Best chart to use

[GSPremington]

I want to create a HIGHLY aesthetically pleasing chart (suitable for permanent display) that shows the time series pattern of a respondant's beliefs or thier confidence level in their predicted outcome of an event. Data will be collected over time as the event unfolds.

A simple example might be asking an American football fan to record whether they predict thier team to win, lose (or draw) and how confident they are in thier predicted outcome multiple times throughout the course of the game. Data captured every 5 minutes of the game yielding 10-20 data points of (outcome, confidence level).

Some considerations
-- The end chart might show just 1-2 respondants or it could show as many as 10-15.
-- outcomes are fixed binary (win, lose or draw).
-- Conviction in the outcome is 0-100%. 0% conviction win = draw.
-- # of times data is collected - As few as 5 as many as 40.

Thank you all for any help.

Edit: Please see my response (post 3) to Luke M in which i clarify the type of chart I am seeking.

 

[Luke M]

Since it's a time series, should be plotted left-to-right. As it's reasonable that the data points are connected dynamically, and not quantum leaps, lines connecting the points are reasonable (straight, not curved). With y-axis on left, no need to have a limit on # of data points.

Do we need to compare different respondants? Or just need to see them all? If former, can plot on single chart, but would want to limit chart clutter (thin lines, gray/no gridlines). If the latter, panel charts would work.

To show the progression, I'd use a positive/negative y-axis with the axis numbers custom number formatted to show win-lose-draw.
In this pic, I'm showing that Bob starts out thinking his team will win, but things don't look good at halftime. Then, his confidence is restored.



Does this help?

 

[GSPremington]

Luke- Thank you for your prompt reply.

I should have (and will go back and edit) clarified that I am looking for HIGHLY aesthetically pleasing charts that can become permanent display pieces .... think of it as a potential art piece for the end consumer who will want to look back at this data over time perhaps displayed in a frame.

The football analogy was to capture the sprit. Hope i did not mislead. My statement above is not to criticize the chart above that does accurately reflect what i am trying to show. Its just that I am hoping that some of you awesome excel folks can point me in the right direction as to a cool way to use a dynamic excel chart like a radar, bubble or donut that can show both the direction and intensity of the persons outcome expectations and how it changed over the course of time till the event was decided.

Edit
I know bubble charts could be used to show direction (win/lose) and intensity (confidence level) but not sure how that would work charted over time ... such as a series of periodic bubbles.

Perhaps radar charts. looking for insightful and pretty way to capture this.

 

[Luke M]

No offence taken GSPremington. I will say, you're request is a reverse of my normal thinking, but I understand that often management/customers want "pretty" over "informative".

 

[NARAYANK991]

Hi Remington ,

Can you clear some confusion on my part ?

You wish to plot something over a period of time ; so can we take it that the horizontal axis is time ? Or is it just a number signifying the response number , starting with 1 , going up to 5 or 40 ?

Can we take it that a loss or a draw are equivalent , since you say it is a binary response ? So one prediction is a win , while a prediction of loss or a draw are the same.

The chart can have the varying responses of one person plotted as a series ; each chart can have one or more series , going up to even 10 - 15 series. Clearly , a chart having 40 multiplied by 15 points will not convey any information aesthetically.

Assuming that , as you say , a bubble chart is used , and we have conviction ( from 0% to 100% ) controlling the bubble size , while the outcome itself ( win or loss ) is plotted on the positive side of the X-axis or the negative side of the X-axis ; what exactly would 30 or 40 bubbles of varying sizes in the upper and lower half of a chart convey ?

Which brings up the point , what exactly are you looking to convey with the help of this chart ?

I would think that people's predictions should be juxtaposed with the actual results of the games !

Narayan

 

[GSPremington]

Hi Remington ,

Can you clear some confusion on my part ? I will try to unconfuse what i asked. responses in blue.

You wish to plot something over a period of time Yes, responses will be gathered over the course of a set period of time. ; so can we take it that the horizontal axis is time ? It could definitely be a horizontal axis (left to right) type of time series chart for sure. Alternatively, i would be interested in some other unique chart (such as a circle reflects time periods from start 0:00 to end 12:00 in which any responses obtained during a specific time window are displayed in that window). Or is it just a number signifying the response number , starting with 1 , going up to 5 or 40 ? It may be there are only 4-5 times in which responses are captured or it could be as many as 25-40 periods depending on how often surveyed. On average i will probably have 4-6 defined periods on average with a high of 8-10.

Can we take it that a loss or a draw are equivalent , since you say it is a binary response ? So one prediction is a win , while a prediction of loss or a draw are the same. Yes, its either "win" or "loss" and they are equivalent. I would also allow for a middle ground of no conviction (neutral).

The chart can have the varying responses of one person plotted as a series ; each chart can have one or more series , going up to even 10 - 15 series. Clearly , a chart having 40 multiplied by 15 points will not convey any information aesthetically. yes, i can see how it will become overly busy and distracting. Seeking any solution that can make the underlying data patterns (information) both elegant and apparent as you step back and look at it.

Assuming that , as you say , a bubble chart is used , and we have conviction ( from 0% to 100% ) controlling the bubble size , while the outcome itself ( win or loss ) is plotted on the positive side of the X-axis or the negative side of the X-axis ; what exactly would 30 or 40 bubbles of varying sizes in the upper and lower half of a chart convey ? Yeah, i dont know that bubble charts are the optimal tool. Just know that i have seen them used to display intensity (conviction) of a data point.

Which brings up the point , what exactly are you looking to convey with the help of this chart ? I'm not trying to be overly vague or obtuse here, but what I am trying/hoping to show in this chart is the overall (and/or filtered by individual respondants) the sentiment fluctuations of respondants and how it changes over time as they participate in a process and wait for an outcome to be realized at the end. One such use could be to display a team's perception of a project's ultimate "success" or "failure" outcome during the course of a development project. Another could be my football game example. A third could be asking parents (friends/family) to record if they are having a boy or girl throughout pregnancy.

I would think that people's predictions should be juxtaposed with the actual results of the games ! Yes, the game example may have painted too black or white a picture as the outcome progress is real time marked to market as the game progresses, so i am more interested in scenarios where the outcome is much less apparent. What i really care about is the respondants beleifs over time as they wait for the outcome. Capturing that is simple. making it informative and interesting to look at is harder (for me).

Narayan

 

[NARAYANK991]

Hi ,

Thanks for the clarifications ; I'll certainly think of what can be done , and I hope others will pitch in too.

Do you happen to have come across any specimen anywhere that we can possibly replicate using Excel ?

Narayan