Download presentation
Presentation is loading. Please wait.
1
IEPI – Participate | Collaborate | Innovate
2
Far North Data Disaggregation Workshop September 2017
Identifying Equity Gaps: Data Disaggregation and Disproportionate Impact in Context Far North Data Disaggregation Workshop September 2017
3
Overview of Session Help foster a conceptual understanding of data disaggregation and disproportionate impact Examine the difference between aggregated and disaggregated data Examine three distinct disproportionate impact methods Using the Disproportionate Impact Calculator Session activity (Scorecard pre-work) Brief introduction to multivariate data disaggregation
4
Data Disaggregation Process
Identification of Outcome Outcome Data for All Groups Combined Age Gender Ethnicity Examples: Course Success Rates, Transfer Rate Aggregated Data Disaggregated Data
5
Aggregated Data The first step to identifying equity gaps is to disaggregate the data Example of Aggregated Data Success Rate (%) 69,400 46,010 66.30%
6
Disaggregated Data(Cont’d)
Example of Disaggregated Data Success Rate (%) African-American 2,547 1, % Asian 9,834 7, % Hispanic 35,055 22, %
7
Disproportionate Impact
“Disproportionate impact is a condition where some students’ access to key resources and supports and ultimately their academic success may be hampered by inequitable practices, policies and approaches to student support.” (California Community College Chancellor’s Office, 2013) Differences in outcomes among subgroups of students is commonly referred to as Disproportionate Impact (DI) Example: Differences in access between subgroups groups may suggest that one group has greater access to support services than others any differences in access and/or completion rates between subgroups groups may suggest that one group has greater access to support services and/or is completing degrees at a higher rate than the other. In this case, the group experiencing less access and/or a lower degree completion is said to be disproportionately impacted.
8
Determining Disproportionate Impact
How do we identify disproportionate impact in disaggregated data? Three Methods: 80% Index Proportionality Index Point Gap Index Each method has its advantages and disadvantages The California Community Colleges Chancellor’s Office recommends the percentage point gap method.
9
80% Index Ethnic Group Cohort Count Outcome Count Success Rate (%)
African-American 2,547 1,388 54.50% 74.79% American Indian 213 144 67.61% 92.78% Asian 9,834 7,166 72.87% 100% Hispanic 35,055 22,304 63.63% 87.32% Multi-Ethnicity 2,261 1,468 64.93% 89.10% Pacific Islander 286 153 53.50% 73.42% White 16,696 11,878 71.14% 97.63% Unknown 2,508 1,509 60.17% 82.57% Total 69,400 46,010 66.30% Source: Fullerton College’s Student Equity Plan
10
80% Index Advantages: Clear benchmark for identifying DI
Effective between-group approach Disadvantages Rigid 80% cutoff can curtail discussion May be subject to sampling size error if sample size very small
11
Proportionality Index
Ethnic Group Cohort Success Outcome Proportionality Index Count Percent African-American 2,547 3.67% 1,388 3.02% 0.82 American Indian 213 0.31% 144 1.02 Asian 9,834 14.17% 7,166 15.57% 1.10 Hispanic 35,055 50.51% 22,304 48.48% 0.96 Multi-Ethnicity 2,261 3.26% 1,468 3.19% 0.98 Pacific Islander 286 0.41% 153 0.33% 0.81 White 16,696 24.06% 11,878 25.82% 1.07 Unknown 2,508 3.61% 1,509 3.28% 0.91 Total 69,400 100% 46,010 1.00 Source: Fullerton College’s Student Equity Plan
12
Proportionality Index
Advantages: Effective for assessing equitable representation Effective within-group approach Recommended benchmark is 0.86 (i.e., values of 0.85 or lower are reflective of DI; Sosa, 2016) Disadvantages No universally accepted benchmark
13
Success Rate (Per Group) Success Rate (Overall)
Point Gap Index Ethnic Group Cohort Outcome Outcome Count Success Rate (Per Group) Success Rate (Overall) Point Gap Index African- American 2,547 1,388 54.50% 66.30% -11.8 American Indian 213 144 67.61% +1.3 Asian 9,834 7,166 72.87% +6.6 Hispanic 35,055 22,304 63.63% -2.7 Multi-Ethnicity 2,261 1,468 64.93% +1.4 Pacific Islander 286 153 53.50% -12.8 White 16,696 11,878 71.14% +4.8 Unknown 2,508 1,509 60.17% -6.1 Total 69,400 46,010 Source: Fullerton College’s Student Equity Plan
14
Point Gap Index Advantages:
Easy to calculate (subtraction-based method) Practical findings CCCCO recommended threshold is 2.99 (i.e., values of 3.00 or higher may be reflective of DI) Disadvantages DI of most well-represented group may be obscured No universally accepted benchmark
15
Disproportionate Impact Calculator
16
Session Activity: Site Specific Group Work
Small group activity Student Success Scorecard Data: Review disaggregated trends in transfer-level math and English attainment from Scorecard Identify strategies that may serve to mitigate any observed DI
17
UNDERSTANDING COMPLEX STUDENT IDENTITIES – MULTIVARIATE DISAGGREGATION
18
Disaggregated Data (Bivariate)
Mean Success Rates by Foster Youth Status at College of Marin (FA 12 and FA 13) Mean Success Rate (%) Pg 14 of MCCD equity plan Bivariate because it disaggregates success rate by foster youth status.
19
Disaggregated Data (Bivariate)
Mean Success Rates by Gender at College of Marin (FA 12 and FA 13) Mean Success Rate (%) Pg 14 of MCCD equity plan
20
Disaggregated Data (Multivariate) – Interaction Present
Mean Success Rates by Foster Youth Status and Gender at College of Marin (FA 12 and FA 13) Mean Success Rate (%) Pg 14 of MCCD equity plan
21
Disaggregated Data (Multivariate) Hypothetical Data Illustrating No Interaction
Hypothetical Data: Mean Success Rates by Foster Youth Status and Gender Mean Success Rate (%) Dummy data indicating no interaction Jared: Not really a note, I just really like the FY by gender interaction example. It really is a good visual demonstration.
22
Overview of Multivariate Disaggregation
Multivariate disaggregation can show interactions between demographic characteristics Example: Does the impact that foster youth status have on success rates depend upon students’ gender? Better approximation of real-world student Students simultaneously belong to multiple demographic groups
23
Overview of Multivariate Disaggregation
Where there are two demographic variables, there are two possible individual effects (or main effects) and there is one possible interaction. Addresses three questions: Does disproportionate impact (DI) exist with respect to first demographic characteristic (e.g., foster youth status)? Does disproportionate impact (DI) exist with respect to second demographic characteristic (e.g., gender)? Does disproportionate impact (DI) exist with respect to the joint influence of both demographic characteristic (e.g., foster youth status and gender)? Better approximation of real-world student lives and experiences
24
Multivariate Disaggregation – Tabular Format
Mean Success Rates by Foster Youth Status and Gender Along with Corresponding DI Index Information Foster Youth Status Male Female Difference 80% Index Point Gap Index Yes 42.26% 62.75% 20.49% 36.55% 12.79 No 68.01% 75.50% 7.49% 100% 0.21 Overall 67.54% 75.24% 7.70% Source: College of Marin’s Student Equity Plan ( )
25
Conclusions Identification of equity gaps does not end with data findings Broad-based institutional dialogue is critical Consider employing more than one DI method to identify potential equity gaps The DI calculator offers users a quick way to compute all of the aforementioned indices along with corresponding equity numbers Multivariate disaggregation allows one to identify DI for each demographic characteristic AND to identify a possible interaction between two or more characteristics Enhanced real-world validity
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.