by Kelly S. Shapley and Joan Bush

Increasing teachers' learning opportunities is currently viewed as one of the most important ways to improve the quality of teaching, and research suggests that mentoring by experienced teachers is an important reform strategy (Darling-Hammond and McLaughlin, 1999; National Center for Research on Teacher Learning, 2000). Assigning mentors to work with beginning teachers is an approach that, according to Little (1999), has the potential benefit of transitioning beginning teachers into the classroom, specific school, and district setting in which they will work.

Policies and practices aimed toward establishing mentoring programs for beginning teachers as a means to improve teaching and learning are based on at least three critical assumptions about the impact of mentoring. First, providing induction support will help new teachers adjust to the demands of teaching and become socialized to the school organization (Feiman-Nemser and Remillard, 1995). Second, mentoring will support the pedagogical development of new teachers (Feiman-Nemser, Carver, Schwille, and Yusko, 1999); and third, mentoring will encourage the retention of beginning teachers in the profession (Huling, 1998). These assumptions about mentoring are embedded within a broader supposition that well-qualified, veteran teachers are available to serve as mentors, and that mentoring programs will tap their "accumulated wisdom" to serve teachers and schools" (Little, 1999, p. 345).

When examining mentoring from a policy perspective, it is important to consider how mentoring programs fit within the professional and social culture of schools, as well as the political context of the district and state. To date, little research on teacher mentoring has focused on school and district contextual factors that might influence the effectiveness of mentor programs and practices. Recognizing this missing element, one component of SEDL's research was a quantitative investigation of contextual variables related to districts, schools, and teachers for three case studies of notable mentoring programs in Texas school districts. Considering the current high rate of beginning teacher attrition in Texas, the study explored conditions that might reveal problems new teachers face during the first induction years.

The purpose for the quantitative study was twofold. First, in-depth descriptive information was provided as contextual background for the case-study sites. The analysis explored how existing school, teacher, and student characteristics pose challenges for the mentoring of beginning teachers. Student diversity was of particular interest. Second, researchers investigated a wide range of teacher and school characteristics that were either known or suspected to be associated with teacher quality and retention. The study attempted to identify district- and campus-level factors that might contribute to beginning teacher attrition/turnover in order to understand how mentoring activities can be designed to support novice teachers. Researchers investigated factors such as teacher experience, age, and degrees held, as well as organizational conditions such as student diversity and academic achievement. More specifically, the study addressed two broad questions:

1. What are the characteristics of the case-study districts, schools, students, and teachers (i.e., demographic and academic)?
2. What is each district's current teacher attrition/turnover status, and what are the associated variables (i.e., school conditions or teacher characteristics)?

Selection of the Case-Study Sites

SEDL researchers selected three Texas school districts to study teacher-mentoring programs. The selection process, which was necessarily purposeful, involved three stages. Researchers first narrowed the pool of potential case study sites to a group of districts that included:

• 20 districts with the highest student enrollments,
• districts recommended by the project's advisory board and others familiar with local mentoring programs, and
• districts that, through the mail survey, self-described their mentoring programs as well established or successful.

Second, districts were eliminated from the initial pool of potential sites if:

• no evidence indicated a successful or well-established mentoring program, or
• demographic data did not indicate a notable degree of diversity in the student population (race/ethnicity, limited English proficiency, and economically disadvantaged).

This preliminary site selection process produced eight "finalist" districts. The list of finalists was sent to the advisory board for recommendations. Based on feedback from advisors and further application of selection criteria (established mentoring programs, diverse student population, geographic diversity), the districts were ranked in order of appropriateness for this study. Based on the ranking, districts were contacted and invited to participate in the study. Two of the first-choice districts agreed to participate in the study. The third first-choice district was unable to participate due to program constraints, so one of the second-choice districts was substituted instead. The three districts that provided case study information are referred to in this report by pseudonyms, as follows:

• Urban Independent School District (UISD)
• County Wide Independent School District (CWISD)
• Mid-City Independent School District (MCISD)

