A Question on Chi-Square

[Saeed Rezaei]

Dear Members,
I am writing to you all to ask a question about the feasibility and applicability of the use of "Chi-square" for the kind of data and research question I have. I have a survey research in which one of the research questions is answered based on the responses of the respondents to one part of the questionnaire. My survey is about the attitude of Iranian language students towards plagiarism issues. I have developed a questionnaire with several items. I have mainly used descriptive statistics to answer the questions. But one question is:
1.    How do Iranian students of Translation Studies, TEFL, Linguistics, and English Literature define plagiarism? Is there any significant difference in their definitions?

In order to answer this research question I have used descriptive statistics for the first part. How about the second part? Do you think Chi-square is fine. I have reported the data based on chi-square but the referee of the journal I submitted my paper to has raised questions about the appropriateness of chi-square.
The participants were 122 language students (Translation, TEFL, Literature, Linguistics)

The items tapping this research question are:

Item 1: Plagiarism is using someone else’s words as if they were your own.
   
Item 2: Plagiarism is using someone else’s ideas as if they were your own.

Item 3: Plagiarism is using someone else’s results as if they were your own.
   
Item 4: Plagiarism is getting your ideas from a text book.

Item 5: Plagiarism is copying and pasting without acknowledging the original source.

Item 6: Plagiarism is getting ideas from a source and paraphrasing them but without acknowledging the original source.

I have reported a descriptive statics and a chi-square. The spss output for chi-square is like this:

Chi-Square Tests
    Value    Df    Asymp. Sig. (2-sided)
Pearson Chi-Square    31.400a    15    .008
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 6.03.

Do you think it is fine to use Chi-square to answer this question?

Your comments are highly appreciated in advance.


Kindest Regards,
Saeed Rezaei
Sharif University of Technology
Department of Languages and Linguistics
Tehran
Iran
  

[Rich Ulrich]

People who say "I used chi-square" usually are referring to the
contingency-table chi-squared test.  Please be more specific in
this description, because:  If you used a contingency table, you
mis-used a contingency table in this instance.

A legitimate contingency table has each person represented just
one time.  You have a sample of 122, and 24 cells (4x6), and thus
the result could not be "a minimum expected count of 6.03" when
the average is barely over 5. 

You could represent each person just once if you considered that
these six items represent a scale of "least-inclusive" to "most-inclusive"
and then use, for each person, the most-inclusive rating that they
provided.  This would presumably be the rating that has the lowest
frequency in the sample.  However, if you are assuming that these
responses are ordered in this fashion, then it will be proper to use
an analysis that makes use of the rank-ordering.  For instance, for
scores of 1 to 6, you could use a simple ANOVA on the scores. 
CROSSTABS gives you a test by Mantel for such data presented in
table form.  (If I recall properly, that happens to be tested by a
"chi-squared" test, though it is not a Pearson contingency-table
chi-squared.)

Alternatively, if the items are not so precisely located by "laxness"
(or whatever dimension that would be), you could count the scores
to get a total of 0 to 6 for each person.  Again, the analysis could be
an ANOVA of a scaled score.

--
Rich Ulrich

---

Date: Sat, 15 Jun 2013 06:17:10 -0700
From: [hidden email]
Subject: A Question on Chi-Square
To: [hidden email]

Dear Members,
I am writing to you all to ask a question about the feasibility and applicability of the use of "Chi-square" for the kind of data and research question I have. I have a survey research in which one of the research questions is answered based on the responses of the respondents to one part of the questionnaire. My survey is about the attitude of Iranian language students towards plagiarism issues. I have developed a questionnaire with several items. I have mainly used descriptive statistics to answer the questions. But one question is:
1.    How do Iranian students of Translation Studies, TEFL, Linguistics, and English Literature define plagiarism? Is there any significant difference in their definitions?

In order to answer this research question I have used descriptive statistics for the first part. How about the second part? Do you think Chi-square is fine. I have reported the data based on chi-square but the referee of the journal I submitted my paper to has raised questions about the appropriateness of chi-square.
The participants were 122 language students (Translation, TEFL, Literature, Linguistics)

The items tapping this research question are:

Item 1: Plagiarism is using someone else’s words as if they were your own.
   
Item 2: Plagiarism is using someone else’s ideas as if they were your own.

Item 3: Plagiarism is using someone else’s results as if they were your own.
   
