![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
percivalOver the past couple of months, as I worked 10-30 hour weeks with two kids and no child care for DS (8 months, cute kid, very good at playing on his own for a little while, bad napper), I often asked myself - why am I doing this? Why am I revising two journal papers, submitting another two (one as first, one as second author), dealing with PhD students, working on project reports?
Because I love it and there's nothing I'd rather do.
I'm one of those people who knows a little about a lot. I'm also one of those people who loves getting stuck into a pile of papers and a huge file of data and finding connections and results. I enjoy crafting a piece of writing that communicates those results.
What I am interested in is helping people interact with computer systems. I'm both a speech scientist and a computer scientist, so the interaction modality I focus on is speech. I specialise in computer voices (aka speech synthesis) and spoken dialogue with computers. I'm not a programmer; rather, my weapons of choice are experiments and statistical data analysis. I can code to support my data work, but I don't implement solutions or new algorithms. Over the past few years, I've moved towards working with older users. I'm trying to understand what aspects of ageing affect how older people interact with computers, and incorporate that into my experimental designs. Older people are challenging because they're so diverse. This is part genetics, but also part life style. If you've been a couch potato all your life and if your main intellectual challenge has been to second-guess Noel Edmonds on Deal or No Deal, chances are you won't be as astute as Peter Higgs (he of the Boson), even if you share the same birthday, went to the same school and started out with the same IQ. If you've been a heavy smoker, your voice will have aged more than if you had done the right thing and abstained (unless you worked as a teacher and had to shout down unruly class rooms for decades. Or maybe you were a heavy smoker AND a teacher.) Older people = variation = difficult statistics = fun! In order to get a handle on the variation, I find myself working with people who understand some of the sources, such as cognitive psychologists and audiologists. I learn a lot during these interactions, mainly a healthy respect for the other field's methods.
And it also means I need to brush up on my stats skills. I need to be careful to select appropriate models and analysis techniques, and NOT to over interpret. I also need to keep reminding myself that a good graph says more than a dumb test. A significance level of 0.05 means nothing if it's attached to ONE significant finding in 45 correlation tests you ran on the same set of 10 variables. If these variables are not normally distributed, and you used Pearson's instead of Spearman's correlation coefficient, you're even more screwed. Teams in medicine usually have resident statisticians; us computer scientists don't tend to have that luxury. You're supposed to be able to handle complex designs and difficult analyses because as a computer scientist, you're mathematically literate, right? Because you know all about exotic logics, you can select and run the appropriate significance test, right? Wrong. So I'm teaching myself and trying to learn from best practice. (By the way, I know nothing about exotic logics, but I know which formal semanticists to ask if I have to.)
And, finally, I'm one of those people who work part-time. If, as I do, you collaborate with people from eight departments from a total of five universities including your own (and that's just the main collaborations!), you start to develop a (un)healthy interest in organisation. Good data management, good time management, realistic goals and expectations. You become ... anal. There, I said it. At first, organising your work is not as cool as doing great research, then, it becomes a good procrastination method, and finally, you realise that organisation will always be a work in progress.
This is a submission to the December Scientiae blog carnival
Because I love it and there's nothing I'd rather do.
I'm one of those people who knows a little about a lot. I'm also one of those people who loves getting stuck into a pile of papers and a huge file of data and finding connections and results. I enjoy crafting a piece of writing that communicates those results.
What I am interested in is helping people interact with computer systems. I'm both a speech scientist and a computer scientist, so the interaction modality I focus on is speech. I specialise in computer voices (aka speech synthesis) and spoken dialogue with computers. I'm not a programmer; rather, my weapons of choice are experiments and statistical data analysis. I can code to support my data work, but I don't implement solutions or new algorithms. Over the past few years, I've moved towards working with older users. I'm trying to understand what aspects of ageing affect how older people interact with computers, and incorporate that into my experimental designs. Older people are challenging because they're so diverse. This is part genetics, but also part life style. If you've been a couch potato all your life and if your main intellectual challenge has been to second-guess Noel Edmonds on Deal or No Deal, chances are you won't be as astute as Peter Higgs (he of the Boson), even if you share the same birthday, went to the same school and started out with the same IQ. If you've been a heavy smoker, your voice will have aged more than if you had done the right thing and abstained (unless you worked as a teacher and had to shout down unruly class rooms for decades. Or maybe you were a heavy smoker AND a teacher.) Older people = variation = difficult statistics = fun! In order to get a handle on the variation, I find myself working with people who understand some of the sources, such as cognitive psychologists and audiologists. I learn a lot during these interactions, mainly a healthy respect for the other field's methods.
And it also means I need to brush up on my stats skills. I need to be careful to select appropriate models and analysis techniques, and NOT to over interpret. I also need to keep reminding myself that a good graph says more than a dumb test. A significance level of 0.05 means nothing if it's attached to ONE significant finding in 45 correlation tests you ran on the same set of 10 variables. If these variables are not normally distributed, and you used Pearson's instead of Spearman's correlation coefficient, you're even more screwed. Teams in medicine usually have resident statisticians; us computer scientists don't tend to have that luxury. You're supposed to be able to handle complex designs and difficult analyses because as a computer scientist, you're mathematically literate, right? Because you know all about exotic logics, you can select and run the appropriate significance test, right? Wrong. So I'm teaching myself and trying to learn from best practice. (By the way, I know nothing about exotic logics, but I know which formal semanticists to ask if I have to.)
And, finally, I'm one of those people who work part-time. If, as I do, you collaborate with people from eight departments from a total of five universities including your own (and that's just the main collaborations!), you start to develop a (un)healthy interest in organisation. Good data management, good time management, realistic goals and expectations. You become ... anal. There, I said it. At first, organising your work is not as cool as doing great research, then, it becomes a good procrastination method, and finally, you realise that organisation will always be a work in progress.
This is a submission to the December Scientiae blog carnival
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2008-11-28 01:55 pm (UTC)Ha, so now *I* know who to ask, good :-) (I have one tame statistician, who is very useful, but she has a day job which isn't for us.) Some time, in all your copious free time, you should give a course on "statistics for computer scientists" or some such.
More seriously, is there a really good book you can recommend? I haven't found a great source yet, and the web doesn't help, since it either assumes I know more stats than I do or assumes I know less maths and is irritating or illogical or both.
no subject
Date: 2008-11-28 08:23 pm (UTC)no subject
Date: 2008-11-28 09:11 pm (UTC)