Bioinformatics survey results

By Grant Jacobs 07/06/2012 2

Eagle Genomics have presented the results of an annual bioinformatics survey.

These types of surveys are invariably limited by understanding precisely who they represent, e.g. do they represent the whole of the target group well, or favour some sector of it. The results of the first question offer some hints of this. (Ideally we’d see a comparison with what balance is considered to represent the fullness of the bioinformatics workspace to put their survey in wider context.)

Their survey is dominated by those from the university sector. (I find the fairly large ‘non-profit’ sector interesting, though.)

Questions 6 and 7 assess respondents views on bioinformatics tools. I think it’s worth considering the results of these questions viewed by who is to use (and/or maintain) the tools –  staff who specialise in hands-on bioinformatics or ‘motivated’ users. (The survey will very likely favour the former.)

They express surprise at the number using ’antiquated’ array expression technology in question 3 – my instincts are that the response might, at least in part, reflect the availability of public data rather than new use of the technology itself.

I would like to have seen the results of question 8 (views on outsourcing) broken down by the sectors from question 1 (respondents’ workplaces), if this were meaningful.

Confirmation bias noted, it’s nice to see their interpretation of the results of question 2 that ’It is anticipated, based on these results, that future attention will be paid most closely to proteomics (25/108), systems biology and epigenetics (19/108 each), […]’ as this overlaps well with my own expertise and intended direction.

I wonder if their final words to their summary of question 2 will have tongues wagging: ’This suggests that despite the increase in availability and use of genomics sequencing facilities, genomics may stealing be decreasing as the most important field of research in future bioinformatics labs.’

2 Responses to “Bioinformatics survey results”

  • I agree with your comments. 108 responses and an evident amount of bias make this study much less useful than it could. I think that one could easily improve on the set of questions and conduct a survey with am much wider set of target responders, and care should be taken in ensuring the ability to stratify responses and lovcate them geographically too. I would be happy to contribute to the planning and implementation.

  • I hope my article isn’t coming across quite so harshly; I only mean the points as interpretative thoughts rather than brickbats!

    These sorts of surveys always have a few catches with respect to who they cover. As much as you can try get a larger sample, in the end you’re stuck with whatever you’ve managed to get. (If you read their article you’ll see they’ve limited themselves to offering what they felt might still offer some insight.)

    Similarly, trying to determine to what extent your participants ‘match’ the overall population isn’t necessarily straight-forward.

    Stratification would help mitigate this (as you imply) in the sense that comparing responses between different sectors can be useful even if you’ve not certain what the exact balance of the different sectors in the total population is, but “of course” that requires (much) larger numbers to work effectively. There’s never a free lunch, eh?!

    Great to hear you’d like to pitch in! Maybe one day I’ll be covering your survey here – let me know.

Site Meter