Lan Liu
Stacked Barchart or Side-by-Side Barchart?
Side-by-Side Barchart
1500
1125
Murders
750
Rapes
375 Robberies
0 Aggravated Assaults
Ames San Francisco Burglaries
Stacked Barchart
3000
2250
1500
Murders
750 Rapes
Robberies
0 Aggravated Assaults
Ames San Francisco Burglaries
Hypothsis: It is easier to read overall counts from a stacked barchart while it is
easier to read group counts from a side-by-side barchart.
When I worked on last assignment, I hesitated over whether to choose a stacked barchart or side by side
barchart. So, in this assignment, I want to do a study to compare these two kinds of barchart. I will use
part of the data that I used in last assignment--the crime statistics of five cities. I plan to ask my friends
from different majors at Iowa State University to do a survey. I will randomly divide them into two
groups. The first group will be asked to read a side-by-side barchart and the second group will read a
stacked barchart. The questions are the same for the two groups: to read the counts for different crimes
across different cities. I plan to show them the charts directly on my screen and will record their answers
in a table as follows. For the first group, they should just read the total counts directly from the chart
rather than by adding sub-categories up.
The make-up of the two groups
Treatment1
Lan Liu
Number Gender Major Graduate/
Undergraduate
1 F Economics Graduate
2 F Computer Graduate
Engineering
3 M Electronic Graduate
Engineering
4 M Chemistry Graduate
5 M Economics Graduate
6 F Economics Undergraduate
7 F Finance Undergraduate
8 M Computer Graduate
Engineering
Treatment2
Number Gender Major Graduate/
Undergraduate
9 F Economics Graduate
10 M Physics Graduate
11 M physics Graduate
12 F Electronic Graduate
Engineering
13 F Economics Graduate
14 F Economics Undergraduate
15 M Biology Graduate
16 F Economics Graduate
Truth Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 46 32 202 691 971
New York 7 13 288 330 271 909
San 12 21 517 326 867
Francisco 1742
Los 12 27 370 377 525
Angeles 1312
Houston 18 41 548 562 1296 2465
1 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 30 20 200 700 950
Lan Liu
New York 10 15 300 340 290 900
San
10 20 500 300 850 1900
Francisco
Los
10 20 380 390 520 1300
Angeles
Houston 5 40 580 590 1350 2600
2 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 1 40 30 200 700 1100
New York 5 10 300 350 270 1000
San
5 10 500 300 900 2000
Francisco
Los
10 20 380 390 500 1200
Angeles
Houston 10 30 580 590 1300 3000
3 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 35 20 200 710 1000
New York 8 15 300 375 295 920
San
5 15 575 310 860 1810
Francisco
Los
15 20 380 390 525 1300
Angeles
Houston 15 40 575 595 1350 2610
4 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 30 20 205 710 900
New York 5 15 290 365 280 920
San
5 15 590 320 850 1800
Francisco
Los
10 20 360 380 520 1300
Angeles
Houston 15 40 580 595 1350 2700
5 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 40 25 200 700 950
New York 1 5 280 310 270 905
San
Francisco 1 3 500 310 870 1650
Los
Angeles 10 15 380 390 570 1350
Houston 10 20 550 580 1290 2290
Lan Liu
6 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 25 15 200 700 980
New York 5 8 290 300 270 950
San
Francisco 3 5 500 310 900 1700
Los
Angeles 15 20 380 395 575 1300
Houston 10 20 560 580 1280 2300
7 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 30 15 200 690 1000
New York 5 10 300 320 280 1000
San
Francisco 4 8 510 310 890 2000
Los
Angeles 10 15 380 385 520 1250
Houston 8 20 580 590 1300 2600
8 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 40 18 200 685 1000
New York 5 8 300 325 280 900
San
Francisco 5 7 515 315 875 1600
Los
Angeles 8 12 380 390 520 1350
Houston 8 15 580 590 1300 2600
9 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 15 10 210 680 980
New York 8 5 300 350 250 900
San
Francisco 10 5 500 380 820 1750
Los
Angeles 8 20 380 370 500 1300
Houston 10 20 580 580 1300 2480
10 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 30 20 200 680 985
Lan Liu
New York 10 3 300 320 280 920
San
Francisco 15 8 500 370 810 1750
Los
Angeles 8 16 400 400 520 1300
Houston 15 20 480 450 1300 2475
11 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 10 9 200 680 980
New York 5 3 300 370 230 910
San
Francisco 10 10 530 370 830 1750
Los
Angeles 8 20 380 400 520 1320
Houston 20 40 570 600 1300 2430
12 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 15 10 210 680 980
New York 8 5 300 350 250 900
San
Francisco 10 5 500 380 820 1750
Los
Angeles 8 20 380 370 500 1300
Houston 10 20 580 580 1300 2450
13 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 35 20 200 690 990
New York 15 10 300 300 280 900
San
Francisco 18 10 500 370 850 1720
Los
Angeles 5 15 400 400 520 1300
Houston 15 25 580 580 1300 2480
14 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 20 8 210 690 985
New York 10 5 300 360 250 910
San
Francisco 10 5 500 385 825 1750
Los
Angeles 10 20 385 375 500 1350
Houston 10 20 580 590 1300 2475
Lan Liu
15 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 40 25 220 680 985
New York 8 5 300 300 290 910
San
Francisco 15 8 510 380 850 1780
Los
Angeles 5 15 400 390 530 1310
Houston 15 25 570 560 1300 2490
16 Aggravated Total
Murders Rapes Robberies Assaults Burglaries
Ames 0 28 15 220 680 980
New York 10 8 300 350 250 900
San
Francisco 10 5 520 380 850 1740
Los
Angeles 10 20 380 380 500 1310
Houston 10 20 580 580 1300 2485
Results of the study
I calculate the differences between the truth and the recorded data for the group counts and total counts,
respectively.
Difference-Group counts Side-by-side Stacked
Mean -1.8 1.7
SD 17.4 21.7
Difference-Total counts Side-by-side Stacked
Mean -42.4 -4.3
SD -130.4 14.8
Lan Liu
For the group counts:
n1=200, n2=200, t-statistic= -1.80, df=398, p-value=0.0712. The difference is significant at 10%.
For the total counts:
n1=40, n2=40, t-statistic= -1.84, df=78, p-value=0.0699. The difference is significant at 10%.
According to the test statistics, I cannot reject the hypothesis: It is easier to read overall counts from a
stacked barchart while it is easier to read group counts from a side-by-side barchart. The difference
between the two barcharts is significant at 10% level.
Overall, the study worked well. My sample is almost all graduate students, but I don’t this affect the
randomness of the sample. I picked five cities’ data and people complained that there is too much data to
read and it takes a long time to finish the survey. I realize that the data of the five cities are actually of the
same kind, and thus maybe a little bit redundant for this study. Next time, I would like to just pick one or
two cites for the chart and at the same time increase the sample size to make the study more efficient.