A little Background

#MTBoS and #iteachmath are two popular online communities of math educators on twitter. Both communities have grown over the years to become valuable “go to” places for many educators across the country and beyond both for personal PD and resources for math classes. Prior to 2017, only #MTBoS existed. #iteachmath came into being in 2017 after a major contributor to the #MTBoS community, Dan Meyer, in a controversial tweet captioned “Let’s Retire #MTBoS”, suggested that community members start using iteachmath in place of #MTBoS. Meyer’s argument was that the harshtag #iteachmath, among other things, is more inclusive and there were many educators that shared his view (See here for more on the story)

My overall goal for this analysis is to compare these two popular mathematics teacher Twitter communities (#MTBoS and #iteachmath) to:

Understand the kinds of resources and opportunities they offer for informal teacher professional learning and possible difference(s) between both of them
Describe the overall sentiment related to the #MTBoS and #iteachmath hashtags, compare them, and potential identify reasons for such sentiments.

Guiding Questions

My analysis will be guided by the flowing research questions:

What kinds of issues/topics do posts in both communities tend to focus on the most? What aspects of teacher professional life or components of teaching and learning, in general, do these topics relate to? What are the similarities between posts with #MTBoS tweets and #iteachmath in this regard? What differences exist?
What is the overall sentiment about #MTBoS tweets and #iteachmath tweets and how do they compare to each other?
Which words tend to occur together in posts with #MTBOS and those with #iteachmath?

Target Audience

My hope is that this analysis will contribute to mathematics teachers and mathematics teacher educators’ awareness of the potentials of teacher networks as a space for teacher informal learning.

Methods

Data Source and Pre-processing The rtweet package in R was used to retrieve 3200 tweets with the harshtags #MTBoS and # iteachmath respectively from twitter. These are tweets that were made 6-9 days preceding the day the data was retrieved. The data was filtered so that only tweets made in the English language were left, stop-words were removed, tokenization was carried out, and four variables– status_id,screen_name, created_at,text– were selected from the original data for tweets with the harshtag #MTBOS and #iteachmath respectively for the analysis.

Data Analysis: Word cloud, word counts and frequencies, quotes related to specific words, Top Trigrams were used to describe findings.

Analysis and Findings

The word cloud representing tweets with the harshtag #MTBOS is given below:

We could also look at the word cloud representing tweets with the harshtag #iteachmath but I suspect that the picture may likely be similar (not saying they will have exactly the same words). So instead, let’s look at word counts for #MTBOS and #iteachmath tweets respectively to get a clearer picture of word frequencies:

So far, we can see that there is a tendency for math educators to use the #MTBOS and #iteachmath harshtags in the same post.Could there be a distinction in the kinds of discussions that posts with these harshtags tend to focus on? I doubt there is, but we can’t jump into conclusion. Let the data speak for itself.

To get a sense of the kinds of resources and opportunities for teacher learning that each community offers (or if you like the kinds of issues that posts with the #MTBOS and #iteachmath harshtags tend to focus on), I will dive into topic modelling. Structural topic modelling holds a better promise for analyses of this nature but let’s use an LDA model for the purpose of this analysis.

Topic Modelling with Latent Dirichlet Allocation (LDA)

To get a general sense of the topics that math educators in these two communities focus on, I begin by fitting an LDA model with k=20. Next, I use the ldatuning package in R to find a suitable value for K. If it turns out that my initial value of K=20 is not suitable, the next step will be to fit a new model using a better value (how that is determined will be described shortly). Why do we care about K? Bail (2018) notes that “selecting the number of topics for your model is a non-trivial decision and can dramatically impact your results”.

