Again, because the data for the RLSS is ordinal, an ANOVA is not the best analysis tool.
The three coaching levels scored differently on three of the six behaviors: democratic behaviors, training and instruction, and social support.
In general, females scored much higher than did the male coaches.
A MANOVA was also used to examine the data for differences between the three levels of coaching (junior high, high school, and college) with regard to leadership behavior in general.
However, due to the nonrandom nature of the sample, the results would not generalizable beyond the 162 participants in the study. In order to reduce threats to internal validity, the participants were asked to respond honestly and confidentiality was stressed so that the coaches might feel more at ease in responding. The researchers mention that the scales were given in a variety of settings.
This could present a threat to the internal validity in that participants might not have been entirely focused on completing the scale, but instead on coordinating practice, completing paperwork, etc.Next, the student described specifics about the research design, including the sample, instrumentation, and data analysis.Ecological and population generalizability were discussed. The student spoke at length on threats to internal validity.Here is a really good example of a scholary research critique written by a student in EDRS 6301.The student who submitted this paper last semester earned a 100 on his critique. A succint summary is provided in the first paragraph.In previous classes we spent more time talking about statistics than the literature review. The sample was nonrandom, including 162 coaches that were chosen on a volunteer basis.That's why you'll see some fairly complex explanations in this paper on the data analysis but no information on the literature review. Within the sample, 118 (0.73) of the coaches were male, while 44 (0.27) were female.This is not consistent with the type of data collected.The RLSS used a Likert scale (ordinal), yet a MANOVA would be most applicable for normally distributed, quantitative data.High school coaches scored much higher than college level coaches in democratic behavior.Junior high coaches were significantly lower in training and instruction than either high school or college coaches.