Posts

Posted on

Flowcharts in Word

I thought I knew pretty much everything about using MS Word but these guys showed me something new. Check out the flow charting video by MakeUseOf blog. https://youtu.be/iiS7aAFI2Cs

Posted on

Adapting Flow Charting to Protocol Analysis

I’ve been working on a project using the think-aloud method of protocol analysis lately. What’s that, you might ask. Protocol analysis is a qualitative process by which one tries to develop a set of steps, the protocol, for some sort of problem solving by tracking how people solve a problem. The think-aloud method asks people to say what they are thinking while solving the problem. Does that make sense? So if I was to try to write a computer algorithm to solve a multiplication problem, I might ask a bunch of people to tell me what they were thinking as they work the problem, then I would look for similarities and patterns in what they did, and develop step-by-step procedure to solve multiplication problems.

That’s what I’ve been doing with presentations of research results. I asked people to watch some presentations and tell me what they were thinking as they interpreted the slides while I recorded what they said. Then I made transcriptions of the recordings and ended up with close to 500 unique phrases. Yikes! Now what was I supposed to do with all those phrases?

I ended up using flow charting to chart each person’s thought process for each slide. I used standard flow chart symbols to represent the stages and processes involved. I did this by hand (because it was easier than inserting a gazillion shapes and symbols in Word) on a separate piece of paper for each slide, then I was able to sort the papers and slice and dice the information as I saw fit.

Here are the flow chart symbols that I used and how I applied them. Let me know if they are not self-explanatory.

Posted on

Best kind of graph

Ever wonder which graphs are easiest for people to understand? Research orders them like this (left to right, top to bottom):

Use the info in this chart to design graphs that are easily understood by viewers.  Notice angles and area down in the middle.  Maybe there is a better way to display your results than by using a pie chart.  Just a thought.

Posted on

Flow Charting 101

Are you trying to figure out how to do a flow chart in Word? Here is a good place to start:  https://www.lucidchart.com/pages/how-to-make-a-flowchart-in-word#top-info.

Word doesn’t have an actual flow chart that is designed to illustrate some sort of algorithm but you can start with a basic hierarchical structure.  To find these go to the Insert SmartArt command and choose your hierarchy.

Once you have the basic chart in your document you can change individual shapes with the Change Shape command on the SmartArt Tools Format ribbon.  You’ll find a whole group of traditional Flowchart symbols in there.

Sorry the images aren’t perfectly clear but I hope this helps.

Posted on

Creating a List Box in Excel

Posted on

Adventures in Coding

I hate doing things twice.  Three times is even worse.

I’ve been working on coding some transcribed audio.  In the past, I have always transcribed directly to a Microsoft Word table, then imported the table into Excel.  Yes, it’s a little redundant but it’s easier for me to type away in Word, tweak the structure to fit Excel, then move it over compared to typing directly into the cell structure of Excel.

Anyway, this time I went searching for some sort of open-source qualitative data software.

After several false starts, I settled on QDA Miner Lite.  The website had a nice chart comparing features of the free Lite version and the full version which gave the impression that the program would do what I wanted.

Several hours of several days later, I had modified my original files to upload into QDA Miner Lite, created codes and variables, and coded the bulk of my data files (or cases as each participant is a different case).  The program displays a fun color-coded graphic of the frequency and duration of each of the codes.  It also gives a count, % of codes, cases, and percent of cases breakdown.  Unfortunately, that’s about all the analysis you can get from the free version of the program.   It will also create some graphs based on the usage of codes. There is no capacity to use the variables that I had painstakingly set up either by sorting or filtering.  As near as I can figure out,  other than some nice bracketing and highlighting of my text, that’s all the lite version will do.

Long story short…I’m back to using Excel to code my qualitative data after multiple attempts to find a better way.  Yes, it seems counter intuitive to use a program designed to do math for what is essentially a verbal task but with Excel it is easy to count usage of a particular code, easy to change codes, easy to use multiple codes for the same passage, and easy to graph my results.  All while still being able to pull quotes from the text to enhance my report when the time come.

 

Posted on

Using Audacity to Transcribe Audio Recordings

I’ve been working on a protocol analysis project lately.  A protocol analysis is where you get someone to describe their thought process (the protocol) as they solve a particular problem.  In this case, I am using the Think Aloud Method in which my subjects are shown a set of images and asked to think out loud as they decipher the images.  It’s pretty easy for them (I think) because they just need to talk out the running stream of thoughts in their heads.  These streams of consciousness were captured in audio recordings.

The streams of consciousness become data when they are transcribed onto the digital equivalent of a piece of paper. Transcription is a rather onerous project because you need to be able to repeat small chunks of the recording to type out the words exactly as spoken.  It helps to slow down the speech of some participants because most people think and speak a lot faster than I can type—and I can type pretty fast.