## A LDA_VEM topic model with 20 topics.

##      Topic 1       Topic 2      Topic 3      Topic 4       Topic 5       
## [1,] "mtbos"       "mtbos"      "mtbos"      "mtbos"       "mtbos"       
## [2,] "exam"        "iteachmath" "learning"   "education"   "iteachmath"  
## [3,] "tip"         "mathschat"  "desmos"     "math"        "math"        
## [4,] "probability" "research"   "iteachmath" "mathematics" "elemmathchat"
## [5,] "chance"      "math"       "math"       "message"     "pm"          
##      Topic 6      Topic 7   Topic 8        Topic 9      Topic 10      
## [1,] "mtbos"      "free"    "mtbos"        "mtbos"      "mtbos"       
## [2,] "iteachmath" "math"    "iteachmath"   "iteachmath" "test"        
## [3,] "math"       "mtbos"   "elemmathchat" "math"       "exam"        
## [4,] "week"       "join"    "msmathchat"   "check"      "significance"
## [5,] "time"       "webinar" "challenge"    "puzzles"    "tip"         
##      Topic 11     Topic 12     Topic 13     Topic 14     Topic 15    
## [1,] "mtbos"      "mtbos"      "mtbos"      "mtbos"      "mtbos"     
## [2,] "iteachmath" "iteachmath" "iteachmath" "iteachmath" "math"      
## [3,] "math"       "math"       "students"   "mathisfun"  "iteachmath"
## [4,] "ideas"      "pd"         "math"       "ss"         "slope"     
## [5,] "teachers"   "summer"     "build"      "math"       "values"    
##      Topic 16     Topic 17     Topic 18      Topic 19     Topic 20    
## [1,] "mtbos"      "mtbos"      "mtbos"       "mtbos"      "mtbos"     
## [2,] "iteachmath" "iteachmath" "math"        "iteachmath" "iteachmath"
## [3,] "real"       "teacher"    "iteachmath"  "math"       "math"      
## [4,] "survey"     "onted"      "post"        "learn"      "science"   
## [5,] "life"       "fraction"   "dbcincbooks" "play"       "elearning"

You will notice from the results (see table above) that mtbos and iteachmath appear on almost all the topics, this is not helpful in making sense of the topics. I will now filter out mtbos and iteach math to make it easier to capture what each topic is about since both words may not be that helpful for my analysis.

## A LDA_VEM topic model with 20 topics.

##      Topic 1    Topic 2    Topic 3   Topic 4   Topic 5       Topic 6        
## [1,] "math"     "math"     "math"    "school"  "math"        "msmathchat"   
## [2,] "students" "teacher"  "maths"   "math"    "mathematics" "challenge"    
## [3,] "solving"  "mathchat" "post"    "love"    "science"     "hsmathchat"   
## [4,] "mathchat" "block"    "blog"    "middle"  "project"     "mathstratchat"
## [5,] "maths"    "register" "student" "puzzles" "elearning"   "math"         
##      Topic 7    Topic 8     Topic 9      Topic 10        Topic 11      
## [1,] "students" "learning"  "free"       "math"          "exam"        
## [2,] "fun"      "mathschat" "math"       "science"       "tip"         
## [3,] "math"     "play"      "edtech"     "elearning"     "significance"
## [4,] "mathchat" "math"      "mathschat"  "geometry"      "test"        
## [5,] "maths"    "page"      "activities" "visualization" "check"       
##      Topic 12     Topic 13   Topic 14 Topic 15     Topic 16   Topic 17  
## [1,] "math"       "math"     "math"   "time"       "math"     "summer"  
## [2,] "join"       "geogebra" "exam"   "day"        "maths"    "pd"      
## [3,] "share"      "function" "book"   "desmos"     "based"    "math"    
## [4,] "free"       "algebra"  "tip"    "mathequity" "students" "schools" 
## [5,] "oratbuyehs" "condo"    "slope"  "activity"   "teachers" "sessions"
##      Topic 18    Topic 19   Topic 20        
## [1,] "desmos"    "ideas"    "join"          
## [2,] "mathisfun" "graders"  "cheers"        
## [3,] "ss"        "life"     "mathwhiteboard"
## [4,] "math"      "real"     "hope"          
## [5,] "mathplay"  "teachers" "fluency"

The new result (see table above)has mtbos and iteachmath filtered out. We can now proceed with finding a suitable value of K using the Griffiths2004 metric as below:

As a rule of thump,to determine a suitable value for K, we look for an inflection point in our plot. Considering that the inflection point is the point of a curve at which a change in the direction of curvature occurs, I can maintain my value of K=20. The Griffiths2004 is one of 3 metrics that can be used for this purpose. That settled, what can one infer about topics that tweets with the #MTBOS harshtag tend to focus on? As you can tell, it’s difficult to authorittively say what each topic is about without being familiar with the #MTBOS twitter community. While this is not the best result possible considering that there are other topic modelling approaches like Dynamic Topic Models, Correlated Topic Models, Hierarchical Topic Models, and more recently, Structural Topic Modeling (STM); it seems some level of familiarity is still required to make authoritative claims. I will now obtain a similar result (table) for tweets with #iteachmath before attempting to make sense of some of the topics.

