Wednesday, August 26, 2015

Facebook Primary 2016, August Update, Does it Predict Support?

Last June I posted on the level of engagement of the presidential candidates on their Facebook pages to see who had the most enthusiastic support on Facebook.  This time I updated the Facebook page data to what has changed since June and to see how the level of Facebook activity correlates with outside polling data.  

I compiled data from the Facebook pages of 22 presidential candidates,(6 Dems and 16 Reps) who are listed on the Real Clear Politics polling average for the Republicans or the Democrats. Lincoln Chafee was listed at 0% and a search on Facebook for his campign page turned up no official campaign page so I entered a zero for each of his Facebook variables.    

I computed correlation coefficients for the variables: The # of page followers (those who click like on the page) in August, gain in followers from June to August, the number engaged with the page (those who clicked on, clicked like, shared or commented on posts from the page), the % engaged (the # engaged divided by the total followers), and the change in the % engagement from June to August with the Real Clear Politics poll average for the candidate.  The full data can be seen at the table in the bottom of the post.  The table of the correlation coefficients is presented below with statistically significant correlations in bold.  There were more significant correlations for Republicans because the sample size was larger.



Real
Clear
Politics
Avg D
Real
Clear
Politics
Avg R
# Followers Aug
0.87
0.69
# Gain in Followers Jun-Aug
0.66
0.84
# Engaged Aug
0.87
0.69
% Engagement Aug
0.47
0.30
% Engagement change, Jun-Aug
0.19
0.26



 The correlation plot for the Democrats is shown above for the relationship between the number of followers (by more than 20,000) and the poll average which was the strongest for them.  The regression line shows the best fit straight line for the data.  The correlation coefficient was 0.87 (75.8% of the variability) which indicates a strong positive linear relationship.  If the correlation was +1.0 (100% of the variability) the candidates would all fall on the regression line. The regression line predicts that for every new Facebook follower there is an increase of 0.000028%.  In other words, for every 100,000 new followers should be associated with a 2.8% increase in the candidates poll average.

Bernie Sanders is slightly ahead of Hillary Clinton are about even in the number of followers and the number engaged with Clinton leading by 19% in the poll average (she had a 50 point lead in June).  Joe Biden (who hasn't officially announced as a candidate) is a distant third in both the poll average and Facebook followers with Chafee, O'Malley and Webb in the lower right hand corner of the chart.  

 
The above chart for the Republicans summarizes the strongest relationship between the poll average and the gain in followers between June and August (0.84 correlation coefficient accounting for 70.6% of the variability).  Donald Trump had the highest gain at 1.49 million in followers followed by Ben Carson at 930,000.  They placed 1st and 3rd in the poll average respectively.  Jeb Bush was an outlier placing second in the poll average but only gaining 72,000 followers.  The other candidates were pretty close to the regression line.  The regression equation predicts that for every new follower there should be an increase of 0.0000116% in their poll average.  In other words, for every gain in followers of 100,000 there should be an increase of 1.16% in the poll average on the Republican side.  

As with all correlation and regression one must be careful about concluding a cause and effect relationship between two new variables.  The candidates popularity may lead to an increase in their Facebook following or vice versa.  The poll averages of Bush and Clinton suggest that they are buoyed by their name recognition and other factors.  However the Facebook presence (and other social media) of a lesser known candidate can circumvent the traditional media in getting their message out.


Candidate
Party
Followers
June
Gain in
Followers,
June
# engaged
June
Engage
% June
Followers
Aug
Gain 
Aug
# Engaged
Aug
%
Engage
Aug
Engage
change
Aug-Jun
Real 
Clear
Politics
Avg
Hillary Clinton
D
871449
32263
164721
0.19
1209179
337730
327399
0.27
0.08
49.3
Bernie Sanders
D
530262
52423
136772
0.26
1225552
695290
355980
0.29
0.03
25
Martin O'Malley
D
70478
4764
15535
0.22
76953
6475
3675
0.05
-0.17
1.7
Lincoln Chafee
D
0
0
0
0
0
0
0
0
0
0
Joe Biden
D
831968
8122
8998
0.01
841644
9676
1596
0
-0.01
12
Jim Webb
D
21745
256
2366
0.11
30544
8799
9956
0.33
0.22
1.3
Donald Trump
R
1792016
113658
53880
0.03
3282608
1490592
743935
0.23
0.2
22
Ted Cruz
R
1249160
4762
134836
0.11
1385876
136716
242349
0.17
0.06
7.3
Mike Huckabee
R
1743599
9519
231343
0.13
1827423
83824
415975
0.23
0.1
4.3
Rick Santorum
R
261168
1982
17289
0.07
264796
3628
11625
0.04
-0.03
1
Ben Carson
R
1487589
74383
168658
0.11
2418003
930414
869374
0.36
0.25
9.7
Rand Paul
R
2001304
78429
281904
0.14
2060430
59126
122158
0.06
-0.08
4.3
Lindsey Graham
R
111795
3914
10982
0.1
130110
18315
5621
0.04
-0.06
0.3
Jeb Bush
R
178172
1888
4681
0.03
250731
72559
30841
0.12
0.09
10.7
Scott Walker
R
283102
4901
24483
0.09
359444
76342
40554
0.11
0.02
7.7
Marco Rubio
R
874898
13893
83916
0.1
978401
103503
54927
0.06
-0.04
7.3
Chris Christie
R
104785
619
3014
0.03
118868
14083
8055
0.07
0.04
3.3
Rick Perry
R
1183094
8979
40992
0.03
1208938
25844
17278
0.01
-0.02
1.3
Bobby Jindal
R
243034
3504
5174
0.02
277777
34743
23878
0.09
0.07
0.7
Carly Fiorina
R
60165
4765
12713
0.21
281637
221472
86799
0.31
0.1
6.3
John Kasich
R
107703
479
7320
0.07
138213
30510
17159
0.12
0.05
0.3
George Pataki
R
15391

4512
29
18330
2939
6892
0.38
-28.62
0.1

**Related Posts**

2016 Facebook Presidential Primary Update: Engagement and Gain 

 

The 2016 Facebook Presidential Primary

 

It's All About The Likes

 

The Impact of the CSI w/o DB Facebook Page on Visits and Engagement