Data Sources

Data for the participating districts and schools were derived from three primary sources: (a) a review of district web sites, (b) the Texas Public Education Information Management System (PEIMS), and (c) the Texas Academic Excellence Indicator System (AEIS). The PEIMS and AEIS are data collection systems designed and overseen by the Texas Education Agency (TEA). The TEA systematically collects and compiles standard information each year to produce a comprehensive database that provides comparative statewide statistics for schools. Reports and databases are available on TEA's web site (www.tea.state.tx.us). The PEIMS database includes descriptive information about organizations, district finances, staff, and students. The AEIS pulls together a wide range of information on student performance in each school and district. Performances on a number of indicators (e.g., criterion-referenced tests, attendance rates, dropout rates) are disaggregated by ethnicity, special education, and low-income status. AEIS reports also provide extensive information on school and district staff, finances, programs, and demographics.

Method

Information was extracted from district-level reports and campus-level files to describe the demographic and achievement characteristics of the three case study districts and their campuses. District-level descriptive information was collected through 1998-99 AEIS reports, 1998-99 district accountability tables, and from information on each district's web site. Campus-level descriptive information was extracted from the TEA web site for the 1998-99 school year. AEIS files included campus reference information, student demographic information, student achievement information, and campus staff information. The downloaded files were imported into Excel and merged into one master file containing campus-level information.

PEIMS data for the 1998-1999 to the 1999-2000 school years were obtained through an open-records request to the TEA. The PEIMS data included teacher/special duty teacher identification information, demographic information, employment and salary information, employee responsibilities, and permit information. PEIMS files for each school year were converted to Excel, imported into the SPSS statistical program, and merged into one master file. The 1998-99 campus-level AEIS data were matched to the master files. In subsequent chapters of this report, additional procedures will be described as findings are presented.

District Characteristics

Information in Tables 4.1 and 4.2 provides academic comparisons for the participating districts. Basic demographic information for each site is summarized in the description of case site mentoring programs in Chapter Five.

Academic Achievement. Texas collects a wide range of information on the performance of students, and information is provided in AEIS reports that are available each year in late fall. A school Accountability Rating is derived from a subset of performance measures. Individual schools and districts are classified as either Exemplary, Recognized, Academically Acceptable, or Low Performing.

Student performance indicators reported for this study included the Texas Assessment of Academic Skills (TAAS) passing rates by subject, attendance rates, dropout rates, as well as statistics related to student college admissions. Accountability ratings are reported for districts and schools, but alternative education campuses have a different rating system and indicators. School ratings are not made for early education centers.

As shown in Table 4.1, both MCISD and CWISD received the Academically Acceptable rating. UISD was rated Unacceptable because of data quality problems and low-performing schools. UISD had the highest percentage of Exemplary campuses (9 percent), while MCISD had far more Recognized campuses (41 percent). The majority of schools received Academically Acceptable ratings.

Table 4.1 District Academic Profiles
	MCISD	CWISD	UISD
AEIS Accountability Ratings
District Rating	Acceptable	Acceptable	Unacceptable
Campus Ratings (Regular)	N=37	N=38	N=96
Exemplary	2 (5%)	2 (5%)	9 (9%)
Recognized	15 (41%)	4 (11%)	7 (7%)
Academically Acceptable	20 (54%)	29 (76%)	63 (66%)
Low Performing	0 (0%)	0 (0%)	16 (17%)
Campus Ratings (Alternative Ed.)	N=3	N=4	N=1
Alternative Ed.: Acceptable	2	4	1
Alternative Ed.: Needs Review	1	0	0
Student College Admissions
Taking SAT/ACT	58%	37%	63%
Scoring Above SAT/ACT Criterion	23%	30%	43%
Note. Statistics based on TEA 1999 district accountability summaries and 1998-99 AEIS Reports. In CWISD, three early education centers were not rated.

District college admission statistics showed that higher percentages of MCISD (58 percent) and UISD (63 percent) students took the SAT/ACT, and a markedly higher percentage of UISD students (43 percent) scored above the SAT/ACT criterion.

District student attendance and dropout rates, and percent passing the state-mandated achievement measure, which are shown in Table 4.2, further explain each district's accountability rating.

