One of the most time-saving techniques that I figured out over the past few years is how to integrate paired samples t-tests into my interactive dashboard. Think about it: By including t-tests, my dashboards are basically calculating scores and scores of paired samples t-tests behind the scenes!

​For instance: Say you want to know if females, in particular, showed statistically significant gains? Boom!...Click on the slicers in my interactive dashboard, and you'll get your answer in seconds! What about females who are underrepresented minority students and who are majoring in Physics? Yep...my dashboard will calculate that for you in an instant!

Want to know the steps?  

To include t-tests, I essentially took the PARCS method and kinda, sorta bent it. In other words, I flipped the script on the PARCS method! Here are the steps:

​Step 1. Insert a pivot table

First, select your raw data...go to your "Insert" ribbon...and click on Pivot Tables.

Helpful Hint: Notice, in the screenshot below, that my raw data includes ID #s for each student in column A, Pre and Post data in columns B through M, and demographic data (major, gender, and race/ethnicity) in columns N, O, and P.  

Once you click "Ok," you should see 2 slicer menus. Start playing with them by clicking on various combinations. If you did this correctly, you will notice that your pivot table is being modified in real time. So, if you selected "Female" in your Gender slicer, the pivot table will showcase only participants who are female. So cool, right?

***Essential: Before we can run calculations, did you notice that there is a "Grand Total" in your Pivot Table? It appears right at the bottom of your Pivot Table. See it? Ummm...we need to get rid of that, like, pronto because it will mess up all of our calculations! To quickly remove "Grand Total," simply right click where it says "Grand Total" and select "Remove Grand Total." Easy fix!

Step 4. Create calculations

Now, we get to calculate statistics! You knew it was coming, and here it is!
We're going to create a summary statistics table with descriptive statistics and t-test formulas.
Create this summary table either on a separate sheet or in the columns to the right of your pivot table (i.e., next to your pivot table). You DO NOT want to create your summary table underneath the pivot table! Bad...very bad!

In the screenshot, below, I created my summary table in column E. For the summary table, I calculated the following:

• N
• Mean
• Standard Deviation
• T-test  

Here are all of the formulas that I used for my particular dashboard:

• Pre N: =Count (B:B)
• Pre Mean: =Average (B:B)
• Pre Std. Dev: =Stdev(B:B)
• T-test: = TTEST(B:B,C:C,1,2)
• Significant?: =IF(F11<0.05, "p<.05","not significant")

Go ahead and click on the slicers and watch your summary table change before your eyes! (I've done this hundreds of times and, trust me, it never loses its magic!)
​

After fiddling with my pre/post chart, I then cut and pasted my chart to a new sheet and called this sheet "Dashboard." I also cut and pasted my two slicers to this new sheet. Once I placed the chart on the dashboard, I then included the t-test results by referencing them from my pivot sheet. 

And, here's the finished product! I added icons and created a cohesive color scheme derived directly from my clients logo. 

Once you have created one set of formulas in your summary table, it's pretty easy to drag your formulas across columns to add more pre/post items. 

Feel free to download a copy of this dashboard: 

​Early one morning, I received a frantic call from a client:

• Client: "Hi, Shelly, ummm...I know it's early, but I've been playing with your dashboard and I noticed that ALL of our female students are dissatisfied with our program. Am I reading this accurately? What is going on with our program? Did we fail our female students somehow?"
• Me: "Good morning! Hmmm...ok...bear with me as I open the dashboard. [Long pause] Okay! Got it open!
• Client: "Thanks, Shelly. I clicked on "Astronomy" majors and "Females" in the slicer menus. When I do that, it looks like 100% of participants said that they are dissatisfied with the overall program. What the heck?"
• Me: "Hmmm...I'm not seeing that. Did you, by chance, filter the data by race/ethnicity using the Race/Ethnicity" slicer?" 
• Client: "Yes...I think I selected 'Latinx/Hispanic.' "
• Me: "So, the total number of females who are Astronomy majors AND identify as Latinx is 4. So, we're actually only talking about 4 students here. Astronomy is the smallest major in the college, and there are very few Latinx students in the department. 
• Client: Oh...only 4 students?!?! I wish that was clearer in the dashboard.