## A LDA_VEM topic model with 20 topics.

##      Topic 1      Topic 2      Topic 3     Topic 4       Topic 5     
## [1,] "learning"   "gt"         "mathisfun" "students"    "students"  
## [2,] "school"     "classroom"  "students"  "frustrating" "looneymath"
## [3,] "math"       "edchat"     "ss"        "perimeter"   "math"      
## [4,] "operations" "edutwitter" "desmos"    "mix"         "listen"    
## [5,] "teachers"   "giving"     "mathplay"  "solving"     "person"    
##      Topic 6      Topic 7       Topic 8    Topic 9    Topic 10   Topic 11   
## [1,] "math"       "mathart"     "math"     "teach"    "join"     "students" 
## [2,] "challenge"  "maths"       "maths"    "math"     "register" "week"     
## [3,] "free"       "mathematics" "day"      "ti"       "geogebra" "check"    
## [4,] "msmathchat" "art"         "adding"   "fractals" "math"     "functions"
## [5,] "hsmathchat" "article"     "students" "project"  "maths"    "free"     
##      Topic 12  Topic 13 Topic 14         Topic 15     Topic 16     
## [1,] "ideas"   "love"   "math"           "math"       "math"       
## [2,] "test"    "class"  "dbcincbooks"    "mathkaveli" "maths"      
## [3,] "classes" "real"   "reagantunstall" "teachers"   "mathigonorg"
## [4,] "join"    "survey" "block"          "mathchat"   "add"        
## [5,] "graders" "data"   "guidedmath"     "equity"     "cheers"     
##      Topic 17        Topic 18      Topic 19         Topic 20  
## [1,] "math"          "mathematics" "hope"           "math"    
## [2,] "science"       "mathteacher" "goformative"    "mathchat"
## [3,] "elearning"     "teachers"    "teachertwitter" "mathed"  
## [4,] "visualization" "dedication"  "copy"           "video"   
## [5,] "geometry"      "passion"     "book"           "share"

Looking at topics 1 through 20 from my LDA model for #MTBOS (see table), one can infer that for #MTBOS tweets:

Topic 1 is probably a discussion about students’ problem solving
Topic 2 is likely about time scheduling or registration for a teacher event
Topic 3 may be about some blog post
Topic 4 and 7 are probably about some interesting puzzles or fun activities for teaching middle school or beyond
Topic 5 and 10 are possibly about some science project or elearning issue related to geometry or some other content area
Topic 6 may be about some online math challenge given the several harshtags under it
Topic 11 is probably about some tips for exams and testing or doing some statistical significance test
Topic 12, 13, 14, and 18 probably have something to do with teaching specific contents like algebra/functions with technology and a possible connection to play, fun, or equity (specifically in topic 15). -Topic 17 seems to be about summer PD opportunities in schools
Topic 20 could be about some PD session teachers are invited to join so they could learn about a tool (maybe math whiteboard) for teaching or assessmnet.

Similarly, looking at topics 1 through 20 from my LDA model for #MTBOS (see table), one can infer that for #MTBOS tweets:

Topic 1 is probably about an issue related to how a school or schools operate or learning math operations
Topic 2 is likely about a harshtag like #edchat/#edutwitter that is being promoted as a harshtag teachers should follow
Topic 3 might be talking about desmos and how it is fun
Topic 4 might be about students’ frustration with some perimeter problem or around learning perimeter
Topic 5 could be an invitation to join a math community e.g looneymath
As was the case with #MTBOS, Topic 6 may be about some online math challenge given the several harshtags under it or a call to follow the msmathchat, hsmathchat harshtags.
Topic 12 seems to be about testing/grading or joining a session or PD opportunity that focuses on assesment.
Topic 13 looks like a discussion around taking a survey or providing some data
Topic 10 might be an a discussion about joining or registering for a PD opportunity for learning about geogebra. -Topic 14 and 19 seem to be about some textbooks or resources for math learning
Equity seems to come in Topic 15 as well
Topic 17 seems to be about elaerning resource related to math or science -Topic 18 seems to be a discussion around good math teacher qualities like dedication and passion -Topic 20 may be about a teacher video that educators are encouraged to share.
Overall, we can see that both twitter opportunities offer different resources for teacher learning. These range from resources for in-person and remote learning, joining different learning communities, opportunities for personal professional development, information about events that teachers could benefit from, discussions about practice as well as opportunity to learn from others’ practice.

