a parameter sweep
Often you do not want one answer, you want to see how the answer changes as you dial one value up and down, what does doubling this resistor do to the gain? A parameter sweep automates that: pick a value, give it a list or range, and SPICE reruns the whole analysis for each setting and overlays the results.
You define a named parameter, say R = {Rval}, then add a sweep directive like '.step param Rval 1k 10k 1k', which reruns from 1 kilohm to 10 kilohm in 1 kilohm steps. Critically, a parameter sweep wraps around another analysis: step the value and rerun a transient to watch an oscillator's frequency shift, or rerun an AC sweep to watch a filter's cutoff move. You get a family of curves on one graph instead of a single trace, turning trial-and-error tweaking into one informative picture.
This is the workhorse for design exploration and optimization: find the resistor that centres your bias, the capacitor that gives the flattest response, the load where gain peaks. It differs from Monte-Carlo in intent, a parameter sweep marches a value deliberately and evenly to map a trend, while Monte-Carlo randomizes many values at once to estimate production spread. Sweep too many values or nest two sweeps and the run time multiplies fast.
Stepping a filter's resistor with '.step param Rval 1k 10k 1k' while running an AC analysis draws ten gain curves at once, each with its cutoff at a different frequency, so you can read off which resistor lands the corner exactly where you want it.
Re-running an analysis across a range of one value, producing a family of curves that reveals a trend.
A parameter sweep maps how one chosen value affects the design; it is deliberate, not random. Do not confuse it with Monte-Carlo, which randomizes many tolerances together to estimate manufacturing spread.