Table 4.2 District Student Accountability Data Percentages
	MCISD	CWISD	UISD
Attendance Rate	96	94	94
Dropout Rate-All	1.9	3.4	2.0*
African American	2.1	3.5	2.7*
Hispanic	2.4	4.5	2.9*
White	1.6	2.4	0.8*
Economically Disadvantaged	1.9	3.4	2.2*
Students Passing TAAS Reading-All	87	81	79
African American	83	72	68
Hispanic	86	74	70
White	91	89	93
Economically Disadvantaged	83	73	66
Students Passing TAAS Math-All	85	79	76
African American	80	64	57
Hispanic	85	75	67
White	90	86	90
Economically Disadvantaged	82	73	63
Students Passing TAAS Writing-All	85	81	82
African American	79	73	72
Hispanic	85	75	73
White	89	88	92
Economically Disadvantaged	84	73	69
Note. Statistics based on percentages in 1998-99 AEIS reports. TAAS refers to the Texas Assessment of Academic Skills. * Dropout rates for UISD were underestimated due to data-quality problems.

Compared to MCISD, dropout statistics for CWISD were considerably higher for all students and subgroups, with the exception of White students. The dropout statistics for UISD were underestimated due to data-quality problems. TAAS reading, mathematics, and writing passing rates were comparable across all districts for White students--however, MCISD's African American, Hispanic, and disadvantaged students performed notably better than those student groups in the other districts. The district's passing rates were generally 10 or more percentage points higher for each of the subjects.

School Characteristics

Additional analyses were undertaken to explore school-level differences within and across districts. Due to the large number of campuses in each district, grouping variables were used to reduce student and teacher data to an understandable form. Considering previous research literature on teacher attrition/retention and emerging district-level differences, three categories were created for comparison purposes: school level, school diversity, and school achievement. School level was defined as elementary, middle, and high schools. School diversity was a dichotomous variable based on two levels: Low Diversity -- the percentage of nonwhite students was less than 55 percent, and High Diversity -- the percentage of nonwhite students was 55 percent or more. School achievement was a dichotomous variable based on two levels: Low Achievement -- a campus accountability rating of Acceptable or Low Performing, and High Achievement -- a campus accountability rating of Exemplary or Recognized.

Student Demographic Characteristics. Student demographics are summarized in Table 4.3 by school type. Across all districts, higher percentages of elementary students were classified as economically disadvantaged. This is probably because parents are more likely to enroll lower-grades students in the national free- and reduced-price lunch program. Higher percentages of elementary students were classified as limited English proficient, but student ethnicity was relatively stable across school levels in all districts.

Important district-level differences emerged for school diversity and school achievement. For MCISD, student characteristics were generally stable for low and high achieving schools, except, as expected, highly diverse schools had higher percentages of nonwhite students. Findings were similar for CWISD. Student characteristics were relatively stable for low and high achieving schools, however, there tended to be more limited English proficient students and economically disadvantaged students in highly diverse schools.

In UISD, dramatic differences were evident for both school diversity and school achievement. Highly diverse schools had significantly greater percentages of limited English proficient (+17 points), economically disadvantaged (+49 points), and nonwhite (+51 points) students. The same pattern was true for achievement. High percentages of students in low achieving UISD schools were limited English proficient (21 percent), economically disadvantaged (70 percent), and nonwhite (82 percent).

Table 4.3 Student Demographic Characteristics–Percentages by School Type
	School Level			Diversity (1)		Achievement (2)
	Elem.	Middle	High	Low	High	Low	High
Mid-City ISD
Campus N	25	8	2	12	31	20	17
Student N	15,673	6,166	4,764	6,960	21,573	16,788	11,610
LEP	6	2	3	4	4	4	6
Economically Disadv	61	41	31	47	53	54	57
African American	39	40	37	25	47	40	38
Hispanic	16	16	17	17	40	16	17
White	41	38	39	56	32	39	42
Nonwhite	59	62	61	44	68	61	58
County Wide ISD
Campus N	29	6	2	13	29	29	6
Student N	16,220	6,455	4,371	10,529	17,860	23,252	3,328
LEP	21	9	6	9	20	17	11
Economically Disadv	68	52	31	44	72	62	54
African American	5	6	5	5	6	5	5
Hispanic	56	49	42	40	62	54	49
White	37	45	52	53	31	40	44
Nonwhite	63	55	48	47	69	60	56
Urban ISD
Campus N	68	15	11	34	68	63	16
Student N	42,947	16,023	19,995	29,525	49,971	46,723	10,820
LEP	19	10	4	4	21	21	4
Economically Disadv	60	50	34	22	71	70	13
African American	18	18	21	8	25	24	4
Hispanic	48	45	39	21	57	56	16
White	31	35	37	68	17	18	76
Nonwhite	69	65	63	32	83	82	24
Note. Statistics based on 1998-99 AEIS reports. Percents may not sum to 100 due to rounding. 1: Low Diversity–nonwhite is less than 55 percent; High Diversity–nonwhite is 55 percent or more. 2: Low Achievement–accountability rating is Acceptable, Low Performing; High Achievement–accountability rating is Exemplary, Recognized.

