Examining Individual and Organizational Growth

(EDEP 768) Week 15: Assignment 3 2

Assignment 3

Examining Individual and Organizational Growth

Due: December 14, 2011 at 4:00 p.m.

For our assignment on studying growth, we will examine growth over three intervals. We will also use two between-group predictors (gender, SES) and two between-organization predictors (organizational processes and organizational size (coded 1 = under 100 employees and 0 = else). The process indicator is assumed to measure various processes related to outcomes (leadership, expectations, values, climate, and communication). There are 1,864 individuals in 154 organizations. [Refer to assignment3dataset.sav]

Model 1: Partition the Variance at Each Level

Determine about how much variance lies within individuals (level 1), between individuals, and between organizations. [Hint: Specify a model with just the math outcome in the model with schools (schcode) at level 3, between individuals (rid*schcode) at level 2, and repeated measures at level 1 (rid*schcode). This will provide a rough estimate.

Model 2: Establishing the Shape of the Growth Trajectory

The first model below examines possible linear and quadratic growth. It turns out that the quadratic polynomial does not vary over individuals and organizations, so we can consider it as fixed. The linear component will be specified as fixed between individuals but varying across organizations. We can also assume an autoregressive error structure for the repeated measures (AR1).

You might wish to begin by examining whether the growth might best be considered as linear, or whether there is also a quadratic (speeding or slowing) component to the growth.

MIXED math WITH timelin timequad lincon quadcon smallsch nprocess female lowses

/CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0,

ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)

/FIXED=lincon quadcon | SSTYPE(3)

/METHOD=REML

/PRINT=G SOLUTION TESTCOV

/RANDOM=INTERCEPT lincon | SUBJECT(schcode) COVTYPE(VC)

/RANDOM=INTERCEPT | SUBJECT(schcode*rid) COVTYPE(VC)

/REPEATED=time | SUBJECT(schcode*rid) COVTYPE(AR1).

You can check to see whether there is a correlation between the intercept and slope at level 3 (organizations) if you wish using an unstructured covariance matrix with correlation coefficient (UNR). If it is not significant you can return to the default diagonal covariance matrix at level 3 (VC or DIAG).

Model 3: Are the Two Predictors Related to Initial Levels of Y and Change in Y over Time?

After establishing the general shape of the growth trajectories, features of organizations are responsible for differences between organizations in terms of their outcomes. Our main interest, therefore, is in the growth parameter (i.e., in this case the linear slope). We can consider the within group variables as adjustments for possible differences in intercept and growth outcomes due to demographic factors.

Things to Do:

1. Provide a short overview of the study.

2. Determine about how much variance lies within individuals (level 1), between individuals, and between organizations.

3. Examine the shape of the trajectory (including only the linear and quadratic components).

4. We can add the intercept predictors as main effects. Predictors of growth are added as interactions (linear*predictor)

5. Report on your findings.