The First Five Years of Mathematics Teaching

Principal Investigator: 
Project Overview
Background & Purpose: 

A study of novice teachers’ development of mathematical knowledge for teaching and the influence of previous preparation, school context and opportunities to learn-on-the-job, on that knowledge. From our TEDS-M study we are learning how pre-service teacher education impacts future teachers’ mathematics knowledge for teaching in 17 countries including the US. We are extending our inquiry into the first 5 years of teaching in a selected group of countries to explore the connections between pre-service preparation and what is learned on the job as it concerns knowledge, skills and curricular content; and the degree to which standards, accountability and other similar mechanisms operate to regulate the support that beginning teachers of mathematics receive during their first years of teaching.


We will survey six representative and independent samples of teachers with zero to 5 years of teaching experience in the U.S. and in 5 to 9 other high achieving countries in international mathematics assessments such as TEDS-M, TIMSS and PISA.

Research Design: 

The project has a comparative research design and will generate evidence that is descriptive [design research] and associative/ correlational [analytic essay and interpretive commentary]. Original data is being collected from beginning primary and secondary teachers with 0-5 years of experience using school records, observation [videography], and survey research [online]. Instrument or measures include Beginning Teacher Questionnaires (BTQ), Principal and Mentor Questionnaires, and Curriculum Analyses.

An important outcome of the analysis will be the creation of a general model of teacher development that can be investigated using multi-group structural equation modeling. Initial analyses of the response data from the teachers’ BTQ will have the goal of examining the intended structure of the results from the instrument. These analyses will be of two types. First, confirmatory analyses will be run using the Conquest and the MPlus software to check if there is adequate fit between the hypothesized data structure and the observed responses. We will do an exploratory multidimensional item response theory analysis to check for unanticipated constructs within the data (DeAyala, 2009). The constructs that are verified through these analyses will be the fine-grained basis for reporting. In addition this analysis will help us determine the composite of skills that best summarizes the complex structure of the instrument. This will be the single summary score used for reporting.

A second part of the initial analyses of the BTQ will be an investigation of the measurement invariance of the mathematics knowledge for teaching and mathematics teaching practice sections of the instrument across countries. Measurement invariance refers to an instrument having the same multidimensional structure across countries (e.g., the instrument measures the same constructs and they have the same inter-correlations across countries). The measurement invariance will be checked using multi-group multivariate procedures, specifically using innovative analyses such as latent variable modeling (LVM) with the software MPlus which contains modules specifically developed to evaluate measurement invariance. If the measurement invariance is not supported by the analysis, then the deviations from invariance will be determined. It may be that the same constructs are assessed, but that they have different correlations; or, that the constructs may be different across countries (these hypotheses can be examined using LVM and Mplus as it readily permits point and interval estimation of latent correlations, as well as the evaluation of country-specific latent structure). A final outcome of this process will be determining common constructs and aligning them so that their measures can be compared across countries. The alignment process may result in different correlations among constructs in different countries. These correlational differences or lack of differences are an important outcome of the study.

Target Population: 
Research Design: 


Subscribe to RSS - Michigan