Student Achievement. Student achievement is summarized in Table 4.4 by school type. For all districts, TAAS passing rates varied only slightly by school level, but there were notable school diversity and achievement trends. First, there were only small differences in average MCISD student outcomes regardless of school diversity and achievement level. Second, achievement differences were greater in CWISD; student performance, on average, was about 10 percentage points lower in highly diverse or lower achieving schools. Third, the average student passing rates in highly diverse and low-performing UISD schools were generally about 20 percentage points lower than the comparison schools. In sum, although all districts were serving diverse populations, the achievement outcomes for nonwhite and disadvantaged students were quite different, and outcomes in some districts were strongly tied to the type of school that students attended.

Table 4.4 Student Achievement–Passing Percentages by School Type
	School Level			Diversity (1)		Achievement (2)
	Elem.	Middle	High	Low	High	Low	High
Mid-City ISD
Campus N	25	8	2	12	31	20	17
Student N	15,673	6,166	4,764	6,960	21,573	16,788	11,610
TAAS Reading-All	90	85	86	91	85	86	91
TAAS Math-All	88	85	75	89	83	82	91
TAAS Writing-All	89	83	89	90	83	85	91
County Wide ISD
Campus N	29	6	2	13	29	29	6
Student N	16,220	6,455	4,371	10,529	17,860	23,252	3,328
TAAS Reading-All	82	79	84	85	77	79	91
TAAS Math-All	82	78	74	83	73	78	92
TAAS Writing-All	84	75	85	86	77	81	92
Urban ISD
Campus N	68	15	11	34	68	63	16
Student N	42,947	16,023	19,995	29,525	49,971	46,723	10,820
TAAS Reading-All	79	75	82	92	72	73	95
TAAS Math-All	77	72	72	90	68	69	94
TAAS Writing-All	81	74	85	93	75	76	96
Note. Statistics based on 1998-99 AEIS reports–percent passing TAAS by subject. 1: Low Diversity–nonwhite is less than 55 percent; High Diversity–nonwhite is 55 percent or more. 2: Low Achievement–accountability rating is Acceptable, Low Performing; High Achievement–accountability rating is Exemplary, Recognized.

Teacher Characteristics. Additional analyses were undertaken to explore how teacher characteristics varied by school types within districts. Using 1999-2000 teacher data, a series of crosstabs were run to review teacher characteristics by campus characteristics. Findings are summarized in Table 4.5 by teacher experience, age, degree, ethnicity, gender, and annual salary. Teacher experience had five categories: beginning teachers in their first teaching year, developing teachers with 1 to 5 years experience, and veteran teachers with 6 to 10 years, 11 to 15 years, and 16 or more years experience. Teacher age was a three-category set: younger (less than 30), middle-aged (30-50), and older (greater than 50). A dichotomous ethnicity variable was created with categories for "white" and "nonwhite."

