Finally, our preferred model groups the estimates into 20-percentage point bins. The estimates are somewhat noisy, but relative to those in the low reactivity group (from −10 percent to +10 percent homeroom cortisol change from baseline to the high-stakes week), those with either large increases or decreases in cortisol from the baseline week performed worse on the standardized test. In other words, decreases and increases in cortisol were associated with underperformance on the high-stakes test. Grouping the “change” bins together, an increase of more than 10 percent or a decrease of more than 10 percent was associated with a 0.443 standard deviation decrease in the test score (p-value = 0.009), relative to those with little cortisol responsivity (−10 percent to +10 percent), holding school-year academic grades, demographic characteristics, and concurrent cortisol constant. Table 3 contains these results. The estimates are fairly similar when broken up by those who increase more than 10 percent (0.437 standard deviation lower scores relative to those with −10 percent to +10 percent change, p-value = 0.015) and those who decrease more than 10 percent (0.458 standard deviation lower scores, p-value = 0.021).
Changes in Test Scores by Cortisol Responsivity to the Test
| . | Preferred . | Low-stakes Scores as a Control . | No Concurrent Cortisol . | Adding Low-stakes Week . | Placebo: Within-baseline Change . | Placebo: Low-stakes Change . | Placebo: High-stakes Change . |
|---|---|---|---|---|---|---|---|
| . | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . |
| Panel A: Change in Test Scores for 10% Above/Below Baseline Cortisol in Testing Week | |||||||
| ±10% from baseline | −0.443** | −0.558* | −0.439** | −0.261* | −0.114 | −0.129 | −0.159 |
| (0.164) | (0.214) | (0.157) | (0.109) | (0.196) | (0.193) | (0.204) | |
| Panel B: Change in Test Scores for 10% Above or 10% Below Baseline Cortisol in Testing Week | |||||||
| 10% above baseline | −0.437* | −0.550* | −0.431* | −0.263* | −0.125 | −0.077 | −0.163 |
| (0.172) | (0.222) | (0.163) | (0.115) | (0.198) | (0.194) | (0.208) | |
| 10% below baseline | −0.458* | −0.589+ | −0.458* | −0.258* | −0.046 | −0.238 | −0.154 |
| (0.192) | (0.292) | (0.176) | (0.127) | (0.218) | (0.207) | (0.217) | |
| Controls: | |||||||
| Q1—3 grades | Y | N | Y | Q1 only | Y | Y | Y |
| Low-stakes test scores | N | Y | N | N | N | N | N |
| Concurrent cortisol | Y | Y | N | Y | Y | Y | Y |
| Cortisol change from baseline | To high-stakes | To high-stakes | To high-stakes | To low- OR high-stakes | Within baseline | To low-stakes | To high-stakes |
| Test outcome | High-stakes | High-stakes | High-stakes | Low- OR high-stakes | High-stakes | High-stakes | Low-stakes |
| N | 67 | 62 | 67 | 136 | 82 | 63 | 63 |
| . | Preferred . | Low-stakes Scores as a Control . | No Concurrent Cortisol . | Adding Low-stakes Week . | Placebo: Within-baseline Change . | Placebo: Low-stakes Change . | Placebo: High-stakes Change . |
|---|---|---|---|---|---|---|---|
| . | (1) . | (2) . | (3) . | (4) . | (5) . | (6) . | (7) . |
| Panel A: Change in Test Scores for 10% Above/Below Baseline Cortisol in Testing Week | |||||||
| ±10% from baseline | −0.443** | −0.558* | −0.439** | −0.261* | −0.114 | −0.129 | −0.159 |
| (0.164) | (0.214) | (0.157) | (0.109) | (0.196) | (0.193) | (0.204) | |
| Panel B: Change in Test Scores for 10% Above or 10% Below Baseline Cortisol in Testing Week | |||||||
| 10% above baseline | −0.437* | −0.550* | −0.431* | −0.263* | −0.125 | −0.077 | −0.163 |
| (0.172) | (0.222) | (0.163) | (0.115) | (0.198) | (0.194) | (0.208) | |
| 10% below baseline | −0.458* | −0.589+ | −0.458* | −0.258* | −0.046 | −0.238 | −0.154 |
| (0.192) | (0.292) | (0.176) | (0.127) | (0.218) | (0.207) | (0.217) | |
| Controls: | |||||||
| Q1—3 grades | Y | N | Y | Q1 only | Y | Y | Y |
| Low-stakes test scores | N | Y | N | N | N | N | N |
| Concurrent cortisol | Y | Y | N | Y | Y | Y | Y |
| Cortisol change from baseline | To high-stakes | To high-stakes | To high-stakes | To low- OR high-stakes | Within baseline | To low-stakes | To high-stakes |
| Test outcome | High-stakes | High-stakes | High-stakes | Low- OR high-stakes | High-stakes | High-stakes | Low-stakes |
| N | 67 | 62 | 67 | 136 | 82 | 63 | 63 |
Notes: Robust standard errors. Analysis conducted at the student level. Outcome is the Z-score of the indicated test. Cortisol data comes from saliva collected in homeroom. Each column represents a different regression estimate. All models also control for school fixed effects; a quadratic of time relative to wake; an indicator for whether the cortisol sample occurred within the CAR timeframe; wake time; age; female; and indicators for economically disadvantaged, Section 504, and McKinney-Vento Act (see Table 1 details). Column 1 is the preferred overall model. Columns 2—4 test alternate specifications by changing the measure of baseline ability (Column 2), removing controls for concurrent cortisol (Column 3), or adding the low-stakes week as an additional observation (Column 4). Column 4 uses quarter 1 grades as the control for baseline ability. Columns 5—7 test placebos of whether general cortisol variability is associated with test scores. Column 5 assesses within-baseline week changes of cortisol predict scores on the high-stakes test. Column 6 (7) tests whether cortisol changes from baseline to the low-stakes (high-stakes) test week affect high-stakes (low-stakes) test outcomes.
+p < 0.10; *p < 0.05; **p < 0.01.