Cortisol, Sleep and Recall in an IB Student

The proposal is best described as a small-N, repeated-measures correlational study, with a quasi-experimental comparison if app users are compared with non-users. Maya is measured on cortisol, sleep and recall over 10 days, but nothing is manipulated, so the design cannot establish causation.

Two threats:

1. Confounding variables. Cortisol rises with caffeine, exercise, time of day and acute exam stress. Sleep is self-reported, so it may be inaccurate. If Maya sleeps badly and also has high cortisol, poor sleep could explain both the hormone level and the recall deficit.
2. Measurement validity. A single 20-word delayed-recall task is narrow; it may measure test anxiety or attention rather than memory. Salivary cortisol is affected by collection timing and food, and one participant gives no baseline.

Demand characteristics are also a risk: Maya knows the hypothesis and may try to perform well or report poor sleep. The app comparison is self-selected unless randomised. At most, the data can show an association. A causal claim needs a controlled manipulation or a much larger longitudinal design with covariates.

First, informed consent and assent. Maya is 17; in most jurisdictions she cannot consent alone. The team should seek parental consent and Maya's written assent, using an information sheet that explains the cortisol swabs, the time burden, data storage and the right to withdraw without academic penalty. Second, protection from harm and confidentiality. Asking about mood and memory can heighten anxiety, so the study needs a clear referral pathway and a named contact. The school's rule that a score below 30% triggers referral conflicts with confidentiality; this must be disclosed in advance, and the referral should go to a counsellor rather than to teachers. The team should also avoid diagnostic labels because Maya has not been clinically assessed.

Data protection matters: saliva and recall data are sensitive, so store them securely and anonymise them. If a commercial app is used, declare any conflict of interest. The ethics committee should review the protocol before any contact, and a pilot should reduce burden. The key reasoning is that consent is not a single signature; it is an ongoing process that must cover unexpected findings and the limits of confidentiality.

No. A negative correlation is consistent with stress impairing memory, but also with memory failure increasing stress, or with a third variable such as poor sleep, caffeine or exam timetable. The biological approach supplies a plausible mechanism: sustained cortisol exposure can affect hippocampal function, and the hippocampus is implicated in declarative memory. But plausibility is not proof. The design lacks a control group, randomisation and repeated standardised testing, and the app comparison is weak if users chose the app.

To support a causal claim, the team would need to manipulate stress or sleep under controlled conditions, or follow a larger cohort with covariates and repeated measures. They could test a dose-response pattern and rule out alternatives. The biological approach is also reductionist here: it may miss that Maya's low mood and forgetting could reflect depression, which itself affects cortisol and memory. The honest conclusion is "associated with". Any recommendation about the app should wait for a randomised trial, because a correlation between app use and recall could simply reflect more motivated students choosing to use it.