Examples of Published Research With Analysis Including a Two-Sample Comparison of Means Using the t Statistic

Candice Lewis provided an example of a two sample I test: Ford D. Y., and Harris, J.J. (1997). A study of the racial identity and achievement of black males and females, Roeper Review, 20, 105-110.

There were 149 participants (54 male participants and 95 female participants). The participants were black students in grades 6 through grade 9. One of the questions this study asked was, “How do black males and females differ in their racial identities?” The variables studied were gender and racial identity. Ford and Harris (1997) used the revised Racial Identity Scale for Black Students (RIS) to assess students’ racial identity. The RIS was administered to all of the students. The revised scale contained 24 Likert-type questions (strongly agree=4 to strongly disagree=1). The RIS is divided into four subscales: pre-encounter, encounter, immersion-emersion, and internalization.

The pre-encounter subscale contained 5 items such as: I am ashamed to be an African American, sometimes I feel like other students do not like me because I am African American. The encounter subscale contained eight items such as: I try to learn more about African Americans by talking to other people about my heritage, I am determined to find my Black identity. The immersion-emersion subscale contained eight items such as: Black is beautiful, My future is tied to the future of other Blacks. The internalization subscale contained three items such as everybody should learn about the cultures of other groups.

A mean of 2.0 was found for the pre-encounter subscale which means the students did not agree with these items. According to the t test, both male participants and female participants had a mean of 2.0 on this sub-scale, therefore the results were not significantly different. A mean of 2.8 was found on the encounter sub-scale. Male participants and female participants had the same mean (2.8), so the results were not statistically significant and students tended to agree with items in the encounter sub-scale. Results of the immersion-emersion t test revealed significant gender difference (p .05), with male participants having a higher response of 3.3. Female participants had a mean response of 3.1. The highest racial identity mean was 3.6 for the internalization sub-scale. According to the t test results, females had a significantly higher mean response of 3.7, compared to males, whose mean was 3.5, (p <. 05).

As far as I can tell this is an appropriate example of an independent samples t test. I am assuming they used the appropriate t test for unequal sample sizes.

Thank you, Candice. Here is my critique:

1. The authors did not report the df.

2. The authors used dehumanizing language ("males" and "females").

3. The authors did not report exact p-values.

4. The authors did not report the standard deviations. Without the standard deviations we cannot determine whether the significant effect reported here is trivial in magnitude or not. I suspect it is trivial. We also cannot determine whether the appropriate t test was employed (was there homogeneity of variance?).

5. There certainly appears to be a lot of crap out there in the journals you all read.

6. I would have liked to see a quote of the author's summary statements.

Here is another summary provided by a previous student:

Chng, C., Carlon, A., & Toynes, B. (2006). HIV on historically black colleges and universities (HBCU): A study of five campuses in Texas, Oklahoma, Louisiana.College Student Journal,40(1), 25-34. Retrieved from PsycINFO database.

Subjects: The subjects were students at a private Historically Black College or University

Variables Studied: (1) Gender of Student (2) HIV knowledge.

Design: Empirical Observational study. The researcher surveyed 1,146 students on their knowledge of HIV.

Conclusions: Female students know more about HIV than Male students

Quote: “Female students (Mean = 6.47.S.D. = 1.06) scored higher on overall HIV knowledge than did their male counterparts (Mean = 6.11, S.D. = 1.16).”… ”As seen in Table 2, we found statistical significance for gender on the overall HIV knowledge scores where women scored higher than men (Men = 6.19, Women = 6.47, t(519) = -2.887, p̲ = .004).”

The standard deviations appear to be approximately 1.5. With a difference between means of .36 that gives an estimated d of about .24, a small difference.