Вы находитесь на странице: 1из 6

0802790 1

A critique of the journal article Gaston, A., Grace Cramp, A., and
Prapavessis, H. (2012). Pregnancyshould women put up their feet or
lace up their running shoes?: Self-presentation and the exercise
stereotype phenomenon during pregnancy. J ournal of Sport and
Exercise Psychology.


This critique sets out to explore the value of Gaston, Grace Cramp, and
Prapavessis (2012) study. It will touch on; the relevance of the title, the niche
in the literature that motivated this research, the reasoning behind the
development of both hypotheses, and the appropriateness of the
methodological and data analysis techniques chosen that prompted their
conclusions. It will end with logical thoughts on the article in questions
reliability and validity, and recommendations on how it could have been
improved.

Firstly it is of note that the purpose of a journal articles title is to permit
the reader a quick overview of the article so they may decide whether it is of
value to them (Belcher, 2009), the sensationalised title of Gaston et al. (2012);
Pregnancyshould women put up their feet or lace up their running shoes?:
Self-presentation and the exercise stereotype phenomenon during pregnancy
is unnecessary and does not adequately represent the intention of the article.
Perhaps, a slightly more appropriate title would be; Does the positive
exercise stereotype apply during pregnancy? this would help alleviate
uncertainty in prospective readers, and allow clearer links between the
authors objective and their findings.

The objective of this article was clearly stated within the abstract,
however only vaguely alluded to within the actual paper. The introduction
helped to support the reasons why the authors felt there was a need for this
research. Although to a fault, there was no literature review, however the
authors presented 25 novel research articles from the last two decades, 23 of
which belonged to a wide variety of peer-reviewed journals. One however
(Lutz, Linder, & Greenwood, 2005) was taken from a website that was
anecdotal at best, the other (Sodenstrom, 1999) was an unpublished doctoral
dissertation from Ume University written in Swedish, and with no native
0802790 2
tongued Swedes in the author list, it must be noted that there is a slight
potential for mistranslation, despite this, the introduction flowed somewhat
logically and provided a certain amount of reasoning for the undertaking of the
study.
Although, the justification for the choice of hypotheses was a little
lacking, as within the research presented nothing indicated that pregnant
women would be seen in a negative light if they were excessive exercisers
and as such, both alternative hypotheses seem to be quite an illogical jump.
To add to this, the inability to prove alternative hypotheses correct is a
detriment to their goal, perhaps a statement of a null hypothesis would have
been helpful as this is at least disprovable (Smith, 2012).

On the whole, the methods employed to collect data that could
substantiate these hypotheses were sufficiently described to allow for
replication of the study. However the settings of the data collection were
omitted, it is presumed that the participants answered the questions from
either their home or work computer, which could present a threat to internal
validity as the multitude of different possible distractions could confound the
data collected (Zaccai, 2003). No other efforts were mentioned to reduce the
internal validity of the study.
The authors support their choice of sample size with a reference to
Cohen (1992), however that study was regarding univariate tests, and so to
maintain a high statistical power during a multivariate analysis of variance
(MANOVA) group sizes of n= >70 are required (Rencher & Christensen,
2012).
Also of note, the quasi-random convenience sample of Canadian
women in the education profession means the results are not generalisable
beyond the 202 participants in the study (Zaccai, 2003). The authors do note
that expanding their inclusion criteria to a more diverse ethnic,
socioeconomic, and males would improve upon this.

The use of a one-way analysis of variance (ANOVA) to assess
differences in the independent groups demographics was correct, as even
though it is unclear whether the data is normally distributed or not due to poor
0802790 3
wording, an ANOVA is a robust enough test to overcome such violations
(Zaccai, 2003).
Attention should also be drawn to the lack of information pertaining to
which chi-square procedures were used to certify homogeneity, as using
different chi-squared tests will produce drastically different results. It is
however presumed the authors used Pearsons chi-squared test, as the
number of expected frequencies in the cells are above 22 (Devore, 2011)
however clarifying this would be of benefit to the results reliability.
It is also not clear what baseline exercise (p. 230) data refers to, it is
likely that this is self-reported exercise (Table 1.), although referring to it as a
baseline measure implies that they will collect repeated measures data, extra
clarity is also required here.

The choice to utilise an omnibus MANOVA test to measure both
personality and physical dimension ratings against the target groups was
quite correct, as this helped reduce type-I error, which would be an incorrect
rejection of a null hypothesis (Smith, 2012). However, again it is unclear
whether or not the normality check mentioned previously was applied to
multivariate data, or solely univariate.
Also, the fact that Boxs M test returned significant, could add
substance to the argument for a larger sample size to increase participants in
each cell (LaTour & Miniard, 1983). Alternatively, implementing Levenes test
of homogeneity of variance could be beneficial, as this tolerates violations of
normality to a greater extent (Glass, Peckham, & Sanders, 1972).
Added clarity on less common statistical tests such as Tamhanes T
2

test would also be helpful, as the current literary recommendation is for the
author to explain them, or at the very least ensure they are referenced
(Zaccai, 2003).