Let’s now look at some quotes with the words desmos, webinar, and student for #MTBOS and #iteachmath respectively.

** #MTBOS **

Desmos:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "Let Mother's Day add to your students' math understanding.  We have 13 acti~
##  2 "Is there a way to make a bar graph (not a histogram) with categorical value~
##  3 "We looked at Mississippi voter registration data from the 1960s this week i~
##  4 "Hey #mtbos - is there a way to prevent students from pasting work from else~
##  5 "<U+0001F44B>While @Give_Me_A_Sine and I presented on @Desmos some people as~
##  6 "How good are you at estimating fractions?  @desmos #iteachmath #mtbos\r\nht~
##  7 "Is there a way to make a bar graph (not a histogram) with categorical value~
##  8 "Hey @FergFlorSchools! Remember the @desmos activity builder training I gave~
##  9 "Let Mother's Day add to your students' math understanding.  We have 13 acti~
## 10 "assigned my students a Desmos activity and when they logged in it says rest~

webinar:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  2 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  3 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  4 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  5 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  6 "<U+0001F39E><U+FE0F> Math Whiteboard Webinar https://t.co/2bfYKsOkPr Watch ~
##  7 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  8 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~
##  9 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~
## 10 "Join me for a free webinar on Mathigon on May 13th at 3:30pm EST. Free math~

student:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "About how many buffalo do you see?\r\nTry out the @MultiplicityNU Image of ~
##  2 "Two more #fractals by my students working on the project \"Regards de géomè~
##  3 "I'm lucky I get to substitute for 2 @AoPSNews classes today - I just finish~
##  4 "It’s frustrating teaching the Order of Operations (BODMAS or PEMDAS).\r\n\r~
##  5 "@pwharris joins #MathKnights for a high energy interview. She speaks to us ~
##  6 "I've a #mathisfun problem idea to ask my younger students @AoPSNews re: eff~
##  7 "Why is \"I do, we do, you do\" ultimately a problem for your students? Tom ~
##  8 "If you know that there has to be a better way to reach all of your students~
##  9 "If you know that there has to be a better way to reach all of your students~
## 10 "Excited to get a sneak peek at @StevenXClontz's new #DoTheMathBook that rel~

** #iteachmath **

Desmos:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "Reviewing solving quadratic equation in #Algebra .  I made this @desmos act~
##  2 "Let Mother's Day add to your students' math understanding.  We have 13 acti~
##  3 "We looked at Mississippi voter registration data from the 1960s this week i~
##  4 "Is there a way to make a bar graph (not a histogram) with categorical value~
##  5 "Reviewing solving quadratic equation in #Algebra .  I made this @desmos act~
##  6 "Reviewing solving quadratic equation in #Algebra .  I made this @desmos act~
##  7 "How good are you at estimating fractions?  @desmos #iteachmath #mtbos\r\nht~
##  8 "The power of choice.  These @desmos options have been EXTREMELY helpful whe~
##  9 "Is there a way to make a bar graph (not a histogram) with categorical value~
## 10 "<U+0001F44B>While @Give_Me_A_Sine and I presented on @Desmos some people as~

Webinar:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "The next @MathigonOrg webinar is on Wed, 5/12 at 7 PM EST and will focus on~
##  2 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~
##  3 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~
##  4 "The next @MathigonOrg webinar is on Wed, 5/12 at 7 PM EST and will focus on~
##  5 "The next @MathigonOrg webinar is on Wed, 5/12 at 7 PM EST and will focus on~
##  6 "The next @MathigonOrg webinar is on Wed, 5/12 at 7 PM EST and will focus on~
##  7 "The next @MathigonOrg webinar is on Wed, 5/12 at 7 PM EST and will focus on~
##  8 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~
##  9 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~
## 10 "Who is joining us tonight at 7 PM ET / 4 PM PT for our webinar:\r\n\r\n5 Re~

student:

## # A tibble: 10 x 1
##    text                                                                         
##    <chr>                                                                        
##  1 "It's so frustrating when students mix up area and perimeter. Here's my take~
##  2 "Working on a solving equations menu for grade 9 math students.  Always nice~
##  3 "\"My philosophy has been for a pretty long time, it's better to learn the b~
##  4 "Looking to help your students build &amp; interpret functions? Check out th~
##  5 "#mtbos #iteachmath Some assistance please: anyone have experience with Savv~
##  6 "It's so frustrating when students mix up area and perimeter. Here's my take~
##  7 "Teaching both remote &amp; in-person students at the same time how to use a~
##  8 "Lesson is making box plots. My room was plastered with 40th bday decoration~
##  9 "Can’t wait to use this and so excited to share this with others.  Let me kn~
## 10 "@claysmythe @SteveMillerOC @DavidSacks Well I’d be happy to send you any of~

