Every calculator on this site produces a population estimate: a statement about what happened across many pregnancies or children whose answers matched yours, scaled to today. None of them measures you.
The most common misreadings come from treating a population figure as a personal one. Others come from treating a range as a prediction of its center, or treating the probability of an event as the probability of a particular day.
A percentage is a frequency
Take a statement that there is a 43% chance of labor in the next seven days. It means that among pregnancies matching the answers entered and still ongoing at this point, about 43 in 100 began labor within a week. It does not mean the calculator is 43% sure about you. Gigerenzer and colleagues (2007) found that probabilities are understood more accurately when expressed as frequencies, and the frequency reading is the correct one for every figure on this site.
The frequency reading also shows what the figure leaves out. The 100 pregnancies share the questionnaire answers and nothing else. Cervical findings, a clinician's assessment, and current symptoms sort them further, and the calculator does not see any of that.
A range is not a prediction of its middle
The labor calculator reports a 50% window and a most likely date. The most likely date is the single day with the highest probability, and that probability is about 4% to 6% in the baseline model. The window contains half the remaining probability. Half is outside it.
The child height predictor reports an estimate with a range from the published error bounds. Khamis and Roche (1994) reported average 90% absolute-error bounds of about 5 cm for boys and 4 cm for girls. The range around the estimate is the part of the result supported by the evidence. The center is the best single guess, and a child at the edge of the range is not an unusual outcome.
The child growth planner reports a central date and a range. Its range is a set of scenarios, not a calibrated interval, and the planner labels it so. A date outside the range is possible.
Conditional on today
The labor forecast is recalculated from today forward each time it runs. A window that read one way at 37 weeks and another way at 39 weeks has not been corrected. It has been conditioned on two more weeks without labor. The question being answered has changed, and the answer changed with it.
A pregnancy that continues past its median date is not late in any sense the forecast recognizes. Smith (2001) found a median of 284 days for first births, and half of first pregnancies continue past it by construction.
Small adjustments are the evidence, not a limitation
Users sometimes expect a previous birth or a risk factor to move the forecast by weeks. The calculator moves the term distribution by days, because the published associations are weak. Ng and Steer found a correlation of 0.10 between successive pregnancy lengths. The odds ratios for smoking and age are modest. A calculator that moved the forecast further would be claiming more than the studies do.
The one input with a large effect is a previous spontaneous preterm birth, which roughly triples the odds of another. The calculator applies it to the before-37-weeks component and reports the result as a separate figure.
Variation is real
Jukic and colleagues (2013) measured ovulation directly in 125 naturally conceived pregnancies and found gestational lengths from 247 to 284 days after excluding preterm births and related conditions. That 37-day range is biological, not a dating artifact, and it is why no calculator can narrow a labor forecast to a few days. A calculator's job is to locate the probability within the range, not to remove the range.
What to do with a result
Use a range as the span to prepare for and a threshold date, such as the 10% date, as the deadline to prepare by. Use conditional figures for questions about a specific date. Use the listed adjustments to see which answers moved the result and by how much. And use the result as one input to a conversation with the person providing care, who has information the calculator does not. The methodology page gives the source and classification of every figure the calculators use.