The Paradox of Choice: Presented at the annual GPACA conference 2015. This presentation is geared towards anyone embarking on the process of purchasing a flow cytometer, specifically, but any laboratory technology more generally.
1. THE PARADOX OF CHOICE
Evaluating & Selecting the Right Cytometer for your Lab
Ryan Duggan
University of Chicago
Adapted from: http://ucflow.blogspot.com/2013/07/10-tips-for-purchasing-your-next.html
3. Paradox of Choice
• Too many options produce paralysis.
• We’re less satisfied with our decision because of FOMO.
• This conflict becomes even harder because choices are
so good.
• This is essentially where we are today with cytometers.
• Some choice leads to competition, too much choice leads
to paralysis.
The Paradox of Choice
By Barry Schwartz
4. Tyranny of the Default
• People are lazy efficient.
• When presented a default option that is good enough,
they will be less inclined to explore other options, even
when other options are better.
• IE on Windows
• Today, we purchase cytometers for which we have
familiarity. The “BD” lab or “Coulter” lab.
5. Instrument Choices
• In ~1999, we purchased the 3-laser BD LSR.
• At the time we had the choice between BD, Beckman-
Coulter, and Partec
• Also the Guava instrument became available in 1998, but
it was too limited.
• Today, we can choose from a number of additional
platforms, not just within those same 3 companies, but
also new companies who joined the world of cytometry.
6. 10-step process
1. Define Needs
2. Query the Userbase
3. Refine Needs
4. Survey the Market
5. Navigate Marketing
6. Create the Matrix
7. Hands-on
8. Social Networks
9. Negotiate
10. Year-end deals
7. Define Needs
• Create a generic
specification of the
needed equipment
based on YOUR
perception.
• What applications do
you see being run
immediately, near
future?
8. Query the Userbase
• Survey the users to find
out what they need from
THEIR perspective.
• Analyze recent historical
data
• Talk to key Investigators
• Research current
industry trends
• Ignore much of what the
users say they need.
9. Refine the Needs
• Find the middle ground
• Realistic specification
with “room to grow”
10. Survey the Market
• Research what’s
available
• Base entry for
consideration on min.
spec.
• Use organizational
tools like Evernote™
or ELN to track
contenders.
11. Navigate the Marketing
• Learn to read marketing materials
• Differentiate between Technical spec (what the instrument
has) and Performance Spec (what the hardware can do).
• Marketing materials are fine for listing Tech. Spec. but
NOT Performance Specification.
• Samsung found itself accused of artificially (and secretly) boosting
benchmark scores on its flagship phone to ensure it would
outperform the competition.” - The Verge (http://goo.gl/5vuCs9)
12. Create the Matrix
• List of specifications vs. available hardware
• Use marketing material tech. spec.
• Include budget as a spec.
# of Lasers
(Total Avail.)
Total
Detectors
Field
Upgradable
Multi-well
Sampler
Event Rate
Total
Events/File
Optical
Upgrades Avail.
Instrument A 1-3 (3) 3-8 No Yes 2,000 100,000 No
Instrument B 2-4 (7) 4-12 Yes Yes 10,000 5,000,000 Yes
Instrument C 2-6 (12) 4-18 Yes Yes 20,000 10,000,000 Yes
Instrument D 2-4 (4) 4-10 No No 20,000 2,000,000 No
Instrument E 2-5 (8) 4-14 Yes Yes 20,000 10,000,000 Yes
Instrument F 1-3 (4) 3-6 No No 2,000 100,000 No
13. Hands-on
• NO CANNED DEMOS
• NO EXAMPLE DATA SETS
• Real data, collected by you, analyzed by you.
• Create your own performance metric if none exists.
• Make THIS, a high level priority that will greatly influence
your decision
14. Get Social
• Read reviews
• Ask peers
• Request OEM
response for any
negative feedback
• Avoid Gen. 0
evangelists.
15. Negotiate Purchase
• Negotiate with multiple OEM’s regardless of intent to
purchase
• Make sure OEMs are aware of their competition
• Competitive bid may be required anyway
• Get everything in writing
• Include Training, extended warranty, shipping, installation,
free upgrades, etc…
16. Be Mindful of Year-end deals
• If you can time it right, year-end numbers can be a
powerful bargaining chip.
18. Hands on evaluation opportunities
• Travel to company to test
• Difficult to prep samples – ship
• Usage based on “ideal” situation
• Includes canned bead demos which have variable utility
• Limited time with instrument
• Travel to lab who currently has the instrument
• Still some difficulty in running samples
• Get a feel for real-life use of the instrument
• Limited time with instrument
• Get in-house instrument demo (hopefully for more than 1
day)
• Prep samples of various kinds to test different aspects
19. What to test
• Performance
• Real event rate
• Resolution
• Linearity
• Carryover
• Standard FL Panel
(Data not shown)
• Ease of use
• Startup
• QC
• Software
• Shutdown
• Administration
29. Ease of Use – Startup/Shutdown
• There should be a startup/shutdown routine (automated
preferably).
0 10 20 30 40
FACSCanto
NovoCyte
Attune NxT
Startup
QC
Dots on Plots
30. Ease of Use – QA/QC
• Either completely managed or not at all.
• BD FACSDiVa CS&T
• Fully managed, pass/fail, historical information at-a-glance.
• Requires specific, single-source bead set
• Can be time-consuming depending on adjustments needed
• MoFlo
• DIY QC/QC
• No built-in platform for tracking or pass/fail confirmation
• Can be fast depending on operator experience
31. Ease of Use - Administration
• Generally overlooked even on modern cytometers
• User on-boarding
• User tracking/permissions
• BD FACSDiVa
• Available via Import/Export
• Runs on Windows allowing for custom methods (labstats, iLab)
• Miltenyi MACSQuant
• Runs in terminal mode so no custom user tracking options
• Limited in usability for large groups
32. Hands-on Summary
• You need to spend some time on the instrument
• Derive experiments that test the things that are most
important to you.
• Resolution, Event Rate, Carryover, Linearity
• Small Particle Detection
• Sample volume flow rate
• Fluorescence spillover
• Usability features can be measured too
• Cold start -> Dots on Plots
• Presence/Absence of fully managed QA/QC
• Presence/Absence of Administrative functions
33. Conclusion
• Recognize when you’re being paralyzed by choice,
• Expand your evolutions beyond your default.
• We’re in a period of great innovation and choice
• Pare down list of contenders according to tech. spec.
• Gain hands-on experience with the short list
• Quantitatively describe performance that matters to you.
• Usability is not a luxury anymore, it’s a necessity.
34. Thank you
• Instrument manufacturers who let me test their stuff
• Acea Biosciences
• Beckman Coulter
• Becton Dickinson
• BioRad
• EMD-Millipore
• Handyem
• Miltenyi Biotec
• Propel Labs
• Stratedigm
• Thermo Fisher
UCFlow.blogspot.com/UC
Editor's Notes
Descriptive specification vs. performance specification. OEM Marketing materials can be used for descriptive spec, but not performance spec. You need to define your own performance spec using various methods.
Study done that looked at retirement plans offered to employees. Result = for every 10 additional plans offered, rate of participation in any of the retirement plans dropped 2%. So, offering 50 plans instead of 5 results in a decrease in participation of almost 10%. It’s so hard to decide among the 50 plans that people just give up.
Whatever you choose, it’s not going to be perfect, which leads you to second guess your original decision. Regret subtracts from satisfaction. The features of the instruments you rejected that were somewhat attractive now become a point of fixation on your part. This leads to FOMO.
Competition breeds innovation, innovation leads to better products, and now that we have so many good options, the Paradox of Choice becomes even more prevalent.
Guava prototypical “next best choice”
As SRL managers, we may see the applications being run in our facilities to be a bit on the modest side, so practically we may suggest something that satisfies the needs of many users. I.e. a 2/3 laser 4-6 color instrument.
Users tend to overestimate their true needs and might suggest the most popular/cutting-edge instrument available.
Filter all the wants into a concrete set of needs to come up with a reasonable set of specifications.
People who are part of a project from day 1 sometimes have clouded judgement. They want the hardware that they helped bring to market succeed.
This is what matters to me. You might be really interested in minimum/maximum sample volume or sample volume flow rates, etc…, but I care about these things.
For this test, I created a concentrated sample using a suspension cell line, which, at 60ul/min should yield 50,000 events per second (50 million per mL yields 50,000 per second at 60uL per minute). I then created serial dilutions from there all the way down until an expected 2,500 events per second. I ran each tube on the instrument and recorded the event rate displayed by the system's counters.