​Noted!

Without knowing the denominator, my client freaked out over 4 students. I can't blame them. Context makes a BIG difference. Often times, dashboards fail to relay this "context" to the user. I would argue that interactive dashboards, in particular, require context to ensure proper interpretation of the data. 

Adding a dynamic "n" count to your dashboards is critical, but it's a little tricky to create in Excel. ​Here are the steps....

Step 1. Insert a pivot table and add a demographic or ID variable to the "value" field

First, select your data...go to your "Insert" ribbon...and click on Pivot Tables.

Once your pivot table has been created, drag and drop a demographic variable or, better yet, an ID variable into the "Values" field. You, basically, want to use a data variable that is "complete." I almost always have an "ID #" column in all of my data sheets. Sometimes the ID #s reflect actual student IDs. But, often, I just make up ID numbers for the purposes of creating a dynamic n count for my dashboards. Sneaky, right? 

Step 2. Change the value from a "sum" to a "count"

By default, Excel will try to "sum" your ID #s. Don't let it! Change the "sums" to "count" by going to the Analyze ribbon and selecting "Field Settings." Within "Field Settings," you'll want to summarize your value field by "count." In other words, select "count." Simple enough? 
​​

Step 3. Insert a Slicer (or make sure that your pivot table is connected to your slicers)

As I've described in Tip #1, Excel slicers make the dream work. Slicers will allow you to slice and dice your dashboards. Insert your slicer by going to the "Insert" ribbon and selecting "Slicer." Once selected, you will be asked to select the variables that will serve as your filters. I usually select demographic variables like gender, race/ethnicity, etc.

Step 4. Insert a text box

We are going to use a text box to feature our new "n" value. Then, we're going to place this text box somewhere in our dashboard to give the user context. Insert a text box by going to the "Insert" ribbon and selecting "text box." Draw your text box to size.

Step 5. Add the dynamic "n" count to your text box.

All right, folks. This is where a lot of people trip up! Excel is finicky sometimes and plays by its own set of rules. To add the dynamic count to your text box, you first need to reference the n count in a cell. In this example, below, I've referenced the n count in cell B66. See that?

The reason why we need to create a referenced cell is because an error message pops up when we try to directly reference a number from a pivot table in a text box. It's a strange thing, I know! Let's just go with it... 

Once you have your referenced cell, then simply select your text box and select your referenced cell. In this example below, I selected my text box and then inserted the formula "=B66" to reference the number in cell B66.
(Place $ signs around your formula to further fix the value: "=$B$66")
​​

The final step is to move your text box to your dashboard. Try it out! Did your n count change as you select the slicers? If not, you might need to fix the formula in your text box to refer to the correct cell. 

These steps are not easy to master...but, including a dynamic n count is going to help you avoid frantic morning calls from clients. ;-)
​​

Once you've created your interactive dashboards in Excel using my famous PARCS method, you will be tempted to share your Excel workbooks with clients and colleagues. You designed something amazing, and it only makes sense that you would want to share it with the world, right?

HOLD ON! 
Aren't you forgetting something???

 

Theoretically, if you shared your Excel workbook as is, your clients and colleagues will have access to not only your amazing dashboard, but also the raw data and all of the pivot tables that were used to painstakingly build your dashboard. Do you really want to give them access to all of this? More importantly, aren't you violating FERPA, HIPA and/or IRB rules and regulations by freely sharing the raw data? Ouch!

Here is a step-by-step guide to protect both your dashboard and your Excel workbook:
​
There are two aspects to protecting your dashboards:

• Protecting your workbook structure, and
• Protecting the integrity of your actual dashboard

Let's start with the first one: Protecting your workbook structure
​

• Step 1: To make sure that your user does not have access to your raw data and pivot tables, you're going to want to first "hide" those sheets in your Excel workbook. To hide a sheet, just right click on it and then select "hide" from the dialogue box. Easy enough?

