[Music] okay hello everyone and thank you so much
for joining us today for this donor box webinar my name is jenna and i am the brand ambassador
for donorbox and i'll also be the moderator for this webinar today for those of you who
are not familiar with donorbox we provide non-profits with simple and effective tools
to manage their online fundraising activities and connect with individual donors on a deeper
level today we've helped 35 000 organizations raise over 700 million dollars to lea
rn more
you can visit us on our site at donorbox.org today we are joined by dataviz expert emilia
comb to talk about making your nonprofit's data visible and useful this is so important you
guys i cannot stress this enough now amelia she is the founder of dataviz for nonprofits
which delivers high quality visualizations that help organizations to quickly grasp their
data improve their work and show their impact amelia has more than 25 years of experience
studying funding and evaluating hum
an services she was a researcher at chapman hall at the
university of chicago which is a policy research center where she conducted studies on a wide
variety of issues impacting children and families prior to chapman hall she worked as a program
officer at this year's roblox foundation at the illinois humanities council she earned her va
from haverford college an m.a from the school of social services administration at the university
of chicago and a phd from the university of bath and she
was trained in data visualization at the
university of washington so needless to say we are in such good hands today and we're really excited
to have her now before i pass the baton over to amelia i just want to mention a couple of items
please note that this webinar will be about 60 minutes long with an additional 5 to ten minutes
allotted at the end for q a this webinar is being recorded so within the next 24 hours after the
webinar ends you will receive that link via email now you will n
otice in the bottom right hand
corner of your screen that you have a chat box please feel free to drop questions in this section
throughout the webinar and we'll track those and answer as many as we can and we'll also be sharing
several helpful links throughout this webinar and those will pop up in the left hand corner of your
screen and you can click those as they appear and the link will open right up all right now
without further ado please everybody give amelia a warm warm welcome thank
you so much jenna it's
great to be with you all today um i'm going to ask you for the beginning of this presentation
to just direct your full attention to the screen um all right um so my plan is to take you on a
somewhat transformative journey today from generic non-profit staff member to superhero
making full use of your super powers or if you prefer your secret weapons okay so i'm sure
each of you have multiple superpowers or secret weapons but today i'm gonna focus on just two of
them
and they are one your ability and it's an ability that you share with your colleagues other
staff members your board members your funders um and it's an ability to process
visual information at lightning speed um and then your other super power is the data
that your organization collects every day data visualization puts these two super powers
into action and it will help you to understand and show what you do and to craft
new strategies to improve what you do so here's the journey which i
've visualized as
a sort of graph this is the journey we're going to go on today we begin now by discussing the
mostly very rational reasons that many people in non-profit organizations often don't make good use
of data then we'll consider your two super powers of secret weapons then we consider your challenges
and um we'll focus on all kinds of challenges including fundraising challenges and how to use
the secret weapons to address those challenges and i'll draw to a close by pointing you
to resources that will allow you to hone your secret weapons beyond this workshop and then
we'll open it up to questions as jenna mentioned um so a little bit more about me thank you for the
wonderful introduction jenna um i love evidence but i hate spreadsheets i've spent more than 25
years studying funding and valuing human services as already mentioned mostly as a researcher at
chapel hall which is at the university of chicago um at chapin hall i worked with a lot of
organizations and t
heir data and came to know the difficulties that many organizations
face in using data well i felt there had to be a way for organizations without high level
data analysis skills in-house to extract meaning quickly and painlessly from their data and to
put it to work for them and that's when i got interested in data visualization or data biz for
short um i now help non-profits consultants and foundations to visualize their data through my
consulting business called dataviz for nonprofits ok
ay so subject number one on our journey why
you don't use data or more correctly why you don't use it as as effectively as you might in
my working with organizations i've come across six primary reasons and let's talk about each one
so reason number one is data naivete or aversion um non-profit staff tend to have expertise
in such issues as the environment the arts health education but not data analysis some have
a downright data aversion they admit or maybe even proudly proclaim that they'
re not numbers people
um nonprofit staff often don't have time for data um of course we're all struggling to stay
afloat to submit the next funding proposal to maintain our programming not to mention
addressing the huge and very needs of our clientele and cultivating donors and digging
through data really gets put on the back burner some understandably worry what the data will show
non-profit staff may fear that they won't be able to control the story or that the data will be
taken out of
context that they'll be compared to other organizations which aren't indeed comparable
they fear that funders will withdraw support based on data no matter how much a funder might talk