Table 4.5 Teacher Demographic Characteristics-Percentages by School Type
	School Level			Diversity (1)		Achievement (2)
	Elem.	Middle	High	Low	High	Low	High
Mid-City ISD
Campus N	25	8	2	12	31	20	17
Teachers N	1,039	435	449	419	1,468	1,139	759
Teacher Experience
First year	10	9	4	10	8	8	10
1-5 years	35	41	35	35	36	36	37
6-10	19	17	18	17	19	18	18
11-15	13	10	15	14	13	13	14
16+	23	23	28	24	25	26	21
Teacher Age
Less than 30	20	18	14	18	18	17	20
30-50	61	62	58	61	61	59	62
51+	19	20	29	22	22	24	18
Highest Degree (3)
Bachelors	87	80	72	86	80	79	87
Masters+	13	20	24	14	18	20	13
Ethnicity
White	81	75	79	80	79	79	79
Nonwhite	19	25	21	20	21	21	21
Male	9	30	40	14	23	27	11
Median Salary	$32,102	$29,700	$34,242	$31,200	$32,102	$32,012	$31,200
County Wide ISD
Campus N	29	6	2	13	29	29	6
Teachers N	791	338	353	554	871	1,234	174
Teacher Experience
First year	5	12	5	5	8	7	4
1-5 years	20	38	24	23	26	26	14
6-10	24	15	18	21	21	20	26
11-15	17	8	18	14	16	15	18
16+	34	27	36	38	30	32	39
Teacher Age
Less than 30	10	17	6	9	11	11	6
30-50	65	61	64	65	64	64	68
51+	25	22	30	26	25	25	26
Highest Degree (3)
Bachelors	86	85	69	80	83	81	87
Masters+	14	15	28	20	16	18	13
Ethnicity
White	75	79	83	83	74	78	81
Nonwhite	25	21	17	17	26	22	19
Male	10	32	43	25	21	25	8
Median Salary	$34,480	$30,800	$31,470	$32,990	$31,470	$31,470	$38,344
Urban ISD
Campus N	68	15	11	34	68	63	16
Teachers N	2,859	944	1,091	1,664	3,209	3,005	662
Teacher Experience
First year	10	10	8	5	12	11	3
1-5 years	29	35	26	25	31	30	23
6-10	17	16	14	17	16	16	18
11-15	14	11	12	14	12	12	17
16+	32	28	40	40	30	31	39
Teacher Age
Less than 30	18	19	14	13	20	19	12
30-50	62	61	58	63	60	60	68
51+	20	19	28	24	21	21	21
Highest Degree (3)
Bachelors	75	74	62	69	72	72	72
Masters+	25	26	37	31	27	28	28
Ethnicity
White	67	72	77	84	63	63	86
Nonwhite	33	28	23	16	37	37	14
Male	10	30	43	20	22	19	7
Median Salary (4)	-	-	-	-	-	-	-
Note. Statistics based on 1998-99 AEIS reports. Percents may not sum to 100 due to rounding. 1: Low Diversity–nonwhite is less than 55 percent; High Diversity–nonwhite is 55 percent or more. 2: Low Achievement–Accountability Rating is Acceptable, Low Performing; High Achievement–Accountability Rating is Exemplary, Recognized. 3: Small percentages of teachers held "no degree." 4: Salary data were omitted due to data-quality problems.

Some teacher differences were evident by school level across all districts. Beginning teachers were more likely to work in elementary or middle schools, and high-school teachers were typically more experienced. Correspondingly, elementary and middle school teachers were generally younger than their high-school peers were. Middle and high-school teachers were more likely to be male, and a higher proportion of high-school teachers had advanced degrees.

For MCISD, teacher characteristics were generally stable by school diversity and achievement, except there were higher percentages of male teachers in highly diverse and low-performing schools. In contrast, teacher characteristics varied greatly by diversity and achievement in the other two districts. Teachers in highly diverse and low achieving schools (which were generally the same schools) were typically less experienced, somewhat younger, and more ethnically diverse. In lower achieving schools, higher proportions of teachers were male. Teacher salary differences were generally small, except in high achieving schools in CWISD. In that district, the median teacher salary in high-achieving schools ($38,344) exceeded the salary in low-achieving schools by almost $7,000.

Teacher mobility was examined for the district and for school campuses. As a first step, data files from two contiguous years (i.e., 1998-99 and 1999-2000) were merged to determine the proportion of teachers who left the school district, continued teaching at the same campus, or moved to a different district campus. Next, operational definitions were created for districts and campuses as follows:

• District Level
  • Teacher Attrition = (leavers)
  • Teacher Retention = (stayers + movers)
• Campus Level
  • Teacher Turnover = (leavers + movers)
  • Teacher Stability = (stayers)

Frequency analyses were conducted to determine the district-level attrition rate (leavers) and retention rate (stayers + movers), as well as the campus-level turnover rate (leavers + movers) and stability rate (stayers) for the three case-study districts. Chi square analyses were calculated for the three districts to note if statistically significant differences existed between teacher attrition/turnover and selected school and teacher characteristics, and to determine if variations between case study districts were present.

