Background for Discussion about Data Collection Integrity

Conclusions of statistical inference are meaningless if the data are not collected properly!

Before beginning a unit on frog anatomy, a seventh-grade biology teacher gives each of the 24 students in the class a pretest to assess their knowledge of frog anatomy. The teacher wants to compare the effectiveness of an instructional program in which students physically dissect frogs with the effectiveness of a different program in which students use computer software that simulates the dissection of a frog. After completing one of the two programs, students will be given a post-test to assess their knowledge of frog anatomy. The teacher will then analyze the changes in the test scores (their score on the post-test minus their score on the pretest). Suppose the teacher decides to allow the students to select which instructional program on frog anatomy (physical dissection or computer simulation) they prefer to take. Furthermore, suppose that 11 students choose actual dissection and 13 students choose computer simulation. Here are the results in a test of significance to compare the two groups:

Computer Simulation Group
Student / Pretest / Posttest / Difference
(posttest-pretest)
A / 58 / 88 / 30
B / 48 / 78 / 30
C / 59 / 89 / 30
D / 55 / 85 / 30
E / 54 / 84 / 30
F / 60 / 90 / 30
G / 48 / 78 / 30
H / 53 / 83 / 30
I / 47 / 77 / 30
J / 56 / 86 / 30
K / 55 / 85 / 30
L / 77 / 88 / 11
M / 75 / 87 / 12
Mean / 27.2
St Dev / 6.95
Actual Frog Dissection Group
Student / Pretest / Posttest / Difference
(posttest-pretest)
N / 88 / 98 / 10
O / 78 / 86 / 8
P / 89 / 99 / 10
Q / 85 / 93 / 8
R / 84 / 94 / 10
S / 90 / 98 / 8
T / 78 / 88 / 10
U / 83 / 91 / 8
V / 77 / 87 / 10
W / 86 / 94 / 8
X / 85 / 95 / 10
Mean / 9.1
St Dev / 1.04

The teacher uses a two sample significance test of means to make a comparison between the two groups. She tests the hypothesis that students who use the computer simulation will see a greater improvement in their scores.

  1. Null & Alternate Hypotheses
  2. H0: μ1 = μ2
  3. HA: μ1 > μ2
  4. Test statistic: t = 9.27
  5. p-value = 0.000
  6. Reject H0. Evidence shows that students who use the computer simulation will see a greater improvement in their scores. In fact, since the p-value is 0, there is strong evidence that this is the case.