During the interview, students reviewed both the general
(G) and examples-based (EB) argument at the same time;
however, each response was printed on a separate page.
Students answered the following questions in the order
shown:
1. Which response convinces you that the statement
is true?
2. Why is the response you chose more convincing than
the other response?
3. If you were to give advice to the student who wrote
the response you did not choose as to how they could
improve their response to make it more convincing, what
would you tell them?
Response A (see Figure 1) is not a valid mathematical
proof. Response A is an EB argument that makes a case for
the statement being true in some cases (namely the case of
5 and 6, 12 and 13, and 1240 and 1241). Response B,
however, attempts to make an argument that applies to all
cases of pairs of consecutive numbers. The response draws
upon the definition of consecutive numbers (“If you take
any two consecutive numbers, you will always get one
even and one odd number”) as well as a fact previously
learned (“And we know that when you add any even
number with any odd number the answer is always odd”).
One reason why it may not be a mathematically valid proof
is that it is unclear whether the student had previously
proven the fact that even + odd = odd. Nevertheless, the
response features deductive argumentation based on defi-
nitions and assumptions to reason why the statement is
true for any two consecutive numbers.
Data Analysis
A member of the research team transcribed students’
responses to the interview task in Figure 1, and we ana-
lyzed those responses in the qualitative research program
HyperResearch, version 2.8.3 (2009, www.researchware
.com/products/hyperresearch.html). Because of audiore-
cording issues, only 22 out of the 25 participants’
responses could be transcribed, which reduced the sample
of interviews for our analysis. Question 1 asked students to
170
either choose Response A or B as most convincing.
Because Response A was an EB argument and Response B
was a more G argument, the first step of analysis was to
translate these choices into codes EB for choice of
Response A, or G for choice of Response B.