The program that opens the audio files by default on my computer is pretty useless for doing these things so I went on a search for something better. Naturally, I would rather not pay for whatever program I might find so I limited my search to open-source software and found Audacity.  Woo hoo! Luck would have it that my first downloaded program does exactly what I want.
You can use the slider on the right to slow down the recording if needed.  This makes the voices sound pretty funny but it’s a little easier to keep up.  Then you hold down the shift key and click the green arrow next to the speed slider.  The green arrow changes to a little loop circle and a text box appears to say “Play-at-Speed/Looped-Play-at-Speed”.  That’s it.  The audio will loop the selection at the speed you select until you manage to get the transcription typed out verbatim. When you finish transcribing your selection, use the little hand mouse pointers to move the start and end of the selection along and repeat the process until your transcription is complete.
My needs are pretty simple but here you can see that I have a file open. In the center panel, you can see that I have made a selection of the audio by the lighter background.  The selection is for about 30 seconds of the audio.

Happy transcribing!

Posted on

SPSS vs Excel for Entering Data with a Calculated Field

I’m in the midst of collecting data for an experiment so I thought I would try a little data entry experiment of my own.  Since I’m comfortable and confident with my Excel skills, I entered the first chunk of data using Excel.  I was starting from scratch, so I had to create all the field labels and develop a coding scheme for some of the responses. Some of the data is in the form of a test with responses that are correct or incorrect, while others are Likert-type survey questions with a range of ‘very likely’ through ‘very unlikely’ choices and I needed to remember what I did with each new record.  Since the number of choices for each field was pretty small I put the codes into comment boxes.  This got me thinking that SPSS might be an easier program to use because I can put the codes in as a field label.

I also needed to calculate a few fields.  This is super easy to do in Excel since I just needed a little equation of the form ‘=Cell 1 + Cell 2 + Cell 3 +…+ Cell n’.  As long as I fill the Excel equation down into the rows where I put new data the equation automatically calculates for every new record.  But since I had an inkling that SPSS might be easier for inputting the data, I tried to create calculated fields with the second chunk of data that needed to be entered.  Here’s what I knew and/or learned in my little experiment.  Writing formulas (creating calculated fields in SPSS terminology) is clunkier than writing them in Excel but no more difficult to understand.

You use the compute variable command in the transform menu.  Simply name your new field, then click away.  Click on a field name, then click the little arrow to put it into the formula, click the math operator, click another field name, and so on and so forth.  That worked fine except…

The big difference with Excel and SPSS is that in SPSS I needed to input all the data first, then create the calculated field if I want the program to calculate for every record while Excel automatically calculates as I add records.  If you create the field and then add more data, you’ll get a bunch of blanks.  On the other hand, if there is missing data in a field, SPSS won’t compute (a nice clue that something is missing), whereas Excel treats blanks as zeros.

My goal was to be able to mess around with the data as I was in the process of entering the data because I impatient to see the direction of my results and messing around with data is easier in SPSS than in Excel.  In the end, I think Excel easier for entering the data with a calculated field. Good thing it’s not too hard to open the Excel file in SPSS when my curiosity gets the best of me.

Posted on

Tracking Down The Broad Street Pump

Hello Data Superstars,

Recently I was touring in London, England when I was overcome by the urge to track down the Broad Street Pump made famous as the source of a cholera outbreak 1854.  At that time the Soho district of London didn’t have proper sewers or water purification plants.  Instead people lived with cesspools under the floor boards in their basements and discarded their household sewage in the streets.  The neighborhood was full of slaughter houses and  everything that accompanies that industry.  All that filth ended up either in the cesspools or drained into the Thames which contaminated the drinking water.

We know now that cholera is causes by bacteria in drinking water but back then people thought it was caused by bad air.  Dr. John Snow was skeptical of this theory so in what was to become one of the first examples of using visualization to understand one’s data, Snow mapped cholera cases from an outbreak that started August 31, 1854.  The map helped Snow pin point the Broad Street pump as being the epicenter of the outbreak with 127 deaths in three days in its immediate vicinity.(source https://en.wikipedia.org/wiki/1854_Broad_Street_cholera_outbreak)

Fast forward to 2018.  It turns out the location of the Broad Street Pump is an easy walk from the Leicester Square and Piccadilly Circus tube stations.  From Leicester Square we walked up hill through a warren of narrow streets that made us realize that Diagon Alley is not the stuff of fantasy but rather the every day existence of Londoners. (Unfortunately, I couldn’t find a magic wand store–will keep looking.)  The pump is no longer there but we did find this sign.

Yes, we actually geeked out and found the spot where data viz solved the cholera crisis.

Here is the red granite “kerbstone” complete with cigarette butt and urban trappings.