Sample Results Section 

---

NOTE: The WWW plays havoc with formatting! This document
has some charts and graphs which are supposed 
to be CENTERED but are not...
There are also one analysis oversight and one 
formatting error in this document. 
They will be revealed at the end!
NOTE ALSO that this document DOES NOT 
comply with good APA format, such as indentations 
or margins, etc. Get the APA manual or 
Seattle U. students can get our short version 
from the fifith floor receptionist.

Sampl e Results Section, Based on the "Fake CC Data"

RESULTS

Introduction

In this section, we will describe the results of our survey of community colleges in the Pacific Northwest. As you will recall from the previous section, 36 of the 39 community colleges in the target region completed their survey forms. The sample excluded vocational-technical colleges and private institutions. While we did not obtain a perfect return rate, we feel our results are quite likely to be representative of the target population of institutions in the Pacific Northwest in the late 1990's, and somewhat representative of community colleges on a national level.

Description of Variables

The surveys gave six variables for analysis. The first, Location, was a categorical variable, with the levels of Urban, Suburban, and Rural. This variable was coded by the researcher, using census data from the 1990 census as follows: if the campus of the college was in a standard metropolitan census area of more than 50,000 persons, it was coded Urban. If the area had between 10,000 and 50,000, the code was Suburban, unless the city was more than 50 mile from a city with population 50,000 or more. In this c ase, the code was Rural. Rural was also coded if the college was in an area with population less than 10,000. This coding scheme follows the prior research of Schmidlap and Dugong (1996) on the effects of location on community colleges.

The n ext two variables, Enrollment and Percent Women, are numerical (interval level) variables taken from the survey and confirmed with the State's most recent enrollment data (from school year 1996-97). The data on percent of women enrolled is considered to have an error rate of +/- 3% due to reporting errors (See Manatee and Sirena, 1994).

The variable Women Leaders was derived from the survey. This was defined (see survey, appendix 1) as the number of women in leadership positions at the colleg e. Leadership positions were defined as roles at or above the level of Dean. The reader should keep in mind that, due to the variation in size of the administration at these colleges, these numbers, which are an actual count, do not reflect the percenta ge of the total number of administrators at each college.

The final two variables are both nominal (yes/no) questions, taken directly from the survey and confirmed by telephone. The variable Federal Grants > $100K asks whether or not the co llege had received in 1996 Federal grants of over one hundred thousand dollars. The variable Womens' Center reports whether or not the college campus has a Womens' Center.

Descriptive and Inferential Statistics

The d escriptive results for the three nominal variables are described first. The first variable examined was Location. Of the 36 colleges, there were 12 in each of the three locations of Urban, Suburban, and Rural. There was also an even distribution of eig hteen college with, and eighteen colleges without, Federal grants over $100,000. As can be seen in Figure 1, below, the variable Womens' Center was not evenly distributed, with thirteen of the colleges with, and twenty-three without, a Womens' Center on campus.

The interrelations of these nominal variables can be seen below. Table 1 examines the relationships between the location variable and the presence or absence of a Womens' Center. As can be seen from both of these analyses, the number of colleges with a Womens' Center was even in the Rural location, (six vs. six), but was two to one against (four with, eight wit hout) in the Urban location. The strongest difference, three with and nine wit hout, occurred in the Suburban location.

Since we are trying to examine the factors that lead to the presence of women leaders on a campus, the interrelations among these nominal variables was exa mined, to check for other possible relations. The relationship between location a nd the presence or absence of a Womens' Center was examined using Chi-squared. While there were differences, as shown above, these were not significant, with c 2 = 1.6, p= .43,.

A seco nd Chi-squared was performed to examine the relationship between presence or absence of a Womens' Center and Federal grants totaling more than $100,000. While there were 18 each in the yes and no categories for grants, the split within these two groups o n a Womens' Center was very different, as can be seen in Table 3.

These results yielded a Chi-squared of 3.01, with one degree of freedom; this approached statistical significance at p=.08. Thus, the collages that did have more than $100,000 in Federal grants were equally likely to have a Womens Center or not. But the colleges with less Federal money were much less likely to have one--four with to fourteen without--and this difference approached statistical significance.

Next, descriptive results were examined for the three interval level measures. The descriptive statistics for Enrollment, percent women enrolled and n umber of women leaders are shown below as Table 4. As you can see from the standard deviations, especially in the variables Enrollment and percent women, there was a great degree of variability in this data. After examining this, it appears that the ran ge of the variables, as well as some negative skewing (with enrollment having several cases under 4,000). Negative skewing also seems responsible for the high standard deviations in percent women (a number of cases with values under 10) and women leaders (cases under 3).

The distribution of the sizes of the colleges (Enrollment) is shown bel ow as Figure 2. One of the analyses necessary to prevent spurious conclusions was to examine the possible effects of Location on the variables enrollment, percent women, and women leaders. Although our hypothesis is concerned with women leaders, there ma y be a variety of causes and interrelations here. The next three analyses are of the effects of Location on the variables mentioned, using ANOVA. There were mean differences in enrollment by location, as shown in Table 5, but these were nonsignificant ( F=.61, DF= 2,33

The variables of percent women and women leaders were also examined with ANOVA. While the percentage of women ranged from 22.3 (Suburban) to 20.5 (Urban) to 19.4 (Rural), this difference did not approach statistical significance. Similarly, Ur ban schools had 12.0 women leaders, while other locations had between ten and eleven leaders, but this difference did not approach significance.

Next, we examined whether the presence or absence of Federal grants of $100,000 or more had an eff ect on enrollment, percent women, or women leaders. The eighteen colleges with Federal grants more than $100,000 had slightly lower enrollments, averaging approximately 6,200 students, while those with less grants money averaged a little more than 7,000 students, a nonsignificant difference. Women enrolled was also nonsignificantly related to grants, with schools with more than $100,000 showing 22.6% women, vs. 18.8% at schools without that level of Federal grant support.

Women leaders averag ed 11.8 at schools with, and 10.5 at schools without Federal grants over $100,000., a nonsignificant difference.

Finally, we examined our hypothesis, that the presence or absence of a Womens' Center on campus would effect the percent of women e nrolled and the number of women leaders. We first checked to see if the campuses with and without Women's Centers had different levels of enrollment. Campuses with (n=13) had approximately 7200 students on the average, vs. approximately 6300 for the 23 that did not. This was a nonsignificant result.

There was virtually no difference (20.5 vs. 20.9) for the campus with and without Womens' centers, so our hypothesis that Womens' Centers would be associated with a higher percentage of women enr olled was not confirmed. It seems that the many other forces determining college enrollment are more powerful than the presence or absence of the Womens' Center.

Our final hypothesis, that the schools with centers would have more women leaders , proved to be stronger, although not as strong as we would have liked. The thirteen schools with Womens' Centers had and average number of women leaders of 13.8, while the other twenty three had an average of 9.7. This strong trend just missed statisti cal significance at p=.06.

In summary...

------------------------ So, what was wrong (Other than formatting, such as indenting, and tables not centered?) The last table had 3 decimal places, completely inappropriate for N=36 (!) A fairly obvious analysis was overlooked!! This fake research wanted to look at the effects of Women's Centers on women enrolled and women leaders. Women's centers had no effect on women's enrollment and an almost significant effect on women leaders. But there was a significant predictor of women leaders which was not analyzed!!! (And it should have been.)