about things like continuous quality improvement and such some nonprofit staff have low level
nonprofit organizations have low level step entering data into management information
systems or spreadsheets other organizations have multiple staff members entering data and
each person is entering data a little di
fferently and the result can be what we call dirty data
that's data that is inaccurate because it's not been entered in a consistent way so for
example if the participant is entered twice into a database once as michael smith
and then again later as michael b smith when he drops out and rejoins a program for
example then tracking this participant's progress is going to be really difficult because as far as
the date is concerned he's two different people all right so many nonprofit staff hav
e data
on their participants and financials but they often lack data to show impact so for example a
tutoring program may not have their participants school grades or test scores because for ferpa
reasons federal government regulations it's hard to get that data out of school districts
also an employment program may not have data on former participants wages over time this kind
of data can be hard to get because it requires tracking down former participants and surveying
them other outcome
s are harder to quantify again it may require a survey which is
time consuming and expensive to conduct and then finally rather than having essential
management information systems smaller nonprofits and nonprofit staff may store their data in
separate excel or google spreadsheets so michael smith's demographics might be on one sheet and
his attendance in various programs might be on other sheets and that makes analysis of say the
relationship of age to program participation really difficul
t because those data
fields are in separate data sources okay so we talked about why you don't use data
and guess what one of your secret weapons is well as i mentioned it's data so most organizations
have lots of data packed away in databases and spreadsheets that they are not putting to good
use um data's the way we know if what we're doing is worthwhile you can think of it as sort of a
tool we don't invest in other tools say computers and then store them away and rarely use them
but tha
t's what we do with data all all the time you know we're entering the data but we're not
taking it out and using it as often as we should even small organizations collect data on
many things here's a sample of the types of data that most nonprofits collect not only
participation in programs and donations but also website traffic volunteers financials audience
okay so this is the point in the presentation where i present my little commercial or
public service announcement if you will about d
ata so now i understand that you may feel that
data doesn't need a commercial it's already been way overhyped we hear and talk about data all the
time we hear about evidence prep evidence-based practices data-driven solutions key performance
indicators but in all the hype around data i think we often lose sight of what its core importance
is so that's why i offer you this little public service announcement um data is important because
of how human perception works what we perceive is based
not only on what we actually observe in
the outside world what we might call data but also what we expect to observe and this is how it works
the brain evaluates which of a variety of probable events are actually occurring and what we perceive
is based on part in part on what decision the brain makes sure what we perceive is also based
on those incoming signals from the outside world but there are far more signals coming from
within our brains that affect our perception so a great example o
f this is a youtube video
that went viral several years ago i'm not going to show you the video because i bet a lot of
you have seen it but if you haven't i'm going to just give you the little um 30 second recap
now so if you're like me you saw it as part of a presentation and the presenter got up and
said i'm about to show you a two-minute video and your job is to watch the video
you're gonna see a bunch of people passing a basketball amongst them your job is to
count how many times the b
asketball gets passed and so if you're like me you dutifully watch the
video you count how many passes the presenter gets back up and says show of hands how many people
saw a guy in a gorilla suit walk through the frame if you're like me you didn't see any guy
in a gorilla suit and sure enough the presenter shows the two video two minute video again the
guy in the gorilla suit is very easy to see um but you often you don't see that guy
because you didn't expect to see him you were primed to
focus on the basketball passes and you
didn't see this very obvious guy in a gorilla suit so these inner brain signals our expectations
can distort our understanding of a situation thus the incoming signals from the outside world
again what we might call data are quite important to confirming or negating our expectations but the
key is you have to pay attention to data to make that happen progress in organizations and indeed
in all of human history often starts with the concession that our
expectations might be wrong
and that data might have something to teach us of course the scientific method that you learned
back in the seventh grade primes us to focus on data rather than expectations we have to formulate
a hypothesis and then use data from the outside world to see if our hypothesis holds up okay so
that's the end of my commercial but i also want to include a little disclaimer here data is part of
the answer but it's never the whole answer perhaps love is the whole answer
who knows but indeed data
is more about questions than it is about answers it can show you what questions you should be
asking and help you to refine those questions it can also point you in the general right
direction but you are never going to have perfect and comprehensive data data will
never take you all the way to your answer so you should you know so you apply data