Leavers, Movers, and Stayers

Figure 4.1 shows the proportion of leavers, movers, and stayers by district for 1999-2000. MCISD had the most stable teaching force, with the highest proportion of stayers (80 percent) and the lowest percentage of within-district movers (4 percent). A markedly higher percentage of teachers left CWISD (27 percent) compared to the other districts (16 percent, 18 percent). This extreme difference is difficult to explain, but TEA data analysts confirmed statistical accuracy. Further investigation is needed to determine why more than one-fourth of the teachers left the district.

As shown in Figure 4.2, there was little difference between the district attrition and campus turnover rates for MCISD. On the other hand, the campus turnover rate exceeded the district rate by eight percentage points for CWISD and nine points for UISD. High campus-level rates meant that between one-fourth to one-third of the teachers either left the districts or moved to different campuses.

District-Level Retention and Attrition

Findings presented in Table 4.6 show how teacher retention and attrition rates varied by school and teacher characteristics. Correspondingly, as reported in Table 4.7, statistical tests for associations among variables were performed using chi-square tests of significance.

Table 4.6 District-Level Retention/Attrition Percentages by School and Teacher Characteristics
	Mid-City ISD				County Wide ISD				Urban ISD
		RET	ATT			RET	ATT			RET	ATT
	N	%	%		N	%	%		N	%	%
School Characteristics
School Level
Elementary	928	85	15	775		74	26	2,076		82	18
Middle	410	81	19	358		70	30	777		80	20
High	425	83	17	336		73	27	840		83	17
Diversity
Low	376	85	15	553		78	22	1,276		87	13
High	1,390	83	17	916		70	30	2,446		80	20
Achievement
Low	1,066	84	16	1,238		73	27	2,239		80	20
High	677	84	16	166		75	25	474		88	12
Teacher Characteristics
Teacher Experience
First year	175	74	26	150		53	47	330		72	28
1-5 years	616	76	24	460		56	44	1,057		73	27
6 or more years	987	90	10	866		85	15	2,373		87	13
Teacher Age
Less than 30	347	69	31	213		50	50	620		67	33
30-50	1,052	86	14	938		76	24	2,337		85	15
51+	379	91	9	325		79	21	803		86	14
Highest Degree
No degree	20	95	5	11		79	21	7		100	0
Bachelors	1,444	84	16	860		72	28	2,673		81	19
Masters+	314	82	18	202		78	22	1,080		85	15
Ethnicity
White	1,192	84	16	1,167		72	28	2,674		82	18
Nonwhite	236	84	16	309		74	26	1,086		82	18
Gender
Male	383	86	14	380		69	31	921		84	16
Female	1,395	83	17	1,096		74	26	2,839		81	19
Note. RET=Retention Rate. ATT=Attrition Rate.

Table 4.7 Chi Square Test Results of School and Teacher Characteristics by District-Level Teacher Retention/Attrition
		MCISD	CWISD	UISD
	df	C2	C2	C2
School Characteristics
School Level	2	3.47	1.67	3.22
Diversity	1	0.71	11.37***	26.32***
Achievement	1	0.00	0.47	14.96***
Teacher Characteristics
Teacher Experience	2	64.11***	168.74***	158.06***
Teacher Age	2	69.15***	64.75***	122.11***
Highest Degree	2	2.90	4.31	10.05**
Ethnicity	1	0.01	0.39	0.09
Gender	1	2.01	4.15*	3.18*
Note. *p<.05. **p<.01. ***p<.001.

School Characteristics. At the campus level, teacher attrition and retention are examined according to school level, diversity of student population, and student achievement.

School level. There was little difference in teacher attrition rates by school level (i.e., elementary, middle, and high). Chi-square analyses revealed no significant school-level differences for any district.

