Upload
julius
View
20
Download
0
Embed Size (px)
DESCRIPTION
Exploring Quasi-Experiments Lab 5: May 9, 2008. Guthrie, J.T., Wigfield, A., & VonSecker, C. (2000). Effects of integrated instruction on motivation and strategy use in reading. Journal of Educational Psychology, 92 , 331-341. Purpose of Research. Causal relationship - PowerPoint PPT Presentation
Citation preview
1
Exploring Quasi-Experiments Lab 5: May 9, 2008
Guthrie, J.T., Wigfield, A., & VonSecker, C. (2000). Effects of integrated instruction on motivation and strategy use in reading. Journal of Educational Psychology, 92, 331-341.
2
Purpose of Research
• Causal relationship– To identify whether the
Concept-Oriented Reading instruction (CORI) intervention produces greater intrinsic motivation and strategy-use than traditional instruction
3
Variables of interest• The Independent Variables (causes)
– Instructional intervention:• CORI intervention• traditional comparison group
• The Dependent Variables (outcomes) – Intrinsic motivation (operationalized: curiosity,
involvement, and preference for challenge)– Extrinsic motivation (operationalized:
recognition and competition)– Strategy use (operationalized: self-report of
cognitive strategies)
• The Covariates– Past achievement (operationalized:
standardized reading achievement scores—CBST/MAT)
4
Participant assignment
• Non-random assignment into intervention and comparison groups (i.e., into classrooms)
• Classrooms assigned based on comparable “subjective matching” of teachers, students, and school settings
5
Cook and Campbell’s UTOS
• Units: 3rd and 5th grade low-achieving students.
• Treatment: Concept-oriented reading instruction (CORI)
• Observations: intrinsic and extrinsic motivation, and strategy use (also past achievement).
• Setting: three different mid-Atlantic grade schools.
6
Threats to Internal ValiditySpecific to Quasi-Experiments• History
– Some event related to the outcome could occur to one group but not the other
• Maturation– Groups may differ in rate of change on the
outcomes prior to treatment
• Instrumentation– Measurement might change from pre to post
test in only one group
• Statistical Regression– Variable used to determine group selection may
be unreliable or unstable
7
Threats to Internal ValiditySpecific to Quasi-experiments• History: A celebrity may visit the schools to
discuss the importance of reading; CORI students may be more susceptible to the message and therefore show greater gains in motivation (due to the celebrity, not the intervention).
• Maturation: Children in the traditional classrooms could be losing motivation at a faster rate than those in the CORI classrooms prior to treatment.
• Instrumentation: Likely not an issue given the lack of pretest design.
• Statistical Regression: The choosing of similar classrooms based on teacher, student, and school make-up could have been based on inaccurate, unreliable, or temporarily skewed subjective judgments.
8
AdditionalThreats to Internal Validity• Lack of pretest makes it difficult to say whether
differences are due to CORI or whether the differences existed at the onset of the research
• Attrition: 11% for grade 3 and 17% for grade 5 due to moving.
• Resentful Demoralization: teachers or students could have shown less motivation knowing that they were not getting the treatment (p.334/ p. 47)
• Compensatory Rivalry: comparison teachers may have tried harder because of the study (mentioned on p. 334 as the John Henry effect).
• Third variables: Teacher expectations could have produced results
• Authors did enhance their design by including extrinsic motivation (p. 332) as a nonequivalent dependent variable (p. 74).
9
Threats to Construct Validity• Diffusion of treatment: Traditional teachers
frequently visited the CORI classrooms and adopted some texts used in the CORI condition. – The authors did collect video, interview, and
questionnaire data regarding the use of the treatment integrity in the CORI classrooms.
• Low internal consistency: Several internal consistencies were low; because reliability puts a lid on validity, this is cause for concern.
• Mono-operation bias: Motivation is captured using only aspects of the Motivation for Reading Questionnaire (MRQ) when other methods may be possible. (e.g., Teacher-reports on individual students).– The authors did address more than one aspect
of motivation (intrinsic and extrinsic)
10
Threats to External Validity
• Schools are chosen based on need, therefore results may not generalize to less needy students/schools.
• Results may not generalize to other grades
• Results may not generalize to interventions that are less intensive.
• Results may not generalize to areas outside of the mid-Atlantic.
• Results may not generalize to participants from dissimilar schools.
11
Threats to Statistical Conclusion Validity
• Violated statistical assumptions: Group administration of treatment violates assumption of independent observations (p. 49).– Addressed by analyzing
the data using HLM
12
Improving the research• Proximal Similarity: The traditional instruction
comparison group could have maintained integrity and differentiation from CORI instruction group. The measures could have more closely approximated intrinsic motivation.– Researchers did a good job employing similar
settings and units to whom they wished to generalize.
– Researchers did a good job ensuring that CORI instruction was being implemented in the CORI classrooms
• Heterogeneous Irrelevancies: Triangulating measures of motivation could have improved construct validity.– The authors did use two different Units and three
different Settings.• Causal explanation: Rule out third variables and
include pretest.