to your own experience to make decisions also we should be wary although not fearful of
data we should also always be
asking where does the data come from is the source likely
to be biased if it's drawn from a sample is it representative of the larger
population it's supposed to represent what's missing from the data are there groups of
people not represented in the data are there time periods not represented in the data and why is
that data missing are some survey respondents for example hard to reach or do they lack trust in the
data collector so these are mainly questions we would ask from outside data
sources but they might
also apply to the data that you collect every day now let's talk about your second secret weapon
remember it's your visual acuity or visual superpowers um so take a look at these two images
which image do you comprehend first the mona lisa or the spreadsheet okay do not put your answers
in the chat window i'm gonna just take a wild guess that you understand the mona lisa first and
that's because our visual system has evolved over millions of years to processes process
images
essentially in parallel so what do i mean by that process in parallel it means that we don't
read the mona lisa from top to bottom and from left to right we take it all in together
and understand almost instantly that this is a picture of a woman in front of a landscape
sporting a dark dress and an inscrutable smile words and numbers which only appeared within
the last few thousand years require us to scan individual characters one at a time recognize them
piece them together into
words or values and then sentences and equations and while all types of
sensory signals barrage the brain the brain is especially geared for visual signals and indeed
uses about half of its volume for visual analysis and that's because for a long time during human
history survival depended on detecting danger such as predators lying in the tall grass using our
visual systems so when we use visual cues like length color and size we transfer much
of the work that is needed to decode data from
the slower conscious energy intensive
parts of the brain to the faster parts of the brain that require less energy that have
been more evolved over these millions of years and that results in more efficient cognition data
visualizations such as charts graphs and maps use these visual cues to communicate data and
thus greatly speeds up our processing of data and so addresses in part that time crunch that
i talked about um database is a quick way to also assess how dirty disconnected or irre
levant
our data is if the picture doesn't look right or complete then we need to do something to improve
our data maybe that means collecting fewer data elements and doing so more accurately so now
let's test our visual processing superpowers so take a look at this very simple spreadsheet and
try to answer the questions below i'm really only going to give you a few seconds to answer these
questions um which are which were consistently higher foundation or corporate grants and what
were the
trends for each type of grant again two more seconds because if i show you the exact same
data using a graph it's so much easier to come up with the answers right we can see right away
that corporate grants were consistently higher foundation grants we can see right away remained
relatively flat over time with that spike in may and that corporate grants exhibited a cyclical
pattern sort of an up up down pattern that repeated itself on a quarterly basis always
reaching the peak in the last
month a quart of the quarter and then declining dramatically
in the first month of the next quarter that pattern would be particularly hard to see
just looking at the numbers in a spreadsheet okay so not only is data displayed visually
easier to process it actually appears to have more impact so a june 2018 washington
post article describes a series of experiments in which data displayed in charts significantly
reduced the misperceptions of subjects both liberal and conservative
on importa
nt political issues so this chart that you're looking at right now
shows the results of just one of those experiments in this experiment republican respondents were
asked whether they believed that the average global temperatures were increasing decreasing
or staying the same over the past 30 years okay so the respondents were divided into three
groups remember the republican respondents there was a control group and they're represented
by the orange bars on these two bar charts they receiv
ed no information at all on global
temperatures they were just asked the temper asked that question about what the trend has been
in global temperatures over the past 30 years another group represented by the gray bars
was given a textual summary of the scientific consensus on global temperatures so they read
like a paragraph about it and then the third group those represented by the blue bars got a
chart showing temperature change since 1940 as measured by four climate institutions and as
you
can see the people who saw the chart again those people represented by the blue bars were the
least likely to draw the wrong conclusion and strangely enough that textual information
actually exacerbated the perception problem for republicans republicans who strongly identified
with the gop the results were exactly the same for liberal or democratic respondents on
other issues that they might get wrong they were much more likely to sort of correct their
perspective um after being presen
ted with a chart all right so you have these so-called superpowers
or secret weapons what are they good for let's consider what your greatest challenges
are and then we'll consider how to address those challenges with data biz okay so we're going to
just focus we have limited time so we'll focus on four really key challenges that i think most
of you can relate to um showing the need for your services to current and prospective donors
and funders showing the impact of what you do distinguish
ing your services from that of other
organizations and evaluating your progress over time and showing that progress to current