School diversity. In CWISD and UISD, teacher attrition was significantly associated with school diversity (C2 = 11.37, df = 1, p < .001 and C2 = 26.32, df = 1, p < .001, respectively). Teachers in highly diverse schools in those districts were more likely to leave than were teachers at less diverse schools. Observed differences were likely related to the nature of student diversity within those districts--white/nonwhite proportions in some schools were extreme, whereas in MCISD schools were more ethnically balanced.

School achievement. For UISD, teacher attrition was strongly associated with school achievement (C2 = 14.96, df = 1, p < .001). Teachers working at low-performing UISD campuses were more likely to leave the district (20% attrition rate) compared to teachers in high achieving schools (12 percent). In CWISD, teacher attrition rates were high for both low performing (27 percent) and high-performing campuses (25 percent).

Teacher Characteristics. The teacher characteristics considered include professional experience in the classroom, age, highest academic degree, ethnicity, and gender.

Teacher experience. Across all districts, teacher attrition rates declined incrementally as years of teaching experience increased. Chi-square analyses revealed that teacher attrition was significantly associated with teacher experience at p < .001 for all districts. First-year and developing teachers (1-5 years) were more likely to leave the districts than their more experienced counterparts. In MCISD and UISD, about one-fourth of the beginning and developing teachers left, whereas in CWISD, over half of the inexperienced teachers left the district.

Teacher age. The same trend was evident for teacher age, which is highly related to teacher experience. Teachers who were less than 30 years old were significantly more likely to leave the districts (p < .001). From one-third to one-half of the youngest teachers left the districts.

Highest degree. The degree teachers held was not highly associated with teacher attrition, except in UISD, where a small number of teachers with no degree were more likely to stay.

Ethnicity. Teachers' ethnicity was not associated with teacher attrition.

Gender. Gender was not strongly associated with teacher attrition, but males in CWISD and UISD were somewhat more likely to leave the districts (p < .05).

Campus-Level Stability and Turnover

Findings presented in Table 4.8 show how campus-level teacher turnover and stability rates varied by school and teacher characteristics within districts. Corresponding Chi square statistical tests for associations among variables are reported in Table 4.9 by district.

School Characteristics. At the campus level, the same characteristics (school level, diversity of student population, student achievement) are considered again in examining campus stability and turnover.

School level. School level (i.e., elementary, middle, and high) was not strongly related to teacher attrition at the district level--however, there were important campus-level differences. Teacher turnover was associated with school level (p <.05) in all districts, and in every case, it was middle school teachers who were more likely to either leave or move to a different campus.

School diversity. School diversity was strongly associated with teacher turnover in CWISD (C2 = 12.14, df =1, p < .01) and UISD (C2 = 20.57, df =1, p < .001). Highly diverse campuses in those districts had markedly higher teacher turnover rates (38 percent, 29 percent, respectively)--thus, school staffs were unstable. One-fourth to one-third of the teachers left diverse schools.

School achievement. A similar teacher turnover trend was evident for school achievement (i.e., low and high). Teacher turnover was associated with school achievement in all districts, but the relationship was strongest in UISD (C2 = 13.60, df = 1, p < .001). High percentages of both CWISD (36 percent) and UISD (29 percent) teachers left low-performing campuses.

Table 4.8 Campus-Level Stability and Turnover by School and Teacher Characteristics
	Mid-City ISD			County Wide ISD			Urban ISD
		STA	TRN		STA	TRN		STA	TRN
	N	%	%	N	%	%	N	%	%
School Characteristics
School Level
Elementary	928	81	19	775	66	34	2,076	73	27
Middle	410	75	25	358	58	42	777	69	31
High	425	81	19	336	70	30	840	79	21
Diversity
Low	376	81	19	553	71	29	1,276	79	21
High	1,390	79	21	916	62	38	2,446	71	29
Achievement
Low	1,066	78	22	1,238	64	36	2,239	72	28
High	677	81	19	166	72	28	474	80	20
Teacher Characteristics
Teacher Experience
First year	175	70	30	150	42	58	330	64	36
1-5 years	616	72	28	460	48	52	1,057	63	37
6 or more years	987	86	14	866	78	22	2,373	79	21
Teacher Age
Less than 30	347	64	36	213	43	57	620	58	42
30-50	1,052	82	18	938	67	33	2,337	75	25
51+	379	88	12	325	75	25	803	80	20
Highest Degree
No degree	20	95	5	7	79	21	14	86	14
Bachelors	1,444	80	20	2,673	64	36	1,202	72	28
Masters+	315	77	23	1,080	69	31	260	77	23
Ethnicity
White	1,428	79	21	1,167	65	35	2,674	74	26
Nonwhite	350	80	20	309	65	35	1,086	72	28
Gender
Male	383	80	20	380	61	39	921	76	24
Female	1,395	79	21	1,096	67	33	2,839	72	28
Note. STA=Stability Rate. TRN=Turnover Rate.

