208d5bda8358b%7C8acbc2c5c8ed42c78169ba438a0dbe2f%7C1&sdata=5Gfnk6cxO9Ud6fUAW%2Bljlif42SNonBcBOvoTzcUzRN0%3D&reserved=0> Virus-free. www.avast.com<https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.avast.com%2Fsig-email%3Futm_medium%3Demail%26utm_source%3Dlink%26utm_campaign%3Dsig-email%26utm_content%3Demailclient&data=01%7C01%7Cthomas.baguley%40NTU.AC.UK%7C3f8cfbd249b74fa6975208d5bda8358b%7C8acbc2c5c8ed42c78169ba438a0dbe2f%7C1&sdata=5Gfnk6cxO9Ud6fUAW%2Bljlif42SNonBcBOvoTzcUzRN0%3D&reserved=0>
DISCLAIMER: This email is intended solely for the addressee. It may contain private and confidential information. If you are not the intended addressee, please take no action based on it nor show a copy to anyone. In this case, please reply to this email to highlight the error. Opinions and information in this email that do not relate to the official business of Nottingham Trent University shall be understood as neither given nor endorsed by the University. Nottingham Trent University has taken steps to ensure that this email and any attachments are virus-free, but we do advise that the recipient should check that the email and its attachments are actually virus free. This is in keeping with good computing practice.
--
--
My employer has nothing to do with this email.