and prospective funders all right so this is really the heart of the presentation how to use
dataviz to conquer the challenges that you have one way organizations use data is to make the case
for the need for their services and i give you now an example from over 200 years ago it's one of
my favorite data visits and maybe you've seen it before it's kind of famous in
the 1800s florence
nightingale and a statistician named william farr analyzed army mortality rates during the
crimean war in england they discovered that most of the soldiers hadn't died in combat but rather
by preventable diseases caused by poor hygiene so nightingale decided to make her case with
pictures rather than tables of numbers or pros in her words quote to effect through the eyes what
we failed to convey to the public through their wordproof ears unquote her invention was the pol
ar
area chart um a sort of variant of the pie chart um that's what you're looking at right now each
slice the pie shows deaths for one month of the war growing larger if deaths increased during
that month and color coded to show causes of death so the blue portions show dust due to preventable
causes the red portions show dust due to wounds and the black portion shows deaths due to other
causes the queen and parliament could see at a glance the importance of hygiene and they quickly
set up
a sanitary commission to improve conditions and indeed death rate spell nightingale became
one of the first people to successfully use data visualization for persuasion
and to influence public policy of course there's lots of
different ways to show need you can show the prevalence of a problem
using icons this chart also uses color quite strategically and it may make more intuitive
sense than a bar chart because we're actually showing people it's not as abstract as a bar chart
where you'r
e just using bars to represent people um so it also takes advantage of the of actually
using 100 people rather than using percentages humans aren't so great at percentages but we could
kind of imagine a hundred people um of course this is kind of visualization you won't want to
overuse you can't have a bunch of icon arrays like this side by side will be overwhelming
but it can be very effective on its own this is what's called a quadrance chart
which shows the relationship between two measu
res and can be another great
way to show exactly where the need is so i created this chart for a client of mine who
was working with a number of school districts to implement what they call deeper learning
strategies and the first thing these consultants did was that they surveyed the principals and
teachers at the schools involved in the project and then they asked them questions to ascertain
how important each of the deeper learning strategies were to them and how often they already
put
these strategies into practice this chart shows the overall results for all principals and
teachers who completed the survey as you can see each dot represents one of the deeper learning
strategies and each dot is positioned to show the average importance assigned to the strategy and
the average frequency of practice of that strategy so the consultants could see really at a glance
that although teachers and principals already felt these strategies were quite important they were
putting them
into practice less than 50 percent of the time with the exception of direct instruction
so they could see the need was in addressing the obstacles that prevent that were preventing
these educators from doing what they already believed was really important they didn't
need to convince them of the importance all right um here's another great sort of
interactive way of showing need um the type of data that people are most interested in seeing
is data about themselves so this is a cool way to
put the viewer in the viz um you're looking at a
static version of this but it just asks a question in this case i'm just showing you a silly question
which is what percent of accents in canada involve moose and i've been told that
actually this is not a moose um i'm not an expert on the difference
between elks and moose so um i apologize for that but you can see you know you could
ask a question like what percent of people um are uninsured or whatever the issue is that
you're trying to sh
ow the need for um and then there's a little slider tool where they can put in
their guests and then when they scroll their mouse over the text in the lower right hand corner
they just get a simple chart that shows them what they guessed what others guessed and what the
data actually show so this is a cool way to sort of put the viewer in the viz ask them to guess and
then show them how they compare to others in the actual data it can be much more engaging than
just throwing a bunch of stat
istics at people another thing you want to show is the efficacy
of what you do um nothing is more effective than a bar chart it may seem tired and old
but the thing about bar charts is we all already know how to read them no one needs to
spend any time on figuring out how to read a bar chart um so the cognitive load is really low so
you can see this is a very effective way to show that you reached your goal in march and surpassed
it in april um this chart makes an even stronger point by all
owing a comparison of the trajectory
of one group that got an intervention to another that didn't get an intervention so we could see
the two groups really diverged after the introduc some kind of intervention was um introduced
um here's another way of showing efficacy this vis uses a number of great techniques
the chart is called a stacked area chart um it also uses a strategy called small multiples
which is placing several small charts side by side to allow for easy comparison and it also
uses that
trick of using 100 people rather than percentages the notations here are really helpful too
we can quickly see that poverty and child mortality have decreased dramatically over
the past 200 years and that basic education literacy democracy and vaccination have all