The lack of clarity in the methodology, and the complex nature of the
tables means the results section takes a few readings to fully comprehend
and therefore might mean confusion to the average reader. The inclusion of
all of the demographics characteristics measured seems to lean towards
unnecessary data torture (Altman, 1998), the authors should also be aware
0802790 4
that the purpose of tables are to assist the readers comprehension, and are
not just a place to store all the data collected (Lang, 2004).
This may be slightly pernickety perhaps, but tables 3 and 4 do not
clarify which descriptive statistic are represented, but as it is ordinal data and
standard deviation is presented, mean is the likely choice. They also do not
clarify what the anchor is referring to. However on a slightly positive note,
within these two tables is the only place in the entire article where the authors
formally state the chosen alpha level of <0.05.

Overall, the conclusions arrived upon by Gaston and colleagues were
quite obviously drawn from the results, however it could be argued that they
didnt concentrate on the research question as thoroughly as perhaps was
necessary. Their data showed that the positive exercise stereotype does
extend to pregnant women, and as such the main objective of the research
was met, however, the authors seemed to be far too concerned with their
hypotheses regarding the excessive exerciser group, that they seemed to
neglect the data that supported their main objective. Even during the
discussion, it is still unclear why they opted for the two alternative hypotheses,
had they stated a formal null hypothesis in the introduction, they would be
able to reject it now.
The authors do highlight limitations of their study and point future
researchers in a direction that might expand the knowledge base around their
findings, however they overlook certain considerations that would aid the
reliability and validity of their study.
A solid literature review may have helped strengthen the reliability of
some of the measures utilised and methodological choices made, whilst an
increased sample size would have added more statistical power to the study.
Whereas the addition of a graphical representation of tables 3 and 4 may
have alleviated some of the confusion within the results section, and
facilitated easier visualization of trends in the data.
The inclusion of estimated confidence intervals should also have been
added wherever possible to bolster the understanding of the magnitude and
size of the effects, as probability values on their own do not offer this
meaningful insight (Strasak, Zaman, Pfeiffer, Gobel, & Ulmer, 2007).
0802790 5
One last quandary of note, is there was no mention of any potential
conflicting interests, however the reader should not be lead to presume there
were none.
Overall, this article is still a noteworthy contribution to the field of
research, even though it has value, and the issues raised appear to have
educational and social importance beyond the scope of just these findings,
many of the methods need reevaluation or at the very least clarity, in an
attempt to tackle the prevalent threats to validity and reliability.




































References

0802790 6
Altman, D. G. (1998). Statistical reviewing for medical journals. Statistics In
Medicine, 17(23), 2661-2674.

Belcher, W. L. (2009). Writing your journal article in twelve weeks: A guide to
academic publishing success. London: Sage Publications.

Devore, J. L. (2011). Probability and statistics for engineering and the
sciences. Oxford: Duxbury Press.

Gaston, A., Grace Cramp, A., & Prapavessis, H. (2012). Pregnancyshould
women put up their feet or lace up their running shoes?: Self-presentation and
the exercise stereotype phenomenon during pregnancy. Journal of Sport and
Exercise Psychology, 34(2), 223-237.

Glass, G. V., Peckham, P. D., & Sanders, J. R. (1972). Consequences of
failure to meet assumptions underlying the fixed effects analyses of variance
and covariance. Review of Educational Research, 42(3), 237-288.

Lang, T. (2004). Twenty statistical errors even you can find in biomedical
research articles. Croatian Medical Journal, 45(4), 361-370.

LaTour, S. A., & Miniard, P. W. (1983). The misuse of repeated measures
analysis in marketing research. Journal of Marketing Research, 45-57.

Lutz, R., Linder, D., & Greenwood, M. (2005). Social influences and physical
activity. In the Association of Exercise Physiology Study Guide (1-28).
Retrieved from http://www. exercisephysiologists.com/ASEP-
EPCresourceMANUAL/index.html

Smith, C. J. (2012). Type I and Type II errors: What are they and why do they
matter?. Phlebology, 27(4), 199-200.

Sodenstrom, T. (1999). Gymkulturens logik: om samverkan mellan kropp,
gym och samhlle. [The logic of gym culture: On the interplay between the
body, gyms, and society]. Unpublished doctoral dissertation, Umea University,
Umea.

Strasak, A. M., Zaman, Q., Pfeiffer, K. P., Gobel, G., & Ulmer, H. (2007).
Statistical errors in medical research-a review of common pitfalls. Swiss
Medical Weekly, 137, 44-49.

Zaccai, J. H. (2004). How to assess epidemiological studies. Postgraduate
Medical Journal, 80(941), 140-147.

Вам также может понравиться