Science Vocabulary/Words that have specific statistical implications

Bias: The difference between the estimated value and true value of a measurement or metric. There can be lots of different types (such as sample bias, which occurs when certain individuals are more likely to be sampled than others), but the extent of the bias is usually unknown. Not describing an unfair preference for something (e.g. NOT: I am biased towards monitoring more colourful organisms when carrying out my surveys).

Confidence / Confidence Interval: How likely is it that the estimated value you have obtained from what you are measuring would be similar to another estimated value if someone replicated the experiment? In science, we typically use 95% confidence intervals to make judgements; they mean we are 95% confident that the true value of what we are measuring would appear within the upper and lower limits of our sample’s interval. Not a subjective description or opinion (e.g. NOT: I am fairly confident that this new drug will speed up recovery time).

Error: Similar to bias in that it is present in any estimate (such as a sample mean) or measurement (such as the weight of an egg) and in that it refers to the difference between the estimated value and the true value. Not a description of a mistake (e.g. NOT: Forgetting the lab equipment was a large error).

Model: A set of rules that can be applied to a set of sampled data, and (theoretically) to a similar set sampled from the wider population. Models are usually explained by mathematical equations that describe the effect random and non-random variables have on the data/measurements being estimated. Not a physical structure that demonstrates something (e.g. NOT: This model of the human heart explains how blood is pumped around the body).

Positive/Negative Trend: A positive trend is when two variables move in the same direction as each other (e.g. as human height increases, weight typically increases as well); a negative trend is when two variables move in the opposite direction to each other (e.g. as latitude increases, species richness typically decreases). Note that positive and negative does not relate to whether a trend shows something good or bad.

p-value: The probability of obtaining at least as extreme results if the null hypothesis (that there is no true difference between what you are comparing) is true. So a p-value of 0.67 means there is a 67% likelihood of observing a difference as large as you observed even if the two population means are identical. We normally require the probability to be below 5% before we accept there is a difference between a comparison. Using the above example with a p-value of 0.67, NOT: there is a 33% chance that the difference you observed was due to chance.

Sensitivity vs Specificity: Sensitivity refers to the number of true positives identified in a study, whereas specificity refers to the number of true negatives identified. These concepts are important in medical research (e.g. when drug trials assess the suitability of drugs to treat/cure patients). A good experiment/method will be both sensitive and specific, but one can be sensitive but not specific, or vice versa. Not used to describe how caring (sensitive) or targeted (specific) your study was.

Significant: A significant result occurs when we are very confident an observed difference between groups/measurements equates to a real difference. In science, we typically compute p-values to assess significance, and usually set the value at 0.05. Only if the p-value is below this, do we reject the null hypothesis (that there is no difference) and support the alternate hypothesis (there is a difference). Not used to describe something subjectively, that seems large to you (e.g. NOT: The new budget means we will save a significant amount of money).

Trend: A pattern in data that implies a relationship between the variables being compared (the closer the relationship, the stronger the trend). Note that this does not necessarily mean variations in one variable cause the observed variations in the other. Also note that there can be a lot of variation in trends (not all individual data follow a perfect linear pattern).

Uncertainty: Because science gathers evidence to support hypotheses and laws, but doesn’t - and can’t - generally seek proofs, there is always a degree of uncertainty attached to any result or conclusion. This may be very small, but it requires scientists to phrase their words appropriately, especially when communicating to non-specialist audiences. Not to be phrased to suggest we are uncertain of the statistical significance that supports our conclusions, but to instead underline we are not presenting a proof/fact.