Table 4.9 Chi Square Test Results of School and Teacher Characteristics By Campus-Level Teacher Stability/Turnover
		MCISD	CWISD	UISD
	df	C2	C2	C2
School Characteristics
School Level	2	7.46*	12.14**	20.57***
Diversity	1	0.93	11.02**	27.25***
Achievement	1	2.15*	3.45*	13.60***
Teacher Characteristics
Teacher Experience	2	60.86***	160.50***	142.75***
Teacher Age	2	71.67***	61.17***	92.38***
Highest Degree	2	3.84	2.97	11.76**
Ethnicity	1	0.18	0.01	1.22
Gender	1	0.14	4.45*	5.54*
Note. *p<.05. **p<.01. ***p<.001.

Teacher Characteristics. The same set of teacher characteristics (professional experience in the classroom, age, highest academic degree, ethnicity, gender) is used again to consider campus stability and turnover.

Teacher experience. Across all districts, teacher experience was a strong predictor of teacher turnover. Chi-square tests showed statistically significant associations for all districts at p < .001. In CWISD, astonishingly, 58 percent of the first-year teachers either moved to another campus or left the district. First-year teacher turnover percentages were somewhat lower for MCISD and UISD (30% and 36 percent, respectively). Teacher turnover rates declined dramatically when teachers had more than six years teaching experience. Overall, high percentages of beginning and developing teachers either left the districts or moved to different campuses.

Teacher age. The same trend was evident by teacher age. Teachers who were less than 30 years old were significantly more likely to leave the districts or their assigned campuses (p < .001).

Highest degree. Teachers' degree was not significantly associated with teacher turnover, except in UISD, where teachers with no degree were less likely to move.

Ethnicity. Teachers' ethnicity was not associated with teacher turnover.

Gender. Gender was not strongly associated with teacher turnover, but males were move likely to leave their campuses in CWISD and females were somewhat more prone to move in UISD.

Summary

Findings from the quantitative analysis revealed important factors that pose challenges for mentoring programs. Foremost, districts and schools vary greatly, so mentoring programs and practices must accommodate distinctive aspects of the school context. Evidence from this study showed that first-year and developing teachers were often assigned to highly diverse and low achieving schools. This suggests that mentoring programs, in many instances, will be challenged to help novice teachers deal with the unique needs of diverse and at-risk student populations.

Results confirmed that district-level teacher attrition is, indeed, a significant problem, and teacher experience is significantly associated with attrition. Between one-fourth to one-half of the beginning and developing teachers left the districts. Teacher mobility is an even more critical problem at the campus level. Campus-level teacher turnover rates, which ranged from 30 percent to 58 percent for first-year teachers, suggested that many schools, particularly middle schools, diverse campuses, and low-performing schools, must continually induct new and inexperienced teachers. Such schools will have difficulty building supportive, collaborative cultures and human resources to support high-quality mentoring for beginning teachers. In those schools, support may be needed from external sources.

Some evidence suggested that a school district's policies and procedures regarding the assignment of beginning teachers might intensify or lessen the teacher attrition/turnover problem. Results for one district showed that teacher turnover rates were considerably lower when schools were ethnically balanced, had equitable percentages of at-risk students, and blended beginning, developing, and veteran teachers. It also seems likely that mentoring programs will be more effective when such supportive conditions exist.

In the final chapter of this report (Chapter Six), conclusions and implications drawn from these findings are incorporated with learnings from the two other research strands.