Sentiment Analysis

Let’s now explore what math educator sentiments look like for each of these these twitter communities (#MTBOS and #iteachmath).How do they compare or differ from each other?. The following plot gives a picture of positive and negative sentiments about the #MTBOS and #iteachmath twitter communities based on the AFINN Lexicon:

Judging from the graphs, the sentiment about both online communities is generally positive. There appears to be more positive sentiment around the iteachmath harshtag and ofcourse, slighly less negative sentiment for it. However, this difference doesn’t look significant. If the difference is anything to go buy, it might possibly be based on the feeling that #iteachmath feels more inclusive. That said, it is difficult to establish a basis for a possible disparity in sentiment even if it were significant considering that educators seem to use both harshtags interchangeably.

Sentiments based on the AFFINN lexicon are limited to positive or negative sentiments. As a result, a more encompassing approach to sentiment is warranted. I will now use the NRC lexicon to capture other types of sentiments about these communities beyond positive/negative sentiments.

Like sentiments for the AFFINN lexicon, sentiments around #iteachmath are better off compared to those of #MTBOS for the NRC lexicon. Again, the difference doesn’t look significant.

Trigrams

Next, let’s look at trigrams for our #MTBOS and #iteachmath harshtags. The following plots show the top three words that tend to appear together.

Making sense of the Trigrams is not as straight forward as one will expect; however, at minimum, we can see that msmathchat, hsmathchat, and mathstratchat tend to appear together for #MTBOS. My hunch is that these are also harshtags for different online teacher communities. In the second graph, we see painting, Jessica, and Wynne together. If like me, you are wondering who Jessica Wynne is, or what math has got to do with painting, look no further. See this link for more on why this photographer has become the darling of mathmaticians.The words that go together with free also reveal that math educators in the #MTBOS online community get the chance to learn about free webinars, math sites, and related resources. The harshtags i mentioned also show up again for #iteachmath.

Discusion and Conclusions

Overall, i consider the nature of approaches and analyses employed in this case study as basic. However, there are a couple of insights that we can take-away as regards the initial guiding questions:

Online spaces/communities are fast becoming a valuable alternative for teacher learning or professional development as seen from the multiple harshtags that appear in #MTBOS and #iteachmath tweets.
Both the #MTBOS and #iteachmath offer several opportunities for teacher informal learning and growth including resources and opportunities related to teaching with technology, elearning,webinars, textbook recommendations, websites, invitations to join diferent learning communities, examples of practice, etc.
There is no remarkable difference in the nature of discussions that posts with either the #MTBOS or the #iteachmath harshtags tend to focus on.
Sentiments for both #MTBOS and #iteachmath are very postive. Although #iteachmath tweets tend to have more postive sentiments compared to #MTBOS, the difference between them is not that significant.
Math educators appear to use the #MTBOS and #iteachmath harshtags interchangeably and sometimes in the same post. Although the suggestion to change from using #MTBOS to #iteachmath by Dan Meyer was met with mixed reactions, it doesn’t seem to reflect on which of the two harshtags math educators use.

Conclusions

Limitations

The approaches employed in this analysis are somewhat basic, more sophisticated approaches will be needed to make better sense of the data.
The size of the data and the time period over which it was retrieved could possibly be factors that will invalidate the results of this analysis if a larger study that uses more data and covers a wider time period were to be conducted.

Implications

A major implication of this case study is that it draws attention to the need for greater awareness of the opportunities that online communities offer for personal teacher learning and growth. School leaders and other stakeholders need to point teachers to these communities and the resources they house.

Future Research - Future research can interview selected teachers who are members of these online teacher communities to see the kinds of resources/support they afford teachers and how they use such resources in their classes. - Another area to look at will be how teacher learning in informal spaces like twitter differs from conventional PD and what teachers’ perception and attitudes are towards both.

References

Silge, J., & Robinson, D. (2017). Text mining with R: A tidy approach. " O’Reilly Media, Inc.".

Teacher learning in informal spaces: A case study of two Math Twitter Communities

Jerome Amedu