Item 4: Plagiarism is getting your ideas from a text book.

Item 5: Plagiarism is copying and pasting without acknowledging the original source.

Item 6: Plagiarism is getting ideas from a source and paraphrasing them but without acknowledging the original source.

I have reported a descriptive statics and a chi-square. The spss output for chi-square is like this:

Chi-Square Tests
    Value    Df    Asymp. Sig. (2-sided)
Pearson Chi-Square    31.400a    15    .008
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 6.03.

Do you think it is fine to use Chi-square to answer this question?

Your comments are highly appreciated in advance.


Kindest Regards,
Saeed Rezaei
Sharif University of Technology
Department of Languages and Linguistics
Tehran
Iran
  

[Saeed Rezaei]

Dear Rich,
Thanks for your informative comments. Regarding your question, I should say that I had a Likert type scale with 1 to 6.
Thanks again for your comments.

Best,
Saeed

 

On Mon, Jun 17, 2013 at 9:46 PM, Rich Ulrich <[hidden email]> wrote:

People who say "I used chi-square" usually are referring to the
contingency-table chi-squared test.  Please be more specific in
this description, because:  If you used a contingency table, you
mis-used a contingency table in this instance.

A legitimate contingency table has each person represented just
one time.  You have a sample of 122, and 24 cells (4x6), and thus
the result could not be "a minimum expected count of 6.03" when
the average is barely over 5. 

You could represent each person just once if you considered that
these six items represent a scale of "least-inclusive" to "most-inclusive"
and then use, for each person, the most-inclusive rating that they
provided.  This would presumably be the rating that has the lowest
frequency in the sample.  However, if you are assuming that these
responses are ordered in this fashion, then it will be proper to use
an analysis that makes use of the rank-ordering.  For instance, for
scores of 1 to 6, you could use a simple ANOVA on the scores. 
CROSSTABS gives you a test by Mantel for such data presented in
table form.  (If I recall properly, that happens to be tested by a
"chi-squared" test, though it is not a Pearson contingency-table
chi-squared.)

Alternatively, if the items are not so precisely located by "laxness"
(or whatever dimension that would be), you could count the scores
to get a total of 0 to 6 for each person.  Again, the analysis could be
an ANOVA of a scaled score.

--
Rich Ulrich

---

Date: Sat, 15 Jun 2013 06:17:10 -0700
From: [hidden email]

Subject: A Question on Chi-Square

To: [hidden email]

Dear Members,
I am writing to you all to ask a question about the feasibility and applicability of the use of "Chi-square" for the kind of data and research question I have. I have a survey research in which one of the research questions is answered based on the responses of the respondents to one part of the questionnaire. My survey is about the attitude of Iranian language students towards plagiarism issues. I have developed a questionnaire with several items. I have mainly used descriptive statistics to answer the questions. But one question is:
1.    How do Iranian students of Translation Studies, TEFL, Linguistics, and English Literature define plagiarism? Is there any significant difference in their definitions?

In order to answer this research question I have used descriptive statistics for the first part. How about the second part? Do you think Chi-square is fine. I have reported the data based on chi-square but the referee of the journal I submitted my paper to has raised questions about the appropriateness of chi-square.
The participants were 122 language students (Translation, TEFL, Literature, Linguistics)

The items tapping this research question are:

Item 1: Plagiarism is using someone else’s words as if they were your own.
   
Item 2: Plagiarism is using someone else’s ideas as if they were your own.

Item 3: Plagiarism is using someone else’s results as if they were your own.
   
Item 4: Plagiarism is getting your ideas from a text book.

Item 5: Plagiarism is copying and pasting without acknowledging the original source.

Item 6: Plagiarism is getting ideas from a source and paraphrasing them but without acknowledging the original source.

I have reported a descriptive statics and a chi-square. The spss output for chi-square is like this:

Chi-Square Tests
    Value    Df    Asymp. Sig. (2-sided)
Pearson Chi-Square    31.400a    15    .008
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 6.03.

Do you think it is fine to use Chi-square to answer this question?

Your comments are highly appreciated in advance.


Kindest Regards,
Saeed Rezaei
Sharif University of Technology
Department of Languages and Linguistics
Tehran
Iran
  

--
Regards
Saeed Rezaei
Lecturer
Languages and Linguistics Center
Sharif University of Technology
Tehran
 IRAN
Skype: [hidden email]