 

• Step 2: Once you have hidden all of the sheets that you do not want users to gain access to, the next step is to password protect your workbook.  To do this, go to the Review ribbon and select "Protect Workbook." In the dialogue box, make sure that "Structure" is checked. Next, enter your password and viola! If you have done this correctly, your user will not able to unhide any of the sheets in your workbook. Try it out for good measure!

​

​Let's do the second one, now:  Protecting the integrity of your dashboard

Sometimes, when a user is playing around with the dashboard and slicing and dicing the data to their hearts content, they might encounter two nuisances: one, the "Format Chart Area" side bar might pop up when they click on specific charts; and/or two, they may inadvertently move your charts and figures around on your dashboard. Ugh!

Both of these things have happened to me...and my clients thought that the dashboard was broken or "not working correctly." 

Please learn from my mistakes and do the following steps to avoid these mishaps:

• Step 1:  Unlock your slicers before protecting your sheet. You still want the user to use the slicers, right? If you protect your sheet  before unlocking your slicers, the user will NOT be able to slice and dice the data. Here's how to unlock your slicers:​
  • First, select all of your slicers by holding down the shift key. Then, right click on your slicers and go to "Size and Properties." Once you select "Size and Properties," a side bar called "Format Slicer" should appear. Within this side bar, you will want to go to "Properties" and UNCHECK "Locked."

***As a bonus, consider fixing your slicers in place. That is, if you want to keep your slicers from moving, you will need check the "Disable resizing and moving" option. To do this, right click on your slicers; select "Size and Properties"; go to "Position and Layout" and make sure that "Disable and resizing and moving" has a checkmark next to it. 

Another fully protected dashboard has been born. You're welcome, HIPA, FERPA and IRB boards!

Do you use traditional pre/post surveys to assess programmatic impact on participants?  If so, you might have encountered the following known problems:

​Retrospective Surveys to the Rescue!

Retrospective surveys, by contrast, ask participants to compare their attitudes before the program to after.  Because a participant completes a retrospective survey in one sitting, responses are more complete.  Not only is there a higher completion percentage with this method, but it also has been found to reduce the Response Shift Bias in participants. Also, isn't it A LOT easier to administer one survey instead of two? You bet! Evaluators love the simplicity of one survey...and instructors and program directors do not need to sacrifice as much time to administer the survey. Less burden all around!
​
Major Advantages...and more Accurate?

In several of my projects, the retrospective survey had clear advantages over the pre/post survey. It yielded more complete datasets and higher response rates. Also, the data from retrospective surveys seem to align more "accurately" to qualitative data.

• For example, in one evaluation project several years ago, participants were telling me, via focus groups, that the summer workshop helped strengthen their confidence and content knowledge in computing. They loved it and are eager to dive deeper into computing!
• But, the traditional pre/post survey did not show this. The pre/post survey actually told me that participants' computing skills were decreasing from before the workshop to after!
• Why did this happen? Well, we can thank the Response Shift Bias for that one! Humans have a tendency to overestimate their knowledge/confidence in a field because they are largely ignorant of the vastness and complexity of that field.
• To use a personal example, if you had asked me 10 years ago how good I was at Excel, I would have said "very good." Now, I would rate myself as "average" even though I currently teach courses in Excel and design some pretty intricate dashboards in Excel.
• Basically, my understanding of the possibilities has fundamentally changed. I know what's achievable in Excel, and I've been exposed to true experts in the field. Only a retrospective survey would "accurately" capture this shift in mindset. A traditional pre/post survey would give evaluators a false sense that my skills and confidence in Excel went from "very good" to "average" over a 10 year period. I was, like, a different person back then, and my understanding of the rating scale (1, poor to 5, very good) shifted. Makes sense?

Any Disadvantages?