dramatically increased during this time period another thing we like to show our outside
audience is uniqueness how we differ from other organizations um this chart shows a
comparison of female and male students on wha
t advanced placement tests they take so it's
not showing non-profit sort of data but it's a great way of showing uniqueness in this case it's
showing how um female test takers are unique from male test takers um so if you kind of read it from
bottom to top you see that there are similar um percentages of male and female test takers for latin and statistics in us government but as
you work your way up the chart you can see that female test takers really diverge in taking
many more tests rela
ted to the arts and language here's another great way of showing uniqueness
this type of this is called a tree map don't ask me why because it looks nothing like
a tree um it compares one organization to others using both shape and color to show the number of
clients served great way to show that the youth together organization serves more clients
than all the other organizations combined okay now let's talk about evaluation and planning
that's something else you want to use data for um log
ic models um are basically what
you're looking at here a logic model is a just as a flow chart which is kind of the theory
behind how a program is supposed to work it shows the inputs the intermediate outputs and the vital
outcomes um you may be familiar with logic models they certainly get a lot of play in proposals
um but in my experience they often collect dust after that you know once you've convinced
the funder to fund you based on your logic model you kind of never see that electric m
odel again um
so i wondered uh what would happen if you plugged logic metal into real time data would that be a
really effective tool for evaluation and planning and this is what i came up with it's called
a living logic model again it's plugged into real-time data it shows progress to
date using color and it's interactive so you can see um on the logic model itself at
the top darker colors show where there's been more progress lighter color show where there's
been less fragrance progress
and when you click on any one of those measures represented by the
squares you get more data in the bar chart below you can also scroll over components to learn
more about the different aspects of the program and this information can
include images and web links okay so um you have some data and you visualized
it and you want to turn your good visualization into a great one well i have 10 data vis
adjustments for improving your visualizations let's go through each one um first is to clean
data we talked about dirty data before and it may seem obvious but before you ever visualize
your data take a look at it look for missing data duplicated data formatting issues for example with
dates everything rises on the quality of your data so don't skip this step number two suggestment is
to encode thoughtfully so what do i mean by encode so encodings are sort of the visual cues that you
choose to represent the words and numbers in your data and researchers cleveland and mcgill as
well
as some others have studied what types of encodings people are able to decode most
accurately and rank them in the following list so you can see that we make the most accurate
comparisons when we're presented with for example lengths along a common scale we're actually pretty
good at length even if it isn't along a common scale but we can only make pretty generic general
comparisons when we are encoding data using color hue volume or color intensity so for example if
you show me a shade of
green two shades of green i'd be hard-pressed to say that one shade is twice
as dark as another but if you show me two bars i could much more confidently say that one bar
is about twice as long as another so i can make sort of those more accurate assessments so
when you want your viewers to make those accurate assessments um use the kind of codings
towards the right hand side of the spectrum but when just generic comparisons are
sufficient you can use the encodings towards the left-hand en
d of the spectrum so pies of
course are delicious but pie charts are often inscrutable when applied to data so look at this
image um the pie chart and the bar chart are showing the exact same data and we can confidently
proclaim that the e bar is the tallest in the part bar chart but we'd be hard pressed to pick out
which of those pie slices is largest e d and c look about the same to me and that's because
we're not so good at angles um so when we're comparing the quantities of several thin
gs bar
charts are almost always better than pie charts the only exception is when you want to compare
a part to a whole um so but once you get beyond two or maybe three slices of pie um i would say
skip the pie chart and dust off the trusty old bar chart or use some other type of chart okay
suggestion number three is to highlight what's important um once you've visualized your data
the story can still remain hidden particularly to those unfamiliar with the data so that's when
it's time to
call out certain data points with color and annotations as i've done in this very
simple bar chart um here is some data from the cdc um on chronic illness which shows an example of
highlighting what's important in a chart this kind of chart is called a parallel coordinates graph
it shows the level or amount of something across several dimensions in this case we're looking
at the percent of adults with chronic illnesses across states in the united states each line
represents a different stat
e and as you can see along the x-axis or the horizontal axis we have
a number of different dimensions in this case different chronic illnesses the notations in
color draw attention to how low residents of hawaii are in relation to other states and rates
of depression um particularly in comparison to oregon which is has the highest rate of depression
um so i'm just calling the comparison i want the viewer to make is between hawaii and the rest
of the states so i just colored in hawaii and ma
de the rest of the states gray this is actually
interactive visualization where you can highlight one state and compare it to all the other states
in this case i just have hawaii highlighted okay suggestion number four is to show order ordered
data should be shown in a way that our perceptual system intrinsically senses as ordered conversely
unordered data should not be shown in a way that perceptually implies an ordering that
doesn't exist so um what you're looking at here is um a bunch of
bar charts on top and on
bottom um the ones on top show the exact same data on the the ones on bottom but the ones on bottom
are the bars are ordered in descending order and what we're looking at is the percent of health
visits that included depression screening across different care teams in family medicine pediatrics
and women's health and when the bars aren't ordered it's much harder to pick out you know
which teams did the best which teams did the worst but when i do it in descending o
rder it's much
easier to um make those comparison among teams my next suggestion is to clarify rather than
confuse with color so don't use the same color hue for two different variables on
the same chart or on side by side charts don't use the same color saturation
for different magnitudes of the same variable again in side by side charts don't use too
many color encodings on one page or dashboard and bright fully saturated color is kind of
like yelling you don't want to yell all the time
people will think you're crazy and stop
listening right so moderate color as you would moderate your voice only yell when
you have a really important point to make and then provide all the other information in
softer less saturated colors to provide contrast so here is a data dashboard that sort of
breaks all the rules that i just talked about it's really you know it gives me a headache
just looking at it it's using the same colors to mean different things in side by side charts
it's using
lots of great fully saturated color all over the place it's really just
kind of a mess and very hard to read um if these two color arrays looked the same to
anyone it might mean that you have red green color blindness um so the chart on the right shows the
full um rainbow array of colors and the chart on the image on the left shows the full array and
the chart on the right shows what that array looks like to someone with red green color blindness um
fun fact is that seven to ten percent of
men are red green colorblind it's a much lower prevalence
in women because it's actually a trait that exists on the x chromosome and the chances of it being
on both x chromosomes for women are really low but because a sort of significant portion of
the population has this problem we should keep that in mind when color coding our data i'm
color coding our charts if you put red and green next to each other it could undermine
the value of the chart for a lot of people these two maps show cdc
chronic illness survey
data um the survey data that we were looking at also in those um parallel coordinates graph and
it uses what's called a diverging color palette to emphasize contrast so the blue shades
show where the states in which rank high on two health indexes so basically the blue shades
where the art show where they're healthier adults the darker the blue the higher the percentage
of healthy adults similarly the orange shades show where there are lower um fewer people who
repor
t healthy lifestyles and then the gray tones are somewhere in the middle so by using
this diverging color palette we're able to pick out for example quarters of poor
health in certain parts of the country okay number six is to delete the unnecessary
if you've read anything by edward tufte who is sort of the grandfather of modern data
visualization you know about the data to ink ratio which should be as close to one as possible
um today we might think more about the data to pixel ratio um bu
t either way it's a good thing to
keep in mind um it means that we should eliminate non-data ink or pixels to help make
a visualization look more clean and to focus attention on the key data points
by eliminating distractive visual elements so here's again same chart on top and
bottom but on the chart on the bottom i've removed the grid lines the data labels i've
even removed the color legend and just directly labeled the lines with fewer distractions viewers
can compare the two trends muc
h more easily this suggestment comes from tamara munzer at the
university of british columbia um and she talks about how those 3d charts those options you see
in excel for example lead to distortions and and really make it more difficult for us to understand
the data so here's an example again the chart on the left and right are showing the exact same
data but the chart on the right uses a drawing technique called foreshortening in which you
reduce or distort in order to convey the illusion
of three dimensions so parts that are supposed to
be perceived as closer in space are made larger you know so that red slice is made larger in this
pie and parts that are supposed to be perceived as farther away are made smaller those are
the red i mean the green and blue slices and as you can see the angles represented
on the 2d chart on the left are distorted on the 3d chart on the right as a result of that
foreshortening making it difficult to judge the relative size of the slices anoth
er technique
to create the illusion of three dimensions is to obscure some objects with others to make it appear
that some are in front of others but of course this is a problem for accurate assessment in data
vis in the top chart the green bars are barely visible rather than have that third axis you can
just use color which the charts already are using to distinguish those three groups a b and c um
really that third axis is totally unnecessary and in the turn on the bottom it's much easier
to determine um that the a group was higher in october than it is um using the 3d chart the a
group was higher than the c group i meant to say okay um this one also comes from tomorrow monster
is beat memory which simply means it's easier to compare two charts that are side by side than to
hold one in memory while you're looking at another one so um this is the power behind those small
multiples charts that i talked about before um in small multiples charts a bunch of small charts
are lai
d out together to make it easy to scan over all of them and to make quick comparisons
this is a great small multiples chart by doug mccune you might want to check out his
blog he's a really good data visualizer so on the x-axis of each one of these charts is
time so we have the times of day and then on the y-axis we have number of crimes um and the daytime
crimes show up on the top and they're represented in yellow and the nighttime crimes are represented
by the blue bars on the bottom half
of each chart so it allows us to quickly see that driving under
the influence and drunkenness occur more often during the night and that suicide and
trespassing occur more often during the day okay suggestion number nine is to zoom in so when
we are showing charts one by one um in a series we should always start with the most general
chart and then zoom in a more specific data so here's an image of a series of visualizations
which i created two of which you've already seen using that cdc c
hronic illness data so the
maps give you a general sense of where there are health concerns according to the data set such
as that corridor of poor health in the southeast the parallel coordinates chart helps to compare
states on specific illnesses and conditions and then there's a last dashboard which i have not
shown you which allows viewers to dig deeper into trends and look at possible predictors of good
and poor health using some interactive features okay my final suggestion to you is
that
visualization is not always the way to go um there are at least two situations in
which i think a table works better than a chart the first is that you already have a
really engaged but diverse audience so these folks are highly motivated to access certain data
and won't be annoyed by having to find the data that they want on a table but they're
also diverse in their interests so different people are going to be interested
in sort of different columns in your data set um tables use pa
per or screen real estate really
efficiently you can fit a lot of rows and columns into a small space allowing users with different
interests to find the data that they want the other case in which you might want to use a
table rather than a chart is when you have lots of units of measure so for example you want to show
the height weight location and satisfaction level of 100 participants in a healthy eating program
this data involves four different units of measure we got inches pounds lat
itude and longitude and
survey ratings such complexity is really difficult to represent in a single visualization because
you have to come up with different encodings for each one of those units of measure but you can do
that quite easily in a table so you might want to use a table under those circumstances okay we're
back to our journey and we're pretty far along by now um i just want to mention some resources
before i open up for questions and answers so um you might want to check out thi
s website
i think jenna's going to put the link up up for you um here's just a sampling from an online
catalog of database software and online tools that you might want to check out so if you're
wondering what you should use to visualize data sort of beyond excel and google sheets you might
want to check out some of these software programs also here's a list of free data
visualization tools that are out there um the one that i use the most is tableau public
um tableau also has a subscripti
on or paid version um but the tableau public the free version
does a lot of the same things that the um paid version does the primary differences are that
you can't save what you create on tableau public to your own desktop or server you have to save it
to the tableau public server so it's not good for very sensitive data because it is a public server
but you can make whatever you create invisible to anyone who doesn't have the url for it so
it's not like someone could google and find it al
so there are fewer data connection options
in tableau public than the paid version in public you can connect to excel and google
sheets and some other common data sources but with the paid version there's a long list
of databases you can connect to in tableau i also highly recommend recommend this simple
um sort of decision tree from andrew abella it's called chart suggestions of thought starter
you're also going to get the link for this um it's based on gene zelazny's classic work
called
saying it with charts um and the decision tree as you can see starts right in the middle
just ask you what what would you like to show and it gives you four options comparison distribution
composition and relationship once you answer that question it kind of leads you
through enough a few other questions about the nature of your data to help
you arrive at an appropriate chart for what kind of data you have and what you want
to show but there are some other really good decision trees out the
re as well this is probably
the simplest and i suggest you start with it um there's also this data visualization catalog
which is simply kind of like a dictionary of different charts so you can click on a chart
and find out more about what it's best to use for um here's some people that you might want to
follow and read i already mentioned edward tufte a couple other names i'll point out on the
list um include ann emery and john schwabisch the other people on the list um most
often visuali
ze more like corporate data or different kind of national data and emory
and john shawbush focus more on public sector and non-profit type data but they're
all you can learn a lot from all of them um again my consulting business is called
database for nonprofits i provide a number of different services for nonprofits to help them to
visualize and thus make better use of their data so feel free to check out my website
you're getting the link for that probably the main thing i do for clients
is help
them to develop interactive data dashboards i often use tableau either the free or the paid
version um to help organizations do that and a lot of my organizations particularly the smaller
ones um really make good use of tableau public i create something really sophisticated for them
um a at the end of the consultancy i teach them how to refresh the data if it's not connected to a
live data source and then they're sort of off and running they don't ever need to learn tableau in
orde
r to use it to show their data the interactive visualizations i create can be embedded in
websites or can be just used for internal purposes um i also have a blog called 60 second data tips
um you're going to get the link for this you can sign up to get a free 60 second data tip each week
you can also check out my website for past data tips they indeed only take about a minute to read
and it's sort of a painless way to sort of up your day-to-day game um slow um not even slowly kind of
quick
ly over time with a small investment of time okay we are at discussion um so i think um
jenna's gonna moderate for me and let me know if there are any questions that i can answer
because i'm not looking at the chat window okay can everybody here and see me okay awesome well amelia that was such an amazing
presentation i am such a data nerd and this is kind of like data and art mixed in all together so
thank you so much for all this useful information and everybody i hope that you're feeling
equipped
with all the knowledge and tools and confidence to start utilizing the power of visualization
data visualization for your organization so before we jump into our q a i want to announce
that donorbox has launched an online platform called the donorbox knowledge community and
this is a place where you can get focused answers to your questions regarding marketing
web development fundraising board management data visualization and so much more and we
are offering exclusive access to t
his invite only community for our webinar attendees today
and the sign up window will remain open for just 30 minutes after this webinar so you can
click the link that just appeared to join and also if we do not get to your question
today this is a great place for you to post your question for amelia so go ahead and join
the community you'll also be redirected to the community after the webinar ends and utilize
all the amazing tools that we have there for you all right so i'm gonna go ahead
and see what kind of questions we have and it looks like we do have a few uh from keith
here it says do some organizations worry about branding with data so colors presentation etc
oh yeah absolutely um whenever i do work with an organization i one of the first things i do
is ask for their style guide if they have it if they don't have a style guide i just ask them you
know what fonts do you use what colors do you use um so yes it's i think it's quite important to
use the um the brand loo
k in the visualizations themselves and programs like tableau allow you
to you know if it doesn't already have your font or colors already in the program you could
import them um you can also import custom shapes into programs like tableau so they have all kinds
of ways to make sure that you stick with your brand look with your visualizations great question
awesome yes that was a great question thank you so much for that and jennifer is asking can tableau
uh public do interactive visualizati
on yeah absolutely i mean all the interactive features
that are in the proprietary version are actually in the free version as well so and it's a it's a
really really strong free tool um what a lot of my org clients do who want to use it um what they
do is they provide me with de-identified data so they remove any you know they're providing with
with data on their clients for example they remove their names their birthdates of course social
security numbers addresses anything that would all
ow someone looking at the data to
identify the individuals in the data and so once you have a really good de-identified
data set it's pretty safe usually to put that on a public server like tableau public again i still
wouldn't do it with health or maybe education data but if it's just like participation data
and organization a lot of organizations use tableau public for that um also you should
know if you want to use the subscription version tableau provides discounts for nonprofits and fo
r
organizations under a certain budget level don't quote me on this but i think it's around 5 million
they'll provide some free subscriptions as well awesome thank you so much for that information
and everybody all these wonderful resources that amelia has provided for us today we will include
in our post webinar email so i know many of you have clicked these links but we will also provide
those for you to explore later on and it looks like we have a question from adam and you may
have alr
eady answered this what program app or software creates that interactive charts like the
moose and elk question that you showed earlier um yeah i should have shown
a better version of that um yeah i use template public for that so yeah
also yeah perfect okay that was easy enough now let me scroll through my chat and make sure
i'm not missing any questions uh this was such a thorough presentation and there were a lot of
questions that i had that you answered throughout your presentation so u
h thank you so much for uh
all of your work with that and let's see here okay scrolling through our chat and i think that
we i think we're good anyone uh we do have one minute left uh so that was perfect timing and
i just want to thank you all for joining us today for this webinar and amelia for sharing your
passion and knowledge with us i know i'm feeling inspired i hope that all of you are as well and
thank you again we are so proud to provide the tools and resources you need to help you
help
others thanks everyone thanks jenna [Music] you
Comments