Are retrospective surveys the answer to all of our problems? Well...not exactly! There are certain conditions that make the retrospective survey unusable and frankly inaccurate:

1. Younger children have difficulty with retrospective surveys. Their memories fade quicker and they forget how they felt before an event occurred. What can I say: Kids live in the moment! That's a good thing, but bad for retrospective surveys, I suppose. Younger children also get confused by a retorspective survey. They need major 'handholding' as they complete the survey.  
2. Memory is less and less accurate the further removed we are from an event. Our memories are faulty and we are subject to false memories (Loftus & Ketchum, 1994). The further we step away from an event, the more hazy our memories get. Retrospective surveys are best to use for short-ish interventions. A 1 week workshop is an ideal set-up for a retrospective survey. A 3 month summer workshop is also pretty good. A semester course? Yeah...that usually works. A one year course? Maybe not.
3. Retrospective Surveys require reflection...and some people have an easier time with this than others: A retrospective survey forces a participant to reflect on their experiences over time. Some people are eager to contemplate how they've changed over time. (Psychology majors live for this stuff!) But, for others, thinking about the past is more difficult and can be off-putting. A painful memory or an unpleasant experience might deter people from looking back. We don't want to change for the worse, right? There might be a subtle positivity bias that emerges when responding to a retrospective survey. That is, it could be that some participants are motivated to show growth from before to after. This "growth" might, in fact, accurately reflect how much they've changed as a result of the intervention. (That's usually the case, I've found!) Alternatively, it might not be accurate; the positivity bias made them do it! It's really hard to tell. Even retrospective surveys have their faults, people! 

More research is clearly needed on this topic. But, there are so many advantages to using retrospective surveys over traditional pre/post surveys;  the good might outweigh the bad.
​What do you think? ​

Here are the steps that I used to create my interactive dashboard. To help me remember all of the steps, I affectionately refer to this method as the PARCS method. (Shoutout to my favorite TV show, Parks & Rec, for inspiring the acronym!) PARCS stands for:

P- Pivot
A- Analyze
R- Rename
C- Chart
S- Slice

​Let's do an example, together...

​Step 1: Insert a Pivot table

​To insert a pivot table, first select your data. Then, go to the INSERT ribbon and select Pivot Table in the upper left-hand corner. A dialogue box will open. Make sure that you place your new pivot table on a "new worksheet." Select "OK." Step 1: Done!
​

 

Ready for the magic?

Step 5: Slice your data

​The secret ingredient when it comes to designing an interactive dashboard in Excel is something called "Slicers." Ever heard of it? Yeah...I didn't discover it until 2-3 years ago, so don't feel bad. Slicers make the dream work. Here's how to insert a slicer:

First, place your cursor anywhere in your Pivot Table. Then, go to the INSERT ribbon and select "slicer." (Yep...what did I tell you! You're basically going to live in the insert ribbon.). Once you select slicer, Excel will ask you to check off the variables that you want to slice and dice the data on. In my case, I selected Major, Gender, and Race/Ethnicity. Once selected, I hit "OK."

Bam! I have 3 slicer menus! So cool, right?   Play around with your slicers and watch your chart and pivot tables change before your very eyes. 
​

Final step: Cut and Paste your charts and slicers on another sheet and call that sheet "Dashboard"

That's all there is to it! Just remember to "cut" not "copy" your charts and slicers. If you have two copies of things in your workbook, Excel tends to freak out a bit. Excel doesn't like twinsies!

If you are curious to know how your slicers are connected to your dashboard, you can right click on one of your slicers and go to "Report Connections." Within Report Connections, you should see a list of all of the Pivot Tables in your entire workbook. (This is why we did Step 3, btw! Without renaming your Pivot Tables, you will get a long list of numbers. Not good.) To further customize your dashboard, you can select which pivot table goes with which slicer by checking and unchecking the boxes within the report connections dialogue box. 

Happy dashboarding!
​Check out some